From nek5000-users at lists.mcs.anl.gov Mon Feb 1 02:52:13 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 1 Feb 2016 14:22:13 +0530 Subject: [Nek5000-users] Need for clarification Message-ID: Dear Sir Thank you for clarifying my doubts. Please forgive me, if you feel the questions I am posting are quite simple. I have just started to learn this. I have one more question in mind & it needs your guidance Sir. In the ".rea file", in the parameter list we have P01 & P02 for density and Reynolds number respectively. Its given in the manual that, when P02 is either negative or positive it will be treated as Reynolds number or reciprocal of Reynolds number. If so, where shall I define my kinematic viscocity? T/R Prashanth Mohanasundaram -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Feb 1 08:37:02 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 01 Feb 2016 14:37:02 +0000 Subject: [Nek5000-users] Does Nek5000 use any kind of parallelism inside one element? In-Reply-To: References: Message-ID: Hi Lars, Nek5000 doesn't implement any sort of parallelism within the elements. Typical element sizes are 8x8x8 or 12x12x12, so that hard coarse granularity limit is generally less restrictive than typical communication vs computation cost limitations. Running with many elements per MPI rank is even better! though this is not the common case. The other main advantage of OpenMP is the ability to share data, but Nek5000 has [almost] no shared state, so its not particularly relevant. The matrix multiplies are handled by compiled code that unrolls the "k" index. See: https://github.com/Nek5000/Nek5000/blob/master/mxm_std.f The matrices are generally too small for MKL: the overhead, from argument checking for example, wipes out any gain from better instructions. There is current work on integrating with LIBXSMM, which provides better performance in flop-bound cases. Hope that helps, Max On Mon, Feb 1, 2016 at 8:20 AM wrote: > Dear Nek5000 developpers and users, > > my name is Lars Haupt and I'am writing my phd-thesis at the moment, which > has something to do with scalable p-multigrid based approaches. Therefore i > have to understand the state of the art multigrid solver Nek5000 and thats > why i'am trying to get some insides into Nek5000 parallelism. > > I read the article about the scalability of the Nek5000 Code which was > achieved on the Juelich Supercomputer in 2010. > I thought that the impressive strong scaling speedup was a result of an > optimised hybrid parallelization strategy (MPI and OpenMP). > > But the general info on the Nek5000 website says that only MPI does the > parallel job. Which means, domain decomposition is the only kind of > parallelism? > What happens with the huge amount of nodes inside one element? The FAQ > tells something about high efficient matrix-matrix multiplications. Which > are implemented using MKL-BLAS 3 dgemm routines, i guess? > If that is the case, does Nek5000 use any OpenMP based parallelism inside > MKL or why not? > > Many thanks in advance > > Lars Haupt > > Dipl.-Math. Lars Haupt > Wissenschaftlicher Mitarbeiter > > Technische Universit?t Dresden > Fakult?t Maschinenwesen > Institut f?r Energietechnik > Professur f?r Geb?udeenergietechnik und W?rmeversorgung > 01062 Dresden > Tel.: +49(351) 463-37619 > Fax.: +49(351) 463-37076 > E-Mail: lars.haupt at tu-dresden.de > > > > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Feb 1 09:21:33 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 1 Feb 2016 08:21:33 -0700 Subject: [Nek5000-users] Need for clarification In-Reply-To: References: Message-ID: Hi Prasanth, Nek solves the navier-stokes equation in dimensional form. So, you can design your geometry in such a way such that your defined length scale L = 1, and your defined velocity scale U = 1 as well. So your Reynolds number = U L / nu = 1/nu. So,if you put P02 = -nu, it will be read as 1/nu. Best Regards, Tanmoy On Mon, Feb 1, 2016 at 1:52 AM, wrote: > Dear Sir > > Thank you for clarifying my doubts. Please forgive me, if you feel the > questions I am posting are quite simple. I have just started to learn > this. > > I have one more question in mind & it needs your guidance Sir. > In the ".rea file", in the parameter list we have P01 & P02 for density > and Reynolds number respectively. Its given in the manual that, when P02 is > either negative or positive it will be treated as Reynolds number or > reciprocal of Reynolds number. If so, where shall I define my kinematic > viscocity? > > T/R > Prashanth Mohanasundaram > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Feb 1 15:17:59 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 1 Feb 2016 16:17:59 -0500 Subject: [Nek5000-users] Clarification on Adjoint solver Message-ID: Hi Neks, To my knowledge, Nek5000 is able to solve adjoint equations of the form: $\partial u_a/\partial t + V.\nabla u_a - u_a.(\nabla V)^T + \nabla \p_a + 1/Re \nabla^2 u_a=0$ and $\nabla.u_a=0$ where u_a and p_a are adjoint velocity and pressure. (Slightly different formulation may be seen in literature). Let's assume the forward problem is already solved so that $V$, i.e. velocity, (and possibly $p$ as the pressure) is (are) already know. Thus, the initial and boundary conditions for adjoint NS are also known. How can we now solve the dual/adjoint NS problem with Nek5000? Is there any example on how to modify .rea files? Specailly, the "convection term of $V.\nabla u_a - u_a.(\nabla V)^T$ needs to be modified compared to forward NS but am not sure how. My search of mailing list entails some modifications in "perturb.f" subroutine, but a little bit more clarification on that would be really appreciated. Yours, Saleh From nek5000-users at lists.mcs.anl.gov Mon Feb 1 23:00:56 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 2 Feb 2016 13:00:56 +0800 Subject: [Nek5000-users] Point votices at inlet Message-ID: Hi neks, I still cannot for the life of me figure out how to apply this vortex method at the inlet to generate turbulence. Let's say i specify a point at the inlet plane say for example at the coordinate (x1, y1). I need to apply the Biot-Savart law at this point which has a term like (*x_i - x) **(Note these are vectors) where x_i is the point i specify and x is all the nodes on that plane. How do i call on the nodes to do this vector subtraction? Can anybody help?* *rif* -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Feb 2 02:01:07 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 2 Feb 2016 16:01:07 +0800 Subject: [Nek5000-users] xm1(i,1,1,1) is? Message-ID: HI neks, Can anybody tell me what xm1(i,1,1,1) means? i think xm1 is mesh 1 but what are the terms in the bracket that i come across so often in some of the examples. rif -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Feb 2 02:07:49 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 2 Feb 2016 09:07:49 +0100 Subject: [Nek5000-users] xm1(i,1,1,1) is? In-Reply-To: References: Message-ID: Hi Rif, xm1 is the x-coordinate in mesh 1 (ym1 and zm1 are the other two coordinates). Each of these variables and in fact all the velocity, pressure, temperature etc. variables are defined in each point of the mesh, which is displayed in the brackets. This is explained well in part 3.4 of the Nek5000 User Manual entitled 'Data Layout'. Hope this helps, Holly Le 02/02/2016 09:01, nek5000-users at lists.mcs.anl.gov a ?crit : > HI neks, > Can anybody tell me what xm1(i,1,1,1) means? i think xm1 is mesh 1 but > what are the terms in the bracket that i come across so often in some > of the examples. > > > rif > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Feb 2 02:09:35 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 2 Feb 2016 09:09:35 +0100 Subject: [Nek5000-users] Clarification on Adjoint solver In-Reply-To: References: Message-ID: Hi The adjoint implementation in nek assumes change of the direction of time evolution. The goal is to use the standard nek solver with minimal modifications (only advection terms have to be changed). You get these equations from you standard one by changing the time and pressure sign (t -> -t; p-> -p). This of course changes your time integration limits, but it shouldn't be a problem. So when you have your linear direct simulation, to turn it into dual one it is enough to include in your userchk include files ADJOINT and TSTEP, and for ISTEP.eq.0 set IFADJOINT to true. Something like: subroutine userchk include 'SIZE' include 'TSTEP' ! ISTEP include 'ADJOINT' ! IFADJ if(ISTEP.eq.0) ADJOINT=.TRUE. There is no need to modify ###.rea or ###.map files. The only problem are boundary conditions for open flows, as they would require non-homogeneous bc. In this case to avoid problems we usually use sufficiently large domain together with zero Dirichlet bc. The last issue is sufficient resolution for both direct and adjoin runs. I hope this gives you some idea about nek implementation. Regards Adam On 2016-02-01 22:17, nek5000-users at lists.mcs.anl.gov wrote: > Hi Neks, > > To my knowledge, Nek5000 is able to solve adjoint equations of the form: > $\partial u_a/\partial t + V.\nabla u_a - u_a.(\nabla V)^T + \nabla > \p_a + 1/Re \nabla^2 u_a=0$ and $\nabla.u_a=0$ > where u_a and p_a are adjoint velocity and pressure. (Slightly > different formulation may be seen in literature). > > Let's assume the forward problem is already solved so that $V$, i.e. > velocity, (and possibly $p$ as the pressure) is (are) already know. > Thus, the initial and boundary conditions for adjoint NS are also > known. How can we now solve the dual/adjoint NS problem with Nek5000? > Is there any example on how to modify .rea files? Specailly, the > "convection term of $V.\nabla u_a - u_a.(\nabla V)^T$ needs to be > modified compared to forward NS but am not sure how. > My search of mailing list entails some modifications in "perturb.f" > subroutine, but a little bit more clarification on that would be > really appreciated. > > Yours, > Saleh > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Tue Feb 2 04:31:13 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 2 Feb 2016 10:31:13 +0000 Subject: [Nek5000-users] Find out which cell a meshpoint belongs to In-Reply-To: References: , Message-ID: Hi neks, I'm just wondering if there's currently a function which acts on a meshpoint and returns the cell it is closest to. I have also tried doing this manually but I need to know the corners of each cell, if that is stored somewhere? Otherwise, is there a way of generating a list of every gridpoint so I can calculate the corners using brute force. Thanks, Lee On Tue, Feb 2, 2016 at 12:10 AM -0800, "nek5000-users at lists.mcs.anl.gov" > wrote: Hi The adjoint implementation in nek assumes change of the direction of time evolution. The goal is to use the standard nek solver with minimal modifications (only advection terms have to be changed). You get these equations from you standard one by changing the time and pressure sign (t -> -t; p-> -p). This of course changes your time integration limits, but it shouldn't be a problem. So when you have your linear direct simulation, to turn it into dual one it is enough to include in your userchk include files ADJOINT and TSTEP, and for ISTEP.eq.0 set IFADJOINT to true. Something like: subroutine userchk include 'SIZE' include 'TSTEP' ! ISTEP include 'ADJOINT' ! IFADJ if(ISTEP.eq.0) ADJOINT=.TRUE. There is no need to modify ###.rea or ###.map files. The only problem are boundary conditions for open flows, as they would require non-homogeneous bc. In this case to avoid problems we usually use sufficiently large domain together with zero Dirichlet bc. The last issue is sufficient resolution for both direct and adjoin runs. I hope this gives you some idea about nek implementation. Regards Adam On 2016-02-01 22:17, nek5000-users at lists.mcs.anl.gov wrote: > Hi Neks, > > To my knowledge, Nek5000 is able to solve adjoint equations of the form: > $\partial u_a/\partial t + V.\nabla u_a - u_a.(\nabla V)^T + \nabla > \p_a + 1/Re \nabla^2 u_a=0$ and $\nabla.u_a=0$ > where u_a and p_a are adjoint velocity and pressure. (Slightly > different formulation may be seen in literature). > > Let's assume the forward problem is already solved so that $V$, i.e. > velocity, (and possibly $p$ as the pressure) is (are) already know. > Thus, the initial and boundary conditions for adjoint NS are also > known. How can we now solve the dual/adjoint NS problem with Nek5000? > Is there any example on how to modify .rea files? Specailly, the > "convection term of $V.\nabla u_a - u_a.(\nabla V)^T$ needs to be > modified compared to forward NS but am not sure how. > My search of mailing list entails some modifications in "perturb.f" > subroutine, but a little bit more clarification on that would be > really appreciated. > > Yours, > Saleh > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Feb 3 11:41:04 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 3 Feb 2016 12:41:04 -0500 Subject: [Nek5000-users] example problem Message-ID: Hey neks i am just trying to learn NEK5000. I WAS TRYING TO DO THE FRST EDDY EXAMPLE AND GOT THE FOLLOWING ERROR WHILE TRYING TO DO './makenek eddy_uv' FATAL ERROR: cannot find mpif77! how do i make the program work further? also, how do i change the path of the fortran mpi compiler and to what? plz help AVR -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Feb 3 11:53:24 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 3 Feb 2016 12:53:24 -0500 Subject: [Nek5000-users] issue Message-ID: hi NEKS i have to solve a steady state problem for flow through an axisymmetric channel. This is my problem which i have to solve: I have the initial conditions (temperature and velocity) which is for turbulent flow and the exit condition(pressure). the system transitions to laminar and reaches steady state over heating. Thus variations are due to temperature and not the time making the system steady state. Is it possible to do it using NEK500? does it have to timescaling involved considering nek5000 can only solve for unsteady state NS equations i really appreciate your help and i m desperate Confused coder -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Feb 3 13:26:55 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 3 Feb 2016 14:26:55 -0500 Subject: [Nek5000-users] example problem In-Reply-To: References: Message-ID: Do you have MPI installed? Try typing which mpif77 Or mpiexec --version Hey neks i am just trying to learn NEK5000. I WAS TRYING TO DO THE FRST EDDY EXAMPLE AND GOT THE FOLLOWING ERROR WHILE TRYING TO DO './makenek eddy_uv' FATAL ERROR: cannot find mpif77! how do i make the program work further? also, how do i change the path of the fortran mpi compiler and to what? plz help AVR _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Feb 3 13:42:56 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 3 Feb 2016 19:42:56 +0000 Subject: [Nek5000-users] example problem In-Reply-To: References: Message-ID: Make sure that you have your mpi module loaded. You can change the compilers in the makenek file. ________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov on behalf of nek5000-users at lists.mcs.anl.gov Sent: Wednesday, February 03, 2016 12:41 PM To: nek5000-users at lists.mcs.anl.gov Subject: [Nek5000-users] example problem Hey neks i am just trying to learn NEK5000. I WAS TRYING TO DO THE FRST EDDY EXAMPLE AND GOT THE FOLLOWING ERROR WHILE TRYING TO DO './makenek eddy_uv' FATAL ERROR: cannot find mpif77! how do i make the program work further? also, how do i change the path of the fortran mpi compiler and to what? plz help AVR -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Feb 4 09:29:43 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 4 Feb 2016 15:29:43 +0000 Subject: [Nek5000-users] Combining multiple case files in Visit Message-ID: Hi All, I have across some videos (https://nek5000.mcs.anl.gov/gallery/) made in Visit that combine multiple case files. I am trying to figure out how to do that. For example: I have result files from two cases: a0.f000* and b0.f000*, is there a way to combine the a.nek5000 and b.nek5000 files that I get from running visnek? Thanks, Ketan -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Fri Feb 5 03:06:53 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 5 Feb 2016 17:06:53 +0800 Subject: [Nek5000-users] is the IFKEPS working? Message-ID: Hi neks, I was just browsing through the .rea file and realized that there is a logical switch for IFKEPS which i assumed is the k epsilon model. If I were to use this model, how do i input the turbulence levels at the inlet boundary? Many thanks in adavnce. Cheers rif -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Fri Feb 5 04:12:31 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 5 Feb 2016 18:12:31 +0800 Subject: [Nek5000-users] xm1(i,1,1,1) is? Message-ID: Hi Holly, It helps a little. Thanks. But I am still a little confused. Suppose I want to identify a coordinate at my inlet, say (1,4) where x is 1 and y is 4. How do I write this in the usr file? The reason i'm asking this is so that i can put in a source point at the inlet. Is the following correct? do i=1,n x=xm1(i,1,1,1) y=ym1(i,1,1,1) x1=1 y1=4 . . . enddo rif -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Fri Feb 5 07:52:41 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 5 Feb 2016 14:52:41 +0100 Subject: [Nek5000-users] P101 and P103 Message-ID: Hi Neks, can someone explain clearly how this parameters (101 and 103) in .rea file, works exactly? From nek5000-users at lists.mcs.anl.gov Fri Feb 5 14:15:42 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 5 Feb 2016 15:15:42 -0500 Subject: [Nek5000-users] Clarification on Adjoint solver In-Reply-To: References: Message-ID: Adams, I really appreciate your detailed answer. I definitely clarified the Nek functionality for adjoint solvers. One question remains is how to include the velocity of the forward solver into adjoint equations. I assume this is also the case when you do linear stability analysis with Nek, since the velocity in convective terms is not completely part of the solution. Mathematically, my question is: in $V. \nabla u_a$ term, as the advection, how can we substitute $V$ (as the forward velocity which is already solved for in a regular NS solution) in adjoint equations. suppose that the velocity, along with other quantities are saved in afield out put such as fld.f0000X for X time-step. So we have access to that velocity but how can we include it in adjoint equations? Many thanks again, Saleh On 2/2/2016 3:09 AM, nek5000-users at lists.mcs.anl.gov wrote: > Hi > > The adjoint implementation in nek assumes change of the direction of > time evolution. The goal is to use the standard nek solver with > minimal modifications (only advection terms have to be changed). You > get these equations from you standard one by changing the time and > pressure sign (t -> -t; p-> -p). This of course changes your time > integration limits, but it shouldn't be a problem. So when you have > your linear direct simulation, to turn it into dual one it is enough > to include in your userchk include files ADJOINT and TSTEP, and for > ISTEP.eq.0 set IFADJOINT to true. Something like: > subroutine userchk > > include 'SIZE' > include 'TSTEP' ! ISTEP > include 'ADJOINT' ! IFADJ > > if(ISTEP.eq.0) ADJOINT=.TRUE. > > There is no need to modify ###.rea or ###.map files. The only problem > are boundary conditions for open flows, as they would require > non-homogeneous bc. In this case to avoid problems we usually use > sufficiently large domain together with zero Dirichlet bc. The last > issue is sufficient resolution for both direct and adjoin runs. I hope > this gives you some idea about nek implementation. > Regards > Adam > > On 2016-02-01 22:17, nek5000-users at lists.mcs.anl.gov wrote: >> Hi Neks, >> >> To my knowledge, Nek5000 is able to solve adjoint equations of the form: >> $\partial u_a/\partial t + V.\nabla u_a - u_a.(\nabla V)^T + \nabla >> \p_a + 1/Re \nabla^2 u_a=0$ and $\nabla.u_a=0$ >> where u_a and p_a are adjoint velocity and pressure. (Slightly >> different formulation may be seen in literature). >> >> Let's assume the forward problem is already solved so that $V$, i.e. >> velocity, (and possibly $p$ as the pressure) is (are) already know. >> Thus, the initial and boundary conditions for adjoint NS are also >> known. How can we now solve the dual/adjoint NS problem with Nek5000? >> Is there any example on how to modify .rea files? Specailly, the >> "convection term of $V.\nabla u_a - u_a.(\nabla V)^T$ needs to be >> modified compared to forward NS but am not sure how. >> My search of mailing list entails some modifications in "perturb.f" >> subroutine, but a little bit more clarification on that would be >> really appreciated. >> >> Yours, >> Saleh >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Sat Feb 6 12:15:14 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Sat, 6 Feb 2016 18:15:14 +0000 Subject: [Nek5000-users] Clarification on Adjoint solver In-Reply-To: References: , Message-ID: Hi Saleh, There is no easily accessible implementation for this, one has to do the change of the velocities himself. In fact we have ongoing work on this and once we are done I guess we can share it. What kind of case are you looking at, long time horizons are quite tricky to handle. We can take this conversation off the mailing list if needed. Oana ________________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] Sent: Friday, February 05, 2016 2:15 PM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] Clarification on Adjoint solver Adams, I really appreciate your detailed answer. I definitely clarified the Nek functionality for adjoint solvers. One question remains is how to include the velocity of the forward solver into adjoint equations. I assume this is also the case when you do linear stability analysis with Nek, since the velocity in convective terms is not completely part of the solution. Mathematically, my question is: in $V. \nabla u_a$ term, as the advection, how can we substitute $V$ (as the forward velocity which is already solved for in a regular NS solution) in adjoint equations. suppose that the velocity, along with other quantities are saved in afield out put such as fld.f0000X for X time-step. So we have access to that velocity but how can we include it in adjoint equations? Many thanks again, Saleh On 2/2/2016 3:09 AM, nek5000-users at lists.mcs.anl.gov wrote: > Hi > > The adjoint implementation in nek assumes change of the direction of > time evolution. The goal is to use the standard nek solver with > minimal modifications (only advection terms have to be changed). You > get these equations from you standard one by changing the time and > pressure sign (t -> -t; p-> -p). This of course changes your time > integration limits, but it shouldn't be a problem. So when you have > your linear direct simulation, to turn it into dual one it is enough > to include in your userchk include files ADJOINT and TSTEP, and for > ISTEP.eq.0 set IFADJOINT to true. Something like: > subroutine userchk > > include 'SIZE' > include 'TSTEP' ! ISTEP > include 'ADJOINT' ! IFADJ > > if(ISTEP.eq.0) ADJOINT=.TRUE. > > There is no need to modify ###.rea or ###.map files. The only problem > are boundary conditions for open flows, as they would require > non-homogeneous bc. In this case to avoid problems we usually use > sufficiently large domain together with zero Dirichlet bc. The last > issue is sufficient resolution for both direct and adjoin runs. I hope > this gives you some idea about nek implementation. > Regards > Adam > > On 2016-02-01 22:17, nek5000-users at lists.mcs.anl.gov wrote: >> Hi Neks, >> >> To my knowledge, Nek5000 is able to solve adjoint equations of the form: >> $\partial u_a/\partial t + V.\nabla u_a - u_a.(\nabla V)^T + \nabla >> \p_a + 1/Re \nabla^2 u_a=0$ and $\nabla.u_a=0$ >> where u_a and p_a are adjoint velocity and pressure. (Slightly >> different formulation may be seen in literature). >> >> Let's assume the forward problem is already solved so that $V$, i.e. >> velocity, (and possibly $p$ as the pressure) is (are) already know. >> Thus, the initial and boundary conditions for adjoint NS are also >> known. How can we now solve the dual/adjoint NS problem with Nek5000? >> Is there any example on how to modify .rea files? Specailly, the >> "convection term of $V.\nabla u_a - u_a.(\nabla V)^T$ needs to be >> modified compared to forward NS but am not sure how. >> My search of mailing list entails some modifications in "perturb.f" >> subroutine, but a little bit more clarification on that would be >> really appreciated. >> >> Yours, >> Saleh >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Sat Feb 6 21:36:05 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Sun, 7 Feb 2016 03:36:05 +0000 Subject: [Nek5000-users] P101 and P103 In-Reply-To: References: Message-ID: These parameters control the filter that is applied to the velocity and thermal fields at the end of each time step. p103 controls the amplitude of the filter, 101 controls the number of modes that are filtered. The modal decomposition is as follows: In each direction, on each element, the solution is represented as a polynomial of degree N in the reference element (i.e., on the interval [-1,1]). For any Nth-order polynomial there is a modal expansion of the form: u(x) = \sum_k=0^N uh_k phi_k(x) where phi_0 = 1, phi_1 = x, and phi_k(x)=L_k(x) - L_{k-1} (x) for k > 1, where L_k(x)=kth Legendre polynomial. With this modal decomposition and p101=0, the effect of the filter is to construct ub(x) = sum_k=0^N sigma_k uh_k phi_k(x) ~= u(x) where sigma_k = 1 for k < N and sigma_N = 1-alpha, where alpha:=p103 is the filter strength. alpha=1 corresponds to a sharp cut-off filter and produces a less smooth result than, say, alpha=0.05--0.25, which are typical filter strengths. If p101 > 0, then more modes are filtered. In this case, one has sigma_k = 1 for k < N-p101. sigma_k = 1-alpha*p(k) for k = N-p101, ..., N where p(k) is the parabola given by: p(k) = [ (k-kc)/(N-kc) ]^2, and kc := N-(p101+1) The net result is a filter that is flat in modal space for k < N-p101 and which tapers off to the value alpha as k-->N. The overall effect of the filter is to suppress high wavenumber instabilities. It is easy to show that the filter does not inhibit exponential convergence, i.e., the effect of the filter is nil when the solution is well resolved. References for this filter are Fischer & Mullen, CRAS 2001, and J.P. Boyd, JCP, 1998. Paul ________________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] Sent: Friday, February 05, 2016 7:52 AM To: nek5000-users at lists.mcs.anl.gov Subject: [Nek5000-users] P101 and P103 Hi Neks, can someone explain clearly how this parameters (101 and 103) in .rea file, works exactly? _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Mon Feb 8 03:26:35 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 8 Feb 2016 10:26:35 +0100 Subject: [Nek5000-users] Clarification on Adjoint solver In-Reply-To: References: Message-ID: Hi Saleh If I understand you correctly, you ask about a base flow, i.e. the velocity field you do your linearisation about (the one you use to advect prturbation). If this is the case, than it is possible to do with nek. Otherwise you have to wait for the code Oana mentioned. What we do in our stability calculations, we take the solution of non-linear NS (steady state reached with e.g. SFD) and next we load it into nek as non-linear filed. One can do it by setting restart options in ###.rea file. It should look like: 3.33333 3.33333 -0.833333 -1.16667 XFAC,YFAC,XZERO,YZERO **MESH DATA** 1st line is X of corner 1,2,3,4. 2nd line is Y. -1472 2 1472 NELT,NDIM,NELV 1 PRESOLVE/RESTART OPTIONS ***** bfext_cyl?.f00000 7 INITIAL CONDITIONS ***** C Default C Default Where bfext_cyl0.f00000 is our restart file (we usually save this velocity field in double precision). Next you have to set linear solver with non-evolving base flow (nek is able to evolve both so be careful). You freeze baseflow by setting negative value of parameter 31 in ###.rea: 0.00000 p030 > 0 ==> properties set in uservp() -1.00000 p031 NPERT: #perturbation modes 0.00000 p032 #BCs in re2 file, if > 0 Nek is able of evolving number of perturbation fields set by abs(param(31)) (in ###.rea) and lpert in SIZE. I have to mention that nek by itself outputs the non-linear fields only, so to see perturbation you should add to your userchk something like this subroutine userchk include 'SIZE' ! NX1, NY1, NZ1, NELV, NIO include 'TSTEP' ! ISTEP, IOSTEP include 'SOLN' ! V[XYZ]P, TP include 'INPUT' ! IF3D, PARAM include 'ADJOINT' ! IFADJ ! direct or adjoint run IFADJ = .FALSE. ! write perturbation field if (mod(ISTEP,int(PARAM(15))).eq.0) then call outpost2(VXP,VYP,VZP,PRP,TP,0,'prt') endif I hope it is an answer to your question. regards adam On 2016-02-06 19:15, nek5000-users at lists.mcs.anl.gov wrote: > Hi Saleh, > > There is no easily accessible implementation for this, one has to do the change of the velocities himself. In fact we have ongoing work on this and once we are done I guess we can share it. > What kind of case are you looking at, long time horizons are quite tricky to handle. We can take this conversation off the mailing list if needed. > > Oana > > > > ________________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > Sent: Friday, February 05, 2016 2:15 PM > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Clarification on Adjoint solver > > Adams, > > I really appreciate your detailed answer. I definitely clarified the Nek > functionality for adjoint solvers. One question remains is how to > include the velocity of the forward solver into adjoint equations. I > assume this is also the case when you do linear stability analysis with > Nek, since the velocity in convective terms is not completely part of > the solution. Mathematically, my question is: > > in $V. \nabla u_a$ term, as the advection, how can we substitute $V$ (as > the forward velocity which is already solved for in a regular NS > solution) in adjoint equations. suppose that the velocity, along with > other quantities are saved in afield out put such as fld.f0000X for X > time-step. So we have access to that velocity but how can we include it > in adjoint equations? > > Many thanks again, > Saleh > > On 2/2/2016 3:09 AM, nek5000-users at lists.mcs.anl.gov wrote: >> Hi >> >> The adjoint implementation in nek assumes change of the direction of >> time evolution. The goal is to use the standard nek solver with >> minimal modifications (only advection terms have to be changed). You >> get these equations from you standard one by changing the time and >> pressure sign (t -> -t; p-> -p). This of course changes your time >> integration limits, but it shouldn't be a problem. So when you have >> your linear direct simulation, to turn it into dual one it is enough >> to include in your userchk include files ADJOINT and TSTEP, and for >> ISTEP.eq.0 set IFADJOINT to true. Something like: >> subroutine userchk >> >> include 'SIZE' >> include 'TSTEP' ! ISTEP >> include 'ADJOINT' ! IFADJ >> >> if(ISTEP.eq.0) ADJOINT=.TRUE. >> >> There is no need to modify ###.rea or ###.map files. The only problem >> are boundary conditions for open flows, as they would require >> non-homogeneous bc. In this case to avoid problems we usually use >> sufficiently large domain together with zero Dirichlet bc. The last >> issue is sufficient resolution for both direct and adjoin runs. I hope >> this gives you some idea about nek implementation. >> Regards >> Adam >> >> On 2016-02-01 22:17, nek5000-users at lists.mcs.anl.gov wrote: >>> Hi Neks, >>> >>> To my knowledge, Nek5000 is able to solve adjoint equations of the form: >>> $\partial u_a/\partial t + V.\nabla u_a - u_a.(\nabla V)^T + \nabla >>> \p_a + 1/Re \nabla^2 u_a=0$ and $\nabla.u_a=0$ >>> where u_a and p_a are adjoint velocity and pressure. (Slightly >>> different formulation may be seen in literature). >>> >>> Let's assume the forward problem is already solved so that $V$, i.e. >>> velocity, (and possibly $p$ as the pressure) is (are) already know. >>> Thus, the initial and boundary conditions for adjoint NS are also >>> known. How can we now solve the dual/adjoint NS problem with Nek5000? >>> Is there any example on how to modify .rea files? Specailly, the >>> "convection term of $V.\nabla u_a - u_a.(\nabla V)^T$ needs to be >>> modified compared to forward NS but am not sure how. >>> My search of mailing list entails some modifications in "perturb.f" >>> subroutine, but a little bit more clarification on that would be >>> really appreciated. >>> >>> Yours, >>> Saleh >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Mon Feb 8 04:05:51 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 8 Feb 2016 11:05:51 +0100 Subject: [Nek5000-users] help Message-ID: Hi neks, i need your help to understand what happen during my simulations runs. 1) - I start a new sim and set the iotime to 2.00 s in .rea file and all work correctly. Now I restart my sim from the last file wrote before, and I see it run but it doesn't write new files. If I drop the iotime to,for example, 0.5s,the sim go on and write more than four files. So why if I set iotime equal 2s nek5000 doesn't write anything and if I set iotime to 0.5 nek write correctly over 2s? This is the last screen printed: Solving for fluid T T T 2307 PRES alph1n 2.4872E-01 1.0591E-01 2.3483E+00 1 2307 PRES halpha 1 1.9233E-02 2307 PRES gmres: 5 8.8294E-06 1.0000E-05 1.5316E-04 2.1182E+00 2.7194E+00 F 2307 1 2.8670E-01 1.7757E-03 1.6146E+02 5 alph1x 2307 halpha 5 1.1392E-01 6.4126E-02 1.9957E-02 1.3173E-02 7.5946E-03 2307 Hmholtz VELX: 4 7.7538E-09 1.7757E-03 1.0000E-08 2307 2 2.9984E-01 9.8403E-04 3.0471E+02 5 alph1x 2307 halpha 5 1.1120E-01 7.9115E-02 2.6727E-02 7.0552E-03 2.0218E-03 2307 Hmholtz VELY: 4 5.3019E-09 9.8403E-04 1.0000E-08 2307 3 3.3424E-01 9.4197E-04 3.5483E+02 5 alph1x 2307 halpha 5 1.2703E-01 8.4245E-02 2.8403E-02 7.2785E-03 1.9445E-03 2307 Hmholtz VELZ: 4 5.1587E-09 9.4197E-04 1.0000E-08 L1/L2 DIV(V) : -2.5232E-17 8.1201E-03 L1/L2 QTL : 0.0000E+00 0.0000E+00 L1/L2 DIV(V)-QTL: -2.5232E-17 8.1201E-03 2307 9.1310E+01 7.0010E+00 Fluid done Step 2308, t= 9.1319540E+01, DT= 1.0000000E-02, C= 0.348 2.0401E+04 7.4189E+00 Solving for fluid T T T 2308 PRES alph1n 2.5051E-01 1.0694E-01 2.3425E+00 2 2308 PRES halpha 2 1.8189E-02 6.7561E-03 2308 PRES gmres: 5 8.0852E-06 1.0000E-05 1.5471E-04 2.5837E+00 3.1443E+00 F 2308 1 2.8670E-01 1.1724E-03 2.4453E+02 6 alph1x 2308 halpha 6 1.0646E-01 7.2246E-02 2.5549E-02 1.6177E-02 1.5435E-02 3.6983E-03 2308 Hmholtz VELX: 4 7.6167E-09 1.1724E-03 1.0000E-08 2308 2 2.9981E-01 6.6475E-04 4.5101E+02 6 alph1x 2308 halpha 6 1.0042E-01 8.8049E-02 3.7024E-02 1.2498E-02 4.7047E-03 2.0487E-03 2308 Hmholtz VELY: 4 5.6393E-09 6.6475E-04 1.0000E-08 2308 3 3.3420E-01 6.6358E-04 5.0363E+02 6 alph1x 2308 halpha 6 1.1612E-01 9.4096E-02 3.9493E-02 1.3028E-02 4.5878E-03 1.9661E-03 2308 Hmholtz VELZ: 4 5.2639E-09 6.6358E-04 1.0000E-08 2) - Another strange thing is the following: i run a simulation, it go on till name0.f000XX file. I restart the same simulation from name0.f000XX but Nek5000 is stucks in the following screen: Beginning session: /datastor/eng/gualtieri/SIMULATIONS_NEK/RIB_3D_109/curve.rea timer accuracy: 9.5367432E-08 sec read .rea file nelgt/nelgv/lelt: 395328 395328 6200 lx1 /lx2 /lx3 : 4 4 4 mapping elements to processors 0 6177 6177 395328 395328 NELV 1 6177 6177 395328 395328 NELV 2 6177 6177 395328 395328 NELV 3 6177 6177 395328 395328 NELV 4 6177 6177 395328 395328 NELV 5 6177 6177 395328 395328 NELV 6 6177 6177 395328 395328 NELV 7 6177 6177 395328 395328 NELV 10 6177 6177 395328 395328 NELV 11 6177 6177 395328 395328 NELV 12 6177 6177 395328 395328 NELV 13 6177 6177 395328 395328 NELV 14 6177 6177 395328 395328 NELV 15 6177 6177 395328 395328 NELV 8 6177 6177 395328 395328 NELV 9 6177 6177 395328 395328 NELV 32 6177 6177 395328 395328 NELV 33 6177 6177 395328 395328 NELV 23 6177 6177 395328 395328 NELV 34 6177 6177 395328 395328 NELV 24 6177 6177 395328 395328 NELV 35 6177 6177 395328 395328 NELV 25 6177 6177 395328 395328 NELV 36 6177 6177 395328 395328 NELV 26 6177 6177 395328 395328 NELV 37 6177 6177 395328 395328 NELV 27 6177 6177 395328 395328 NELV 38 6177 6177 395328 395328 NELV 28 6177 6177 395328 395328 NELV 39 6177 6177 395328 395328 NELV 29 6177 6177 395328 395328 NELV 40 6177 6177 395328 395328 NELV 30 6177 6177 395328 395328 NELV 41 6177 6177 395328 395328 NELV 31 6177 6177 395328 395328 NELV 42 6177 6177 395328 395328 NELV 16 6177 6177 395328 395328 NELV 43 6177 6177 395328 395328 NELV 17 6177 6177 395328 395328 NELV 44 6177 6177 395328 395328 NELV 18 6177 6177 395328 395328 NELV 45 6177 6177 395328 395328 NELV 19 6177 6177 395328 395328 NELV 46 6177 6177 395328 395328 NELV 20 6177 6177 395328 395328 NELV 47 6177 6177 395328 395328 NELV 21 6177 6177 395328 395328 NELV 22 6177 6177 395328 395328 NELV 60 6177 6177 395328 395328 NELV 61 6177 6177 395328 395328 NELV 62 6177 6177 395328 395328 NELV 63 6177 6177 395328 395328 NELV 48 6177 6177 395328 395328 NELV 49 6177 6177 395328 395328 NELV 50 6177 6177 395328 395328 NELV 51 6177 6177 395328 395328 NELV 52 6177 6177 395328 395328 NELV 53 6177 6177 395328 395328 NELV 54 6177 6177 395328 395328 NELV 55 6177 6177 395328 395328 NELV 56 6177 6177 395328 395328 NELV 57 6177 6177 395328 395328 NELV 58 6177 6177 395328 395328 NELV 59 6177 6177 395328 395328 NELV element load imbalance: 0 6177 6177 done :: mapping elements to processors 0 objects found done :: read .rea file 25.415 sec setup mesh topology Right-handed check complete for 395328 elements. OK. setvert3d: 4 7585356 10747980 7585356 7585356 call usrsetvert done :: usrsetvert gs_setup: 534804 unique labels shared From nek5000-users at lists.mcs.anl.gov Tue Feb 9 12:25:16 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 9 Feb 2016 13:25:16 -0500 Subject: [Nek5000-users] output ascii format Message-ID: Hello Nek5000 users, I'm new in Nek and I'm trying to output field results in an ascii format. I added negative value at p66 of .rea file but I'm still getting binary files. Any help will be appreciated. Thanks -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Feb 9 13:18:21 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 9 Feb 2016 19:18:21 +0000 Subject: [Nek5000-users] output ascii format In-Reply-To: References: Message-ID: Hi, Look whether you have an overwrite param(66) = 4 somewhere in your user file Aleks ________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] Sent: Tuesday, February 09, 2016 12:25 PM To: nek5000-users at lists.mcs.anl.gov Subject: [Nek5000-users] output ascii format Hello Nek5000 users, I'm new in Nek and I'm trying to output field results in an ascii format. I added negative value at p66 of .rea file but I'm still getting binary files. Any help will be appreciated. Thanks -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Feb 10 01:02:23 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 10 Feb 2016 08:02:23 +0100 Subject: [Nek5000-users] SYM boundary conditions Message-ID: Hi Neks, I am looking for a boundary condition that leads the flow to go out or to go into the domain. Is 'SYM' valid for this purpose? I think that what I am looking for is, in the outlet surface: - If the velocity vector at the outlet points out of the domain, then the boundary condition for the velocity will be of the Neumann type - However, if the velocity at the outlet points into the domain, then we can imagine that the outflow is no longer an outflow, but an inflow, and that we therefore would like to specify a Dirichlet boundary condition for velocity Thanks in advance SL From nek5000-users at lists.mcs.anl.gov Wed Feb 10 03:20:52 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 10 Feb 2016 10:20:52 +0100 Subject: [Nek5000-users] SYM boundary conditions In-Reply-To: References: Message-ID: Hi, SYM is essentially means that the velocity through the interface is zero and for the inplane velocities that the normal gradients vanish. So it does not allow transpiration/penetration through the boundary. So I suspect that this condition is not what you are looking for. However, I am not exactly sure what you really mean. Perhaps you can say a bit more about your case? Best regards, Philipp On 2016-02-10 08:02, nek5000-users at lists.mcs.anl.gov wrote: > Hi Neks, > > I am looking for a boundary condition that leads the flow to go out or > to go into the domain. > > Is 'SYM' valid for this purpose? > > > I think that what I am looking for is, in the outlet surface: > > - If the velocity vector at the outlet points out of the domain, then > the boundary condition for the velocity will be of the Neumann type > > - However, if the velocity at the outlet points into the domain, then > we can imagine that the outflow is no longer an outflow, but an inflow, > and that we therefore would like to specify a Dirichlet boundary > condition for velocity > > > Thanks in advance > SL > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Wed Feb 10 04:45:46 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 10 Feb 2016 11:45:46 +0100 Subject: [Nek5000-users] SYM boundary conditions In-Reply-To: References: Message-ID: Hi, Thank you. I am trying to simulate a flow whose inlet boundary condition is time periodic ux=sin(omega*time). Since there is an injection and a suction, and the mean flow is zero, I think that the boundary conditions should let the transpiration/penetration of the flow... (I have tried to run the simulation using outflow BC but it crashes). Is there any BC condition in Nek5000 that suits to this problem? Thank you for your help in advance. Best regards, SL El 10-02-2016 10:20, nek5000-users at lists.mcs.anl.gov escribi?: > Hi, > SYM is essentially means that the velocity through the interface is > zero and for the inplane velocities that the normal gradients vanish. > So it does not allow transpiration/penetration through the boundary. > So I suspect that this condition is not what you are looking for. > > However, I am not exactly sure what you really mean. Perhaps you can > say a bit more about your case? > > Best regards, > Philipp > > On 2016-02-10 08:02, nek5000-users at lists.mcs.anl.gov wrote: >> Hi Neks, >> >> I am looking for a boundary condition that leads the flow to go out or >> to go into the domain. >> >> Is 'SYM' valid for this purpose? >> >> >> I think that what I am looking for is, in the outlet surface: >> >> - If the velocity vector at the outlet points out of the domain, >> then >> the boundary condition for the velocity will be of the Neumann type >> >> - However, if the velocity at the outlet points into the domain, >> then >> we can imagine that the outflow is no longer an outflow, but an >> inflow, >> and that we therefore would like to specify a Dirichlet boundary >> condition for velocity >> >> >> Thanks in advance >> SL >> >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Wed Feb 10 05:17:28 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 10 Feb 2016 12:17:28 +0100 Subject: [Nek5000-users] SYM boundary conditions In-Reply-To: References: Message-ID: one initial suggestion would be to use ON at the outflow. Is the flow turbulent at the outflow boundary? Philipp On 2016-02-10 11:45, nek5000-users at lists.mcs.anl.gov wrote: > Hi, > > Thank you. I am trying to simulate a flow whose inlet boundary condition > is time periodic ux=sin(omega*time). Since there is an injection and a > suction, and the mean flow is zero, I think that the boundary conditions > should let the transpiration/penetration of the flow... (I have tried to > run the simulation using outflow BC but it crashes). Is there any BC > condition in Nek5000 that suits to this problem? > > Thank you for your help in advance. > Best regards, > SL > > El 10-02-2016 10:20, nek5000-users at lists.mcs.anl.gov escribi?: >> Hi, >> SYM is essentially means that the velocity through the interface is >> zero and for the inplane velocities that the normal gradients vanish. >> So it does not allow transpiration/penetration through the boundary. >> So I suspect that this condition is not what you are looking for. >> >> However, I am not exactly sure what you really mean. Perhaps you can >> say a bit more about your case? >> >> Best regards, >> Philipp >> >> On 2016-02-10 08:02, nek5000-users at lists.mcs.anl.gov wrote: >>> Hi Neks, >>> >>> I am looking for a boundary condition that leads the flow to go out or >>> to go into the domain. >>> >>> Is 'SYM' valid for this purpose? >>> >>> >>> I think that what I am looking for is, in the outlet surface: >>> >>> - If the velocity vector at the outlet points out of the domain, then >>> the boundary condition for the velocity will be of the Neumann type >>> >>> - However, if the velocity at the outlet points into the domain, then >>> we can imagine that the outflow is no longer an outflow, but an inflow, >>> and that we therefore would like to specify a Dirichlet boundary >>> condition for velocity >>> >>> >>> Thanks in advance >>> SL >>> >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Wed Feb 10 06:02:02 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 10 Feb 2016 13:02:02 +0100 Subject: [Nek5000-users] SYM boundary conditions In-Reply-To: References: Message-ID: Yes, it is turbulent... I will try to use ON. Please, could you help me to understand a bit more why it could work? Thanks again! Cheers SL El 10-02-2016 12:17, nek5000-users at lists.mcs.anl.gov escribi?: > one initial suggestion would be to use ON at the outflow. Is the flow > turbulent at the outflow boundary? > Philipp > > On 2016-02-10 11:45, nek5000-users at lists.mcs.anl.gov wrote: >> Hi, >> >> Thank you. I am trying to simulate a flow whose inlet boundary >> condition >> is time periodic ux=sin(omega*time). Since there is an injection and a >> suction, and the mean flow is zero, I think that the boundary >> conditions >> should let the transpiration/penetration of the flow... (I have tried >> to >> run the simulation using outflow BC but it crashes). Is there any BC >> condition in Nek5000 that suits to this problem? >> >> Thank you for your help in advance. >> Best regards, >> SL >> >> El 10-02-2016 10:20, nek5000-users at lists.mcs.anl.gov escribi?: >>> Hi, >>> SYM is essentially means that the velocity through the interface is >>> zero and for the inplane velocities that the normal gradients vanish. >>> So it does not allow transpiration/penetration through the boundary. >>> So I suspect that this condition is not what you are looking for. >>> >>> However, I am not exactly sure what you really mean. Perhaps you can >>> say a bit more about your case? >>> >>> Best regards, >>> Philipp >>> >>> On 2016-02-10 08:02, nek5000-users at lists.mcs.anl.gov wrote: >>>> Hi Neks, >>>> >>>> I am looking for a boundary condition that leads the flow to go out >>>> or >>>> to go into the domain. >>>> >>>> Is 'SYM' valid for this purpose? >>>> >>>> >>>> I think that what I am looking for is, in the outlet surface: >>>> >>>> - If the velocity vector at the outlet points out of the domain, >>>> then >>>> the boundary condition for the velocity will be of the Neumann type >>>> >>>> - However, if the velocity at the outlet points into the domain, >>>> then >>>> we can imagine that the outflow is no longer an outflow, but an >>>> inflow, >>>> and that we therefore would like to specify a Dirichlet boundary >>>> condition for velocity >>>> >>>> >>>> Thanks in advance >>>> SL >>>> >>>> _______________________________________________ >>>> Nek5000-users mailing list >>>> Nek5000-users at lists.mcs.anl.gov >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >> >> >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Wed Feb 10 08:04:27 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 10 Feb 2016 15:04:27 +0100 Subject: [Nek5000-users] SYM boundary conditions In-Reply-To: References: Message-ID: ok, I see. If it is turbulent, then it is a bit more difficult. You can try ON, but perhaps it will not work. But in the real case, what type of inflow are you going to have, something turbulent? What about damping the turbulence before it actually reaches the outflow boundary, e.g. using a forcing? ONce it is laminar, I think that ON could work. Philipp On 2016-02-10 13:02, nek5000-users at lists.mcs.anl.gov wrote: > Yes, it is turbulent... > > I will try to use ON. Please, could you help me to understand a bit more > why it could work? > > Thanks again! > Cheers > SL > > > El 10-02-2016 12:17, nek5000-users at lists.mcs.anl.gov escribi?: >> one initial suggestion would be to use ON at the outflow. Is the flow >> turbulent at the outflow boundary? >> Philipp >> >> On 2016-02-10 11:45, nek5000-users at lists.mcs.anl.gov wrote: >>> Hi, >>> >>> Thank you. I am trying to simulate a flow whose inlet boundary condition >>> is time periodic ux=sin(omega*time). Since there is an injection and a >>> suction, and the mean flow is zero, I think that the boundary conditions >>> should let the transpiration/penetration of the flow... (I have tried to >>> run the simulation using outflow BC but it crashes). Is there any BC >>> condition in Nek5000 that suits to this problem? >>> >>> Thank you for your help in advance. >>> Best regards, >>> SL >>> >>> El 10-02-2016 10:20, nek5000-users at lists.mcs.anl.gov escribi?: >>>> Hi, >>>> SYM is essentially means that the velocity through the interface is >>>> zero and for the inplane velocities that the normal gradients vanish. >>>> So it does not allow transpiration/penetration through the boundary. >>>> So I suspect that this condition is not what you are looking for. >>>> >>>> However, I am not exactly sure what you really mean. Perhaps you can >>>> say a bit more about your case? >>>> >>>> Best regards, >>>> Philipp >>>> >>>> On 2016-02-10 08:02, nek5000-users at lists.mcs.anl.gov wrote: >>>>> Hi Neks, >>>>> >>>>> I am looking for a boundary condition that leads the flow to go out or >>>>> to go into the domain. >>>>> >>>>> Is 'SYM' valid for this purpose? >>>>> >>>>> >>>>> I think that what I am looking for is, in the outlet surface: >>>>> >>>>> - If the velocity vector at the outlet points out of the domain, >>>>> then >>>>> the boundary condition for the velocity will be of the Neumann type >>>>> >>>>> - However, if the velocity at the outlet points into the domain, >>>>> then >>>>> we can imagine that the outflow is no longer an outflow, but an >>>>> inflow, >>>>> and that we therefore would like to specify a Dirichlet boundary >>>>> condition for velocity >>>>> >>>>> >>>>> Thanks in advance >>>>> SL >>>>> >>>>> _______________________________________________ >>>>> Nek5000-users mailing list >>>>> Nek5000-users at lists.mcs.anl.gov >>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>>> _______________________________________________ >>>> Nek5000-users mailing list >>>> Nek5000-users at lists.mcs.anl.gov >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> >>> >>> >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Wed Feb 10 08:36:12 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 10 Feb 2016 14:36:12 +0000 Subject: [Nek5000-users] SYM boundary conditions In-Reply-To: References: , Message-ID: Is this an internal flow, or external flow? If internal, you can use the turn_outflow() routine with sufficient divergence to ensure that the flow is always leaving the domain. We use this for some of our moving domain simulations. Paul ________________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] Sent: Wednesday, February 10, 2016 8:04 AM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] SYM boundary conditions ok, I see. If it is turbulent, then it is a bit more difficult. You can try ON, but perhaps it will not work. But in the real case, what type of inflow are you going to have, something turbulent? What about damping the turbulence before it actually reaches the outflow boundary, e.g. using a forcing? ONce it is laminar, I think that ON could work. Philipp On 2016-02-10 13:02, nek5000-users at lists.mcs.anl.gov wrote: > Yes, it is turbulent... > > I will try to use ON. Please, could you help me to understand a bit more > why it could work? > > Thanks again! > Cheers > SL > > > El 10-02-2016 12:17, nek5000-users at lists.mcs.anl.gov escribi?: >> one initial suggestion would be to use ON at the outflow. Is the flow >> turbulent at the outflow boundary? >> Philipp >> >> On 2016-02-10 11:45, nek5000-users at lists.mcs.anl.gov wrote: >>> Hi, >>> >>> Thank you. I am trying to simulate a flow whose inlet boundary condition >>> is time periodic ux=sin(omega*time). Since there is an injection and a >>> suction, and the mean flow is zero, I think that the boundary conditions >>> should let the transpiration/penetration of the flow... (I have tried to >>> run the simulation using outflow BC but it crashes). Is there any BC >>> condition in Nek5000 that suits to this problem? >>> >>> Thank you for your help in advance. >>> Best regards, >>> SL >>> >>> El 10-02-2016 10:20, nek5000-users at lists.mcs.anl.gov escribi?: >>>> Hi, >>>> SYM is essentially means that the velocity through the interface is >>>> zero and for the inplane velocities that the normal gradients vanish. >>>> So it does not allow transpiration/penetration through the boundary. >>>> So I suspect that this condition is not what you are looking for. >>>> >>>> However, I am not exactly sure what you really mean. Perhaps you can >>>> say a bit more about your case? >>>> >>>> Best regards, >>>> Philipp >>>> >>>> On 2016-02-10 08:02, nek5000-users at lists.mcs.anl.gov wrote: >>>>> Hi Neks, >>>>> >>>>> I am looking for a boundary condition that leads the flow to go out or >>>>> to go into the domain. >>>>> >>>>> Is 'SYM' valid for this purpose? >>>>> >>>>> >>>>> I think that what I am looking for is, in the outlet surface: >>>>> >>>>> - If the velocity vector at the outlet points out of the domain, >>>>> then >>>>> the boundary condition for the velocity will be of the Neumann type >>>>> >>>>> - However, if the velocity at the outlet points into the domain, >>>>> then >>>>> we can imagine that the outflow is no longer an outflow, but an >>>>> inflow, >>>>> and that we therefore would like to specify a Dirichlet boundary >>>>> condition for velocity >>>>> >>>>> >>>>> Thanks in advance >>>>> SL >>>>> >>>>> _______________________________________________ >>>>> Nek5000-users mailing list >>>>> Nek5000-users at lists.mcs.anl.gov >>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>>> _______________________________________________ >>>> Nek5000-users mailing list >>>> Nek5000-users at lists.mcs.anl.gov >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> >>> >>> >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Tue Feb 9 15:42:25 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 9 Feb 2016 21:42:25 +0000 Subject: [Nek5000-users] Help Message-ID: I have to change my makenek file every time i open linux or ubuntu Sent from my iPhone From nek5000-users at lists.mcs.anl.gov Wed Feb 10 11:58:34 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 10 Feb 2016 12:58:34 -0500 Subject: [Nek5000-users] Clarification on Adjoint solver In-Reply-To: References: Message-ID: Adam and Oana, Thanks again for your responses. Let me summarize what I understood. 1) We may have either evolving (unsteady) or non-evolving (steady) base flow. As for the latter, we may use restart option in rea file. 2) For non-evolving case (which is what I have in mind in the end) Nek "is able to" handle and to do so parameter #31 in rea file needs to be modified. Can you please clarify how. Suppose that there is velocity and a passive scalar or even temperature, so there are three adjoint fields (velocity, pressure and the scalar). Also, is what Oana mentioned, "one has to do the change of the velocities himself" related to remark number 2 above? 3) There's also a confusion in userchk: First Adam mentioned that we need to include 'ADJOINT' plus if(ISTEP.eq.0) ADJOINT=.TRUE. for adjoint solver in order to i) changing the time and pressure sign and 2) modification of advection terms. But in the second example, while he still uses include 'ADJOINT' in userchk but then there is this statement IFADJ = .FALSE. Can you please clarify what's the difference here? I'm really grateful that you take the time to answer these questions. Please let me know if we can take this conversation off the mailing list. Brief summary of the problem we have in mind: we'd like to optimize the flow with respect to some parameters (either BCs or ICs) in terms of a defined cost function. So this can be regarded as either design or optimal control problem. Calculus of variations and Lagrange multipliers lead to adjoint equations which are very similar to NS equations. We already solved the problem using NEK and we'd like to use this again for adjoint equations. All the results will be in an outer loop for optimization (based on, say, steepest gradient). In general, the problem is unsteady, but we can start off with steady and non-evolving cases at early stages. Best, Saleh On 2/8/2016 4:26 AM, nek5000-users at lists.mcs.anl.gov wrote: > Hi Saleh > > If I understand you correctly, you ask about a base flow, i.e. the > velocity field you do your linearisation about (the one you use to > advect prturbation). If this is the case, than it is possible to do > with nek. Otherwise you have to wait for the code Oana mentioned. What > we do in our stability calculations, we take the solution of > non-linear NS (steady state reached with e.g. SFD) and next we load it > into nek as non-linear filed. One can do it by setting restart options > in ###.rea file. It should look like: > 3.33333 3.33333 -0.833333 -1.16667 > XFAC,YFAC,XZERO,YZERO > **MESH DATA** 1st line is X of corner 1,2,3,4. 2nd line is Y. > -1472 2 1472 NELT,NDIM,NELV > 1 PRESOLVE/RESTART OPTIONS ***** > bfext_cyl?.f00000 > 7 INITIAL CONDITIONS ***** > C Default > C Default > > > Where bfext_cyl0.f00000 is our restart file (we usually save this > velocity field in double precision). Next you have to set linear > solver with non-evolving base flow (nek is able to evolve both so be > careful). You freeze baseflow by setting negative value of parameter > 31 in ###.rea: > 0.00000 p030 > 0 ==> properties set in uservp() > -1.00000 p031 NPERT: #perturbation modes > 0.00000 p032 #BCs in re2 file, if > 0 > > Nek is able of evolving number of perturbation fields set by > abs(param(31)) (in ###.rea) and lpert in SIZE. I have to mention that > nek by itself outputs the non-linear fields only, so to see > perturbation you should add to your userchk something like this > > > subroutine userchk > > include 'SIZE' ! NX1, NY1, NZ1, NELV, NIO > include 'TSTEP' ! ISTEP, IOSTEP > include 'SOLN' ! V[XYZ]P, TP > include 'INPUT' ! IF3D, PARAM > include 'ADJOINT' ! IFADJ > > ! direct or adjoint run > IFADJ = .FALSE. > > ! write perturbation field > if (mod(ISTEP,int(PARAM(15))).eq.0) then > call outpost2(VXP,VYP,VZP,PRP,TP,0,'prt') > endif > > I hope it is an answer to your question. > regards > adam > > On 2016-02-06 19:15, nek5000-users at lists.mcs.anl.gov wrote: >> Hi Saleh, >> >> There is no easily accessible implementation for this, one has to do >> the change of the velocities himself. In fact we have ongoing work on >> this and once we are done I guess we can share it. >> What kind of case are you looking at, long time horizons are quite >> tricky to handle. We can take this conversation off the mailing list >> if needed. >> >> Oana >> >> >> >> ________________________________________ >> From: nek5000-users-bounces at lists.mcs.anl.gov >> [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of >> nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> Sent: Friday, February 05, 2016 2:15 PM >> To: nek5000-users at lists.mcs.anl.gov >> Subject: Re: [Nek5000-users] Clarification on Adjoint solver >> >> Adams, >> >> I really appreciate your detailed answer. I definitely clarified the Nek >> functionality for adjoint solvers. One question remains is how to >> include the velocity of the forward solver into adjoint equations. I >> assume this is also the case when you do linear stability analysis with >> Nek, since the velocity in convective terms is not completely part of >> the solution. Mathematically, my question is: >> >> in $V. \nabla u_a$ term, as the advection, how can we substitute $V$ (as >> the forward velocity which is already solved for in a regular NS >> solution) in adjoint equations. suppose that the velocity, along with >> other quantities are saved in afield out put such as fld.f0000X for X >> time-step. So we have access to that velocity but how can we include it >> in adjoint equations? >> >> Many thanks again, >> Saleh >> >> On 2/2/2016 3:09 AM, nek5000-users at lists.mcs.anl.gov wrote: >>> Hi >>> >>> The adjoint implementation in nek assumes change of the direction of >>> time evolution. The goal is to use the standard nek solver with >>> minimal modifications (only advection terms have to be changed). You >>> get these equations from you standard one by changing the time and >>> pressure sign (t -> -t; p-> -p). This of course changes your time >>> integration limits, but it shouldn't be a problem. So when you have >>> your linear direct simulation, to turn it into dual one it is enough >>> to include in your userchk include files ADJOINT and TSTEP, and for >>> ISTEP.eq.0 set IFADJOINT to true. Something like: >>> subroutine userchk >>> >>> include 'SIZE' >>> include 'TSTEP' ! ISTEP >>> include 'ADJOINT' ! IFADJ >>> >>> if(ISTEP.eq.0) ADJOINT=.TRUE. >>> >>> There is no need to modify ###.rea or ###.map files. The only problem >>> are boundary conditions for open flows, as they would require >>> non-homogeneous bc. In this case to avoid problems we usually use >>> sufficiently large domain together with zero Dirichlet bc. The last >>> issue is sufficient resolution for both direct and adjoin runs. I hope >>> this gives you some idea about nek implementation. >>> Regards >>> Adam >>> >>> On 2016-02-01 22:17, nek5000-users at lists.mcs.anl.gov wrote: >>>> Hi Neks, >>>> >>>> To my knowledge, Nek5000 is able to solve adjoint equations of the >>>> form: >>>> $\partial u_a/\partial t + V.\nabla u_a - u_a.(\nabla V)^T + \nabla >>>> \p_a + 1/Re \nabla^2 u_a=0$ and $\nabla.u_a=0$ >>>> where u_a and p_a are adjoint velocity and pressure. (Slightly >>>> different formulation may be seen in literature). >>>> >>>> Let's assume the forward problem is already solved so that $V$, i.e. >>>> velocity, (and possibly $p$ as the pressure) is (are) already know. >>>> Thus, the initial and boundary conditions for adjoint NS are also >>>> known. How can we now solve the dual/adjoint NS problem with Nek5000? >>>> Is there any example on how to modify .rea files? Specailly, the >>>> "convection term of $V.\nabla u_a - u_a.(\nabla V)^T$ needs to be >>>> modified compared to forward NS but am not sure how. >>>> My search of mailing list entails some modifications in "perturb.f" >>>> subroutine, but a little bit more clarification on that would be >>>> really appreciated. >>>> >>>> Yours, >>>> Saleh >>>> _______________________________________________ >>>> Nek5000-users mailing list >>>> Nek5000-users at lists.mcs.anl.gov >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Wed Feb 10 14:06:13 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 10 Feb 2016 20:06:13 +0000 Subject: [Nek5000-users] help : compiling error Message-ID: I have been compiling nek for now 2 weeks and still it is not done. in the user's manual I have done till maketools all. after that I was I did cd mkdir eddy cd eddy cp ~/nek5_svn/examples/eddy/* . cp ~/nek5_svn/trunk/nek/makenek . than for applying changes vi makenek I changed F77 to gfortran , CC to gcc and uncommented the ifmpi = false than for compiling I did ./makenek eddy_uv it does not get compile and gives me following please help what should I do? gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/mvmesh.f -o obj/mvmesh.o In file /home/users/mansi/nek5_svn/trunk/nek/turb.f:578 COMMON /SCRUZ/ V1(LX1,LY1,LZ1,LELV) 1 Warning: Named COMMON block 'scruz' at (1) shall be of the same size gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/ic.f -o obj/ic.o gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/ssolv.f -o obj/ssolv.o gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/planx.f -o obj/planx.o gfortran -c -O3 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I./ /home/users/mansi/nek5_svn/trunk/nek/math.f -o obj/math.o gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/mxm_wrapper.f -o obj/mxm_wrapper.o gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/hmholtz.f -o obj/hmholtz.o gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/gfdm_par.f -o obj/gfdm_par.o In file /home/users/mansi/nek5_svn/trunk/nek/hmholtz.f:638 COMMON /CPRINT/ IFPRINT, IFHZPC 1 Warning: Named COMMON block 'cprint' at (1) shall be of the same size gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/gfdm_op.f -o obj/gfdm_op.o gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/gfdm_solve.f -o obj/gfdm_solve.o gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/subs1.f -o obj/subs1.o In file /home/users/mansi/nek5_svn/trunk/nek/hmholtz.f:2143 common /ctmp1/ ur(lt),us(lt),ut(lt) 1 Warning: Named COMMON block 'ctmp1' at (1) shall be of the same size gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/subs2.f -o obj/subs2.o In file /home/users/mansi/nek5_svn/trunk/nek/subs1.f:1578 common /ctmp1/ exx(lx1*ly1*lz1*lelt) 1 Warning: Named COMMON block 'ctmp1' at (1) shall be of the same size gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/genbox.f -o obj/genbox.o gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/gmres.f -o obj/gmres.o gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/hsmg.f -o obj/hsmg.o gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/convect.f -o obj/convect.o gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/induct.f -o obj/induct.o In file /home/users/mansi/nek5_svn/trunk/nek/hsmg.f:296 common /hsmgw/ work(0:lwk-1),work2(0:lwk-1) 1 Warning: Named COMMON block 'hsmgw' at (1) shall be of the same size In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:189 COMMON /SCRNS/ TA1(LX1,LY1,LZ1,LELV) 1 Warning: Named COMMON block 'scrns' at (1) shall be of the same size In file /home/users/mansi/nek5_svn/trunk/nek/convect.f:722 tr(i,2)=rx(i,4,e)*fx(i)+rx(i,5,e)*fy(i)+rx(i,6,e)*fz(i) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/convect.f:722 tr(i,2)=rx(i,4,e)*fx(i)+rx(i,5,e)*fy(i)+rx(i,6,e)*fz(i) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/convect.f:723 tr(i,3)=rx(i,7,e)*fx(i)+rx(i,8,e)*fy(i)+rx(i,9,e)*fz(i) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/convect.f:723 tr(i,3)=rx(i,7,e)*fx(i)+rx(i,8,e)*fy(i)+rx(i,9,e)*fz(i) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/convect.f:723 tr(i,3)=rx(i,7,e)*fx(i)+rx(i,8,e)*fy(i)+rx(i,9,e)*fz(i) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/convect.f:919 cs(i,e)=rx(i,4,e)*fx(i)+rx(i,5,e)*fy(i)+rx(i,6,e)*fz(i) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/convect.f:919 cs(i,e)=rx(i,4,e)*fx(i)+rx(i,5,e)*fy(i)+rx(i,6,e)*fz(i) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/convect.f:920 ct(i,e)=rx(i,7,e)*fx(i)+rx(i,8,e)*fy(i)+rx(i,9,e)*fz(i) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/convect.f:920 ct(i,e)=rx(i,7,e)*fx(i)+rx(i,8,e)*fy(i)+rx(i,9,e)*fz(i) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/convect.f:920 ct(i,e)=rx(i,7,e)*fx(i)+rx(i,8,e)*fy(i)+rx(i,9,e)*fz(i) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/convect.f:946 common /ctmp1/ work(lxyz1,lelt) 1 Warning: Named COMMON block 'ctmp1' at (1) shall be of the same size In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:472 c1 = lf(i,2)*b3(i) - lf(i,3)*b2(i) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:473 c2 = lf(i,3)*b1(i) - lf(i,1)*b3(i) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:477 lf(i,3) = c3 1 Warning: Array reference at (1) is out of bounds gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/perturb.f -o obj/perturb.o In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:1079 $ ,vproj(1,5),ivproj(1,5),binvm1) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:1083 $ ,vproj(1,6),ivproj(1,6),binvm1) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:1379 $ rx(i,4,e)*zm(i,1)+rx(i,5,e)*zm(i,2)+rx(i,6,e)*zm(i,3) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:1379 $ rx(i,4,e)*zm(i,1)+rx(i,5,e)*zm(i,2)+rx(i,6,e)*zm(i,3) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:1381 $ rx(i,7,e)*zm(i,1)+rx(i,8,e)*zm(i,2)+rx(i,9,e)*zm(i,3) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:1381 $ rx(i,7,e)*zm(i,1)+rx(i,8,e)*zm(i,2)+rx(i,9,e)*zm(i,3) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:1381 $ rx(i,7,e)*zm(i,1)+rx(i,8,e)*zm(i,2)+rx(i,9,e)*zm(i,3) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:1408 $ rx(i,4,e)*zp(i,1)+rx(i,5,e)*zp(i,2)+rx(i,6,e)*zp(i,3) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:1408 $ rx(i,4,e)*zp(i,1)+rx(i,5,e)*zp(i,2)+rx(i,6,e)*zp(i,3) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:1410 $ rx(i,7,e)*zp(i,1)+rx(i,8,e)*zp(i,2)+rx(i,9,e)*zp(i,3) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:1410 $ rx(i,7,e)*zp(i,1)+rx(i,8,e)*zp(i,2)+rx(i,9,e)*zp(i,3) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:1410 $ rx(i,7,e)*zp(i,1)+rx(i,8,e)*zp(i,2)+rx(i,9,e)*zp(i,3) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/perturb.f:343 $ +rx(i,7,e)*utz(i)) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/perturb.f:341 $ +rx(i,7,e)*uty(i))+ 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/perturb.f:339 $ +rx(i,7,e)*utx(i))+ 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/perturb.f:349 $ +rx(i,8,e)*utz(i)) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/perturb.f:348 $ fz(i)*(rx(i,2,e)*urz(i)+rx(i,5,e)*usz(i) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/perturb.f:347 $ +rx(i,8,e)*uty(i))+ 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/perturb.f:346 $ fy(i)*(rx(i,2,e)*ury(i)+rx(i,5,e)*usy(i) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/perturb.f:345 $ +rx(i,8,e)*utx(i))+ 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/perturb.f:344 uf5(i)=fx(i)*(rx(i,2,e)*urx(i)+rx(i,5,e)*usx(i) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/perturb.f:355 $ +rx(i,9,e)*utz(i)) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/perturb.f:354 $ fz(i)*(rx(i,3,e)*urz(i)+rx(i,6,e)*usz(i) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/perturb.f:353 $ +rx(i,9,e)*uty(i))+ 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/perturb.f:352 $ fy(i)*(rx(i,3,e)*ury(i)+rx(i,6,e)*usy(i) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/perturb.f:351 $ +rx(i,9,e)*utx(i))+ 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/perturb.f:350 uf6(i)=fx(i)*(rx(i,3,e)*urx(i)+rx(i,6,e)*usx(i) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:1552 call intp_rstd(rx(1,5,e),sym1(1,1,1,e),nx1,nxd,if3d,0) ! 0 - 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:1553 call intp_rstd(rx(1,6,e),szm1(1,1,1,e),nx1,nxd,if3d,0) ! 0 - 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:1554 call intp_rstd(rx(1,7,e),txm1(1,1,1,e),nx1,nxd,if3d,0) ! 0 - 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:1555 call intp_rstd(rx(1,8,e),tym1(1,1,1,e),nx1,nxd,if3d,0) ! 0 - 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/induct.f:1556 call intp_rstd(rx(1,9,e),tzm1(1,1,1,e),nx1,nxd,if3d,0) ! 0 - 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/perturb.f:454 COMMON /SCRNS/ TA1(LX1,LY1,LZ1,LELV) 1 Warning: Named COMMON block 'scrns' at (1) shall be of the same size In file /home/users/mansi/nek5_svn/trunk/nek/convect.f:1538 common /scrdg/ uf(lf),uxf(lf),uyf(lf),uzf(lf),upwind_wgt(lf) 1 Warning: Named COMMON block 'scrdg' at (1) shall be of the same size In file /home/users/mansi/nek5_svn/trunk/nek/hsmg.f:2535 common /scrvh/ h1 (lx1,ly1,lz1,lelv) 1 Warning: Named COMMON block 'scrvh' at (1) shall be of the same size In file /home/users/mansi/nek5_svn/trunk/nek/perturb.f:700 common /scruz/ ta (lx1,ly1,lz1,lelt) 1 Warning: Named COMMON block 'scruz' at (1) shall be of the same size In file /home/users/mansi/nek5_svn/trunk/nek/hsmg.f:2924 common /ctmp0/ w(100000) 1 Warning: Named COMMON block 'ctmp0' at (1) shall be of the same size gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/navier5.f -o obj/navier5.o gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/navier6.f -o obj/navier6.o gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/navier7.f -o obj/navier7.o In file /home/users/mansi/nek5_svn/trunk/nek/navier5.f:1822 s21 = sij(j1,j2,1,4,e) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/navier5.f:1823 s31 = sij(j1,j2,1,6,e) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/navier5.f:1825 s12 = sij(j1,j2,1,4,e) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/navier5.f:1827 s32 = sij(j1,j2,1,5,e) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/navier5.f:1829 s13 = sij(j1,j2,1,6,e) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/navier5.f:1830 s23 = sij(j1,j2,1,5,e) 1 Warning: Array reference at (1) is out of bounds gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/navier8.f -o obj/navier8.o In file /home/users/mansi/nek5_svn/trunk/nek/navier5.f:2706 common /scravg/ u (lyavg) 1 Warning: Named COMMON block 'scravg' at (1) shall be of the same size gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/fast3d.f -o obj/fast3d.o In file /home/users/mansi/nek5_svn/trunk/nek/navier8.f:2022 common /scrns/ ind,etuple 1 Warning: Named COMMON block 'scrns' at (1) shall be of the same size In file /home/users/mansi/nek5_svn/trunk/nek/fast3d.f:1585 llr(e) = l(1,2,2,e)-lmr(e) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/fast3d.f:1586 lrr(e) = l(n,2,2,e)-lmr(e) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/fast3d.f:1587 lls(e) = l(2,1,2,e)-lms(e) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/fast3d.f:1588 lrs(e) = l(2,n,2,e)-lms(e) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/navier5.f:3056 xt(l,3) = z(i,j,k) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/navier5.f:3217 s21 = sij(j1,j2,1,4,e) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/navier5.f:3218 s31 = sij(j1,j2,1,6,e) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/navier5.f:3220 s12 = sij(j1,j2,1,4,e) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/navier5.f:3222 s32 = sij(j1,j2,1,5,e) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/navier5.f:3224 s13 = sij(j1,j2,1,6,e) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/navier5.f:3225 s23 = sij(j1,j2,1,5,e) 1 Warning: Array reference at (1) is out of bounds gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/fasts.f -o obj/fasts.o /home/users/mansi/nek5_svn/trunk/nek/navier8.f: In function 'setvert3d': /home/users/mansi/nek5_svn/trunk/nek/navier8.f:2354: internal compiler error: in expand_expr_real_1, at expr.c:6846 Please submit a full bug report, with preprocessed source if appropriate. See for instructions. make: *** [obj/navier8.o] Error 1 make: *** Waiting for unfinished jobs.... In file /home/users/mansi/nek5_svn/trunk/nek/fasts.f:343 l(i,j,2,e)=1 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/fasts.f:384 w(i,j,1,3,eb)=1.0/l(i+1,j+1,2,e) 1 Warning: Array reference at (1) is out of bounds In file /home/users/mansi/nek5_svn/trunk/nek/navier5.f:4085 common /ctmp1/ w(lt,lelt),ur(lt),us(lt),ut(lt),w1(2*lt) 1 Warning: Named COMMON block 'ctmp1' at (1) shall be of the same size [mansi at maea203-01 eddy]$ -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Feb 10 15:10:26 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 10 Feb 2016 22:10:26 +0100 Subject: [Nek5000-users] Clarification on Adjoint solver In-Reply-To: References: Message-ID: Dear all, Let me add some answers below. On 2016-02-10 18:58, nek5000-users at lists.mcs.anl.gov wrote: > Adam and Oana, > Thanks again for your responses. > Let me summarize what I understood. > 1) We may have either evolving (unsteady) or non-evolving (steady) base > flow. As for the latter, we may use restart option in rea file. The repo version is for linear adjoint, i.e. for a non-evolving base flows. > 2) For non-evolving case (which is what I have in mind in the end) Nek > "is able to" handle and to do so parameter #31 in rea file needs to be > modified. Can you please clarify how. Suppose that there is velocity and > a passive scalar or even temperature, so there are three adjoint fields > (velocity, pressure and the scalar). The repo version only has the adjoint equations implemented for the velocity. The temperature adjoint has first been implemented at ANL, and we have further validated it. We hope that we can commit it back to the repo within a few weeks. Parameter 31 gives the number of disturbance fields, negative for a fixed base flow. For typical adjoint runs it should be -1 (one adjoint variable for fixed base flow). > Also, is what Oana mentioned, "one has to do the change of the > velocities himself" related to remark number 2 above? Exactly, for a changing base flow, one had to do that manually. > 3) There's also a confusion in userchk: First Adam mentioned that we > need to include 'ADJOINT' plus if(ISTEP.eq.0) ADJOINT=.TRUE. for adjoint > solver in order to i) changing the time and pressure sign and 2) > modification of advection terms. But in the second example, while he > still uses include 'ADJOINT' in userchk but then there is this statement > IFADJ = .FALSE. Can you please clarify what's the difference here? It was a typo, the relevant variable is IFADJ and nothing else. This one had to be true or false depending on the equation to be solved. > I'm really grateful that you take the time to answer these questions. > Please let me know if we can take this conversation off the mailing list. > > Brief summary of the problem we have in mind: we'd like to optimize the > flow with respect to some parameters (either BCs or ICs) in terms of a > defined cost function. So this can be regarded as either design or > optimal control problem. Calculus of variations and Lagrange multipliers > lead to adjoint equations which are very similar to NS equations. We > already solved the problem using NEK and we'd like to use this again for > adjoint equations. All the results will be in an outer loop for > optimization (based on, say, steepest gradient). In general, the problem > is unsteady, but we can start off with steady and non-evolving cases at > early stages. We have calculated optimal initial conditions etc. using Nek, see e.g. the results of the jet in crossflow, http://www.sciencedirect.com/science/article/pii/S0997754614001009. Is your intended optimisation similar to that? Best regards, Philipp > > Best, > Saleh > > On 2/8/2016 4:26 AM, nek5000-users at lists.mcs.anl.gov wrote: >> Hi Saleh >> >> If I understand you correctly, you ask about a base flow, i.e. the >> velocity field you do your linearisation about (the one you use to >> advect prturbation). If this is the case, than it is possible to do >> with nek. Otherwise you have to wait for the code Oana mentioned. What >> we do in our stability calculations, we take the solution of >> non-linear NS (steady state reached with e.g. SFD) and next we load it >> into nek as non-linear filed. One can do it by setting restart options >> in ###.rea file. It should look like: >> 3.33333 3.33333 -0.833333 -1.16667 >> XFAC,YFAC,XZERO,YZERO >> **MESH DATA** 1st line is X of corner 1,2,3,4. 2nd line is Y. >> -1472 2 1472 NELT,NDIM,NELV >> 1 PRESOLVE/RESTART OPTIONS ***** >> bfext_cyl?.f00000 >> 7 INITIAL CONDITIONS ***** >> C Default >> C Default >> >> >> Where bfext_cyl0.f00000 is our restart file (we usually save this >> velocity field in double precision). Next you have to set linear >> solver with non-evolving base flow (nek is able to evolve both so be >> careful). You freeze baseflow by setting negative value of parameter >> 31 in ###.rea: >> 0.00000 p030 > 0 ==> properties set in uservp() >> -1.00000 p031 NPERT: #perturbation modes >> 0.00000 p032 #BCs in re2 file, if > 0 >> >> Nek is able of evolving number of perturbation fields set by >> abs(param(31)) (in ###.rea) and lpert in SIZE. I have to mention that >> nek by itself outputs the non-linear fields only, so to see >> perturbation you should add to your userchk something like this >> >> >> subroutine userchk >> >> include 'SIZE' ! NX1, NY1, NZ1, NELV, NIO >> include 'TSTEP' ! ISTEP, IOSTEP >> include 'SOLN' ! V[XYZ]P, TP >> include 'INPUT' ! IF3D, PARAM >> include 'ADJOINT' ! IFADJ >> >> ! direct or adjoint run >> IFADJ = .FALSE. >> >> ! write perturbation field >> if (mod(ISTEP,int(PARAM(15))).eq.0) then >> call outpost2(VXP,VYP,VZP,PRP,TP,0,'prt') >> endif >> >> I hope it is an answer to your question. >> regards >> adam >> >> On 2016-02-06 19:15, nek5000-users at lists.mcs.anl.gov wrote: >>> Hi Saleh, >>> >>> There is no easily accessible implementation for this, one has to do >>> the change of the velocities himself. In fact we have ongoing work on >>> this and once we are done I guess we can share it. >>> What kind of case are you looking at, long time horizons are quite >>> tricky to handle. We can take this conversation off the mailing list >>> if needed. >>> >>> Oana >>> >>> >>> >>> ________________________________________ >>> From: nek5000-users-bounces at lists.mcs.anl.gov >>> [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of >>> nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >>> Sent: Friday, February 05, 2016 2:15 PM >>> To: nek5000-users at lists.mcs.anl.gov >>> Subject: Re: [Nek5000-users] Clarification on Adjoint solver >>> >>> Adams, >>> >>> I really appreciate your detailed answer. I definitely clarified the Nek >>> functionality for adjoint solvers. One question remains is how to >>> include the velocity of the forward solver into adjoint equations. I >>> assume this is also the case when you do linear stability analysis with >>> Nek, since the velocity in convective terms is not completely part of >>> the solution. Mathematically, my question is: >>> >>> in $V. \nabla u_a$ term, as the advection, how can we substitute $V$ (as >>> the forward velocity which is already solved for in a regular NS >>> solution) in adjoint equations. suppose that the velocity, along with >>> other quantities are saved in afield out put such as fld.f0000X for X >>> time-step. So we have access to that velocity but how can we include it >>> in adjoint equations? >>> >>> Many thanks again, >>> Saleh >>> >>> On 2/2/2016 3:09 AM, nek5000-users at lists.mcs.anl.gov wrote: >>>> Hi >>>> >>>> The adjoint implementation in nek assumes change of the direction of >>>> time evolution. The goal is to use the standard nek solver with >>>> minimal modifications (only advection terms have to be changed). You >>>> get these equations from you standard one by changing the time and >>>> pressure sign (t -> -t; p-> -p). This of course changes your time >>>> integration limits, but it shouldn't be a problem. So when you have >>>> your linear direct simulation, to turn it into dual one it is enough >>>> to include in your userchk include files ADJOINT and TSTEP, and for >>>> ISTEP.eq.0 set IFADJOINT to true. Something like: >>>> subroutine userchk >>>> >>>> include 'SIZE' >>>> include 'TSTEP' ! ISTEP >>>> include 'ADJOINT' ! IFADJ >>>> >>>> if(ISTEP.eq.0) ADJOINT=.TRUE. >>>> >>>> There is no need to modify ###.rea or ###.map files. The only problem >>>> are boundary conditions for open flows, as they would require >>>> non-homogeneous bc. In this case to avoid problems we usually use >>>> sufficiently large domain together with zero Dirichlet bc. The last >>>> issue is sufficient resolution for both direct and adjoin runs. I hope >>>> this gives you some idea about nek implementation. >>>> Regards >>>> Adam >>>> >>>> On 2016-02-01 22:17, nek5000-users at lists.mcs.anl.gov wrote: >>>>> Hi Neks, >>>>> >>>>> To my knowledge, Nek5000 is able to solve adjoint equations of the >>>>> form: >>>>> $\partial u_a/\partial t + V.\nabla u_a - u_a.(\nabla V)^T + \nabla >>>>> \p_a + 1/Re \nabla^2 u_a=0$ and $\nabla.u_a=0$ >>>>> where u_a and p_a are adjoint velocity and pressure. (Slightly >>>>> different formulation may be seen in literature). >>>>> >>>>> Let's assume the forward problem is already solved so that $V$, i.e. >>>>> velocity, (and possibly $p$ as the pressure) is (are) already know. >>>>> Thus, the initial and boundary conditions for adjoint NS are also >>>>> known. How can we now solve the dual/adjoint NS problem with Nek5000? >>>>> Is there any example on how to modify .rea files? Specailly, the >>>>> "convection term of $V.\nabla u_a - u_a.(\nabla V)^T$ needs to be >>>>> modified compared to forward NS but am not sure how. >>>>> My search of mailing list entails some modifications in "perturb.f" >>>>> subroutine, but a little bit more clarification on that would be >>>>> really appreciated. >>>>> >>>>> Yours, >>>>> Saleh >>>>> _______________________________________________ >>>>> Nek5000-users mailing list >>>>> Nek5000-users at lists.mcs.anl.gov >>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>>> _______________________________________________ >>>> Nek5000-users mailing list >>>> Nek5000-users at lists.mcs.anl.gov >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Wed Feb 10 16:46:11 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 10 Feb 2016 23:46:11 +0100 Subject: [Nek5000-users] SYM boundary conditions In-Reply-To: References: , Message-ID: The flow is turbulent in the real case. It is also an external flow... About damping the turbulence using a forcing, do you mean that I could add something like ffx=-sin(omega*time) in the userf subroutine (being ux=sin(omega*time))? Or do you mean that I should use a selective frequency damping subroutine? About turn_outflow() routine, it is working when the flow is external? Does it make sense using SYM in the sides of the domain and ON in the outlet surface? Or the idea is using ON in both, the sides and the outlet surface? Thank you very much for your help. Cheers SL El 10-02-2016 15:36, nek5000-users at lists.mcs.anl.gov escribi?: > Is this an internal flow, or external flow? > > If internal, you can use the turn_outflow() routine with sufficient > divergence to ensure > that the flow is always leaving the domain. We use this for some of > our moving domain > simulations. > > Paul > > ________________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov > [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of > nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > Sent: Wednesday, February 10, 2016 8:04 AM > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] SYM boundary conditions > > ok, I see. If it is turbulent, then it is a bit more difficult. You can > try ON, but perhaps it will not work. But in the real case, what type > of > inflow are you going to have, something turbulent? What about damping > the turbulence before it actually reaches the outflow boundary, e.g. > using a forcing? ONce it is laminar, I think that ON could work. > > Philipp > > On 2016-02-10 13:02, nek5000-users at lists.mcs.anl.gov wrote: >> Yes, it is turbulent... >> >> I will try to use ON. Please, could you help me to understand a bit >> more >> why it could work? >> >> Thanks again! >> Cheers >> SL >> >> >> El 10-02-2016 12:17, nek5000-users at lists.mcs.anl.gov escribi?: >>> one initial suggestion would be to use ON at the outflow. Is the flow >>> turbulent at the outflow boundary? >>> Philipp >>> >>> On 2016-02-10 11:45, nek5000-users at lists.mcs.anl.gov wrote: >>>> Hi, >>>> >>>> Thank you. I am trying to simulate a flow whose inlet boundary >>>> condition >>>> is time periodic ux=sin(omega*time). Since there is an injection and >>>> a >>>> suction, and the mean flow is zero, I think that the boundary >>>> conditions >>>> should let the transpiration/penetration of the flow... (I have >>>> tried to >>>> run the simulation using outflow BC but it crashes). Is there any BC >>>> condition in Nek5000 that suits to this problem? >>>> >>>> Thank you for your help in advance. >>>> Best regards, >>>> SL >>>> >>>> El 10-02-2016 10:20, nek5000-users at lists.mcs.anl.gov escribi?: >>>>> Hi, >>>>> SYM is essentially means that the velocity through the interface is >>>>> zero and for the inplane velocities that the normal gradients >>>>> vanish. >>>>> So it does not allow transpiration/penetration through the >>>>> boundary. >>>>> So I suspect that this condition is not what you are looking for. >>>>> >>>>> However, I am not exactly sure what you really mean. Perhaps you >>>>> can >>>>> say a bit more about your case? >>>>> >>>>> Best regards, >>>>> Philipp >>>>> >>>>> On 2016-02-10 08:02, nek5000-users at lists.mcs.anl.gov wrote: >>>>>> Hi Neks, >>>>>> >>>>>> I am looking for a boundary condition that leads the flow to go >>>>>> out or >>>>>> to go into the domain. >>>>>> >>>>>> Is 'SYM' valid for this purpose? >>>>>> >>>>>> >>>>>> I think that what I am looking for is, in the outlet surface: >>>>>> >>>>>> - If the velocity vector at the outlet points out of the >>>>>> domain, >>>>>> then >>>>>> the boundary condition for the velocity will be of the Neumann >>>>>> type >>>>>> >>>>>> - However, if the velocity at the outlet points into the >>>>>> domain, >>>>>> then >>>>>> we can imagine that the outflow is no longer an outflow, but an >>>>>> inflow, >>>>>> and that we therefore would like to specify a Dirichlet boundary >>>>>> condition for velocity >>>>>> >>>>>> >>>>>> Thanks in advance >>>>>> SL >>>>>> >>>>>> _______________________________________________ >>>>>> Nek5000-users mailing list >>>>>> Nek5000-users at lists.mcs.anl.gov >>>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>>>> _______________________________________________ >>>>> Nek5000-users mailing list >>>>> Nek5000-users at lists.mcs.anl.gov >>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>>> >>>> >>>> >>>> _______________________________________________ >>>> Nek5000-users mailing list >>>> Nek5000-users at lists.mcs.anl.gov >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >> >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Wed Feb 10 17:51:31 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 10 Feb 2016 18:51:31 -0500 Subject: [Nek5000-users] visnek Message-ID: hey i was trrying out the eddy example i am able to compile and run it. i installed visit 2.10.0 and then was abel to open visit as well. However when i am trying to give the command '~/nek5_svn/trunk/tools/scripts/visnek eddy_uv' i get the following error: bash: /home//nek5_svn/trunk/tools/scripts/visnek: /bin/tcsh: bad interpreter: No such file or directory. i am not able to create a .nek5000 file for visit. plz help AVR -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Feb 10 19:40:08 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 10 Feb 2016 20:40:08 -0500 Subject: [Nek5000-users] visnek In-Reply-To: References: Message-ID: Hi AVR, If I remember correctly, you need tcsh for visnek to work. I believe this error fix is already in the archive as well. Ubuntu uses apt-get install tcsh David On Feb 10, 2016 6:51 PM, wrote: > hey i was trrying out the eddy example > i am able to compile and run it. i installed visit 2.10.0 and then was > abel to open visit as well. However when i am trying to give the command > '~/nek5_svn/trunk/tools/scripts/visnek eddy_uv' i get the following error: > bash: /home//nek5_svn/trunk/tools/scripts/visnek: /bin/tcsh: bad > interpreter: No such file or directory. > i am not able to create a .nek5000 file for visit. > plz help > > AVR > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Feb 10 22:50:23 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 10 Feb 2016 23:50:23 -0500 Subject: [Nek5000-users] 1. help : compiling error In-Reply-To: References: Message-ID: mansi? Hi... In general, things will compile with a lot of warnings about array sizes, etc..; don't know if there are any specific problems in your case or not... The compiler failure error is what is stopping the final compile and link. Not sure, but this could be a compiler/library version problem. I seem to remember that I had problems in the past when the version of Linux and the gcc/gfortran compiler was too old... Ask your system admin to do a "yum update ; yum upgrade". In some cases, the version of RHEL/CentOS/Scientific Linux is so old that the latest compiler available for that version is still too old. If that's the case, you must install and use your own version of a newer gfortran - this is very difficult because of all the dependencies, references, and libraries required; or, upgrade the base system version to the latest - much easier, but you may have problems getting an admin to do this ;-). If you control the machine, do it yourself... Note that this problem was referenced in another NEK5000 mailing list entry; do a search of the archives to find it using "...internal compiler error: in expand_expr_real_1" See Ya, Murph -- Murphy Leo O'Dea PhD Candidate Oakland University mlodea at oakland.edu -------------------------------------------------------------------------------------------------------------------------------------------- s23 = sij(j1,j2,1,5,e) 1 Warning: Array reference at (1) is out of bounds gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/fasts.f -o obj/fasts.o /home/users/mansi/nek5_svn/trunk/nek/navier8.f: In function ?setvert3d?: /home/users/mansi/nek5_svn/trunk/nek/navier8.f:2354: internal compiler error: in expand_expr_real_1, at expr.c:6846 Please submit a full bug report, with preprocessed source if appropriate. See for instructions. make: *** [obj/navier8.o] Error 1 make: *** Waiting for unfinished jobs.... In file /home/users/mansi/nek5_svn/trunk/nek/fasts.f:343 -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Feb 10 23:46:49 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 11 Feb 2016 13:46:49 +0800 Subject: [Nek5000-users] inlet points Message-ID: Hi neks, I am trying to incorporate the vortex method to generate inlet turbulence. The fluctuations are follow the Biot-Savart law which has a form of the greens function ie x-x_i. the vortex core is placed at x_i. At the inlet of say a box computational domain, my plane is the Y-Z plane. Let's say I have a vortex core at the point (y,z)=(-2,5). At the rest of the nodes of my inlet plane I would need to have (Y,Z) - (2,5). How is this done in the usr file? I dont think i can use xm1(i,1,1,1) because I only want the points at the inlet and not the whole domain. Can anybody help please? Cheers Rif -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Feb 11 08:35:23 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 11 Feb 2016 14:35:23 +0000 Subject: [Nek5000-users] Link Fortran90 code via Fortran Interface Message-ID: Hi Neks, Is it possible to link Fortran90 code via the Fortran Interface (b3d.usr). Best regards, Philipp Weiss -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Feb 11 11:00:25 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 11 Feb 2016 17:00:25 +0000 Subject: [Nek5000-users] Link Fortran90 code via Fortran Interface In-Reply-To: References: Message-ID: Hi Philipp, You should be able to define subprograms in a separate F90 file and call them from the usr file. Then, just compile your new file and add it to the link line using the "USR_LFLAGS" option in makenek. However, it could be a bit tricky to access global data stored Nek5000's common blocks, so I'd strongly suggest making those routines self-contained with all data passing via arguments. This approach can be used to link in C or C++ code as well, as they are all object compatible. The usr file itself needs to be F77, if for no other reason than the include files ('SIZE', for example) are in F77 format. Depending on your problem, you may be able to use NekBox, which is written in F90 and uses F90 for the usr file. NekBox is more of a prototyping code than a production code, so its functionality is limited. See: https://github.com/Nek5000/NekBox Max On Thu, Feb 11, 2016 at 8:35 AM wrote: > Hi Neks, > > Is it possible to link Fortran90 code via the Fortran Interface (b3d.usr). > > Best regards, Philipp Weiss > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Feb 11 11:21:47 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 11 Feb 2016 22:51:47 +0530 Subject: [Nek5000-users] Need for clarification - source term Message-ID: Dear Sir, In the userq() subroutine: I wish to include the source term as density*amplitude*omega*omega*cos(omega*time) How shall I do this? T/R Prashanth Mohanasundaram -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Feb 11 16:06:56 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 11 Feb 2016 23:06:56 +0100 Subject: [Nek5000-users] Link Fortran90 code via Fortran Interface In-Reply-To: References: Message-ID: I guess you could just use common blocks also in the Fortran90 code. Philipp On 2016-02-11 18:00, nek5000-users at lists.mcs.anl.gov wrote: > Hi Philipp, > > You should be able to define subprograms in a separate F90 file and call > them from the usr file. Then, just compile your new file and add it to > the link line using the "USR_LFLAGS" option in makenek. However, it > could be a bit tricky to access global data stored Nek5000's common > blocks, so I'd strongly suggest making those routines self-contained > with all data passing via arguments. This approach can be used to link > in C or C++ code as well, as they are all object compatible. > > The usr file itself needs to be F77, if for no other reason than the > include files ('SIZE', for example) are in F77 format. > > Depending on your problem, you may be able to use NekBox, which is > written in F90 and uses F90 for the usr file. NekBox is more of a > prototyping code than a production code, so its functionality is > limited. See: > https://github.com/Nek5000/NekBox > > Max > > On Thu, Feb 11, 2016 at 8:35 AM > wrote: > > Hi Neks, > > Is it possible to link Fortran90 code via the Fortran Interface > (b3d.usr). > > Best regards, Philipp Weiss > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > From nek5000-users at lists.mcs.anl.gov Thu Feb 11 16:22:52 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 11 Feb 2016 22:22:52 +0000 Subject: [Nek5000-users] Link Fortran90 code via Fortran Interface In-Reply-To: References: , Message-ID: Hi Philipp, If you only want to link a patch you can do so by modifying your makenek and makefile_usr.inc as in the stability toolbox for example https://nek5000.mcs.anl.gov/toolboxes/ Oana ________________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] Sent: Thursday, February 11, 2016 4:06 PM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] Link Fortran90 code via Fortran Interface I guess you could just use common blocks also in the Fortran90 code. Philipp On 2016-02-11 18:00, nek5000-users at lists.mcs.anl.gov wrote: > Hi Philipp, > > You should be able to define subprograms in a separate F90 file and call > them from the usr file. Then, just compile your new file and add it to > the link line using the "USR_LFLAGS" option in makenek. However, it > could be a bit tricky to access global data stored Nek5000's common > blocks, so I'd strongly suggest making those routines self-contained > with all data passing via arguments. This approach can be used to link > in C or C++ code as well, as they are all object compatible. > > The usr file itself needs to be F77, if for no other reason than the > include files ('SIZE', for example) are in F77 format. > > Depending on your problem, you may be able to use NekBox, which is > written in F90 and uses F90 for the usr file. NekBox is more of a > prototyping code than a production code, so its functionality is > limited. See: > https://github.com/Nek5000/NekBox > > Max > > On Thu, Feb 11, 2016 at 8:35 AM > wrote: > > Hi Neks, > > Is it possible to link Fortran90 code via the Fortran Interface > (b3d.usr). > > Best regards, Philipp Weiss > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Thu Feb 11 16:25:35 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 11 Feb 2016 22:25:35 +0000 Subject: [Nek5000-users] Link Fortran90 code via Fortran Interface In-Reply-To: References: Message-ID: Sure, but that would have to be done explicitly, not via `include` statements, because the include files contain F77-style comments and line continuations. Then, if one of those common block names changes in the include file, the compiler interprets your common block as being a new declaration and you get a silent bug. The include files seem to be relatively stable, so just be careful... On Thu, Feb 11, 2016 at 4:07 PM wrote: > I guess you could just use common blocks also in the Fortran90 code. > > Philipp > > On 2016-02-11 18:00, nek5000-users at lists.mcs.anl.gov wrote: > > Hi Philipp, > > > > You should be able to define subprograms in a separate F90 file and call > > them from the usr file. Then, just compile your new file and add it to > > the link line using the "USR_LFLAGS" option in makenek. However, it > > could be a bit tricky to access global data stored Nek5000's common > > blocks, so I'd strongly suggest making those routines self-contained > > with all data passing via arguments. This approach can be used to link > > in C or C++ code as well, as they are all object compatible. > > > > The usr file itself needs to be F77, if for no other reason than the > > include files ('SIZE', for example) are in F77 format. > > > > Depending on your problem, you may be able to use NekBox, which is > > written in F90 and uses F90 for the usr file. NekBox is more of a > > prototyping code than a production code, so its functionality is > > limited. See: > > https://github.com/Nek5000/NekBox > > > > Max > > > > On Thu, Feb 11, 2016 at 8:35 AM > > wrote: > > > > Hi Neks, > > > > Is it possible to link Fortran90 code via the Fortran Interface > > (b3d.usr). > > > > Best regards, Philipp Weiss > > _______________________________________________ > > Nek5000-users mailing list > > Nek5000-users at lists.mcs.anl.gov Nek5000-users at lists.mcs.anl.gov> > > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > > > > > > _______________________________________________ > > Nek5000-users mailing list > > Nek5000-users at lists.mcs.anl.gov > > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Fri Feb 12 14:10:51 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 12 Feb 2016 20:10:51 +0000 Subject: [Nek5000-users] Nek5000-users Digest, Vol 84, Issue 13 In-Reply-To: References: Message-ID: Hello Murph, I tried doing the which gfortran and gfortran -v.I also tried the yum update and yum upgrade. All I got is nothing. I am so badly stuck nothing is working.[mansi at maea203-01 eddy]$ which gfortran /usr/bin/gfortran [mansi at maea203-01 eddy]$ gfortran v- gfortran: v-: No such file or directory [mansi at maea203-01 eddy]$ gfortran -v Using built-in specs. Target: x86_64-redhat-linux Configured with: ../configure --prefix=/usr --mandir=/usr/share/man --infodir=/usr/share/info --enable-shared --enable-threads=posix --enable-checking=release --with-system-zlib --enable-__cxa_atexit --disable-libunwind-exceptions --enable-libgcj-multifile --enable-languages=c,c++,objc,obj-c++,java,fortran,ada --enable-java-awt=gtk --disable-dssi --enable-plugin --with-java-home=/usr/lib/jvm/java-1.4.2-gcj-1.4.2.0/jre --with-cpu=generic --host=x86_64-redhat-linux Thread model: posix gcc version 4.1.2 20071124 (Red Hat 4.1.2-42) [mansi at maea203-01 eddy]$ From nek5000-users at lists.mcs.anl.gov Mon Feb 15 05:38:23 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 15 Feb 2016 17:08:23 +0530 Subject: [Nek5000-users] How to include high reynolds number Message-ID: Dear Sir Whenever I give reynolds number as 20000, my simulation stops running after few timesteps. The logfile shows some error related to Courrant number. Kindly help me out in solving this issue. T/R Prashanth Mohanasundaram -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Feb 15 05:52:18 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 15 Feb 2016 17:22:18 +0530 Subject: [Nek5000-users] Clarification on including high reynolds number Message-ID: Dear Sir Whenever I give reynolds number as 20000, my simulation stops running after few timesteps. The logfile shows some error related to Courrant number. Kindly help me out in solving this issue. T/R Prashanth Mohanasundaram -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Feb 15 06:02:00 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 15 Feb 2016 13:02:00 +0100 Subject: [Nek5000-users] Clarification on including high reynolds number In-Reply-To: References: Message-ID: Dear Prashanth What did you do to investigate your problem? I guess you checked the most obvious reasons like time and spatial resolution already? Cheers Stefan > On 15.02.2016, at 12:52, nek5000-users at lists.mcs.anl.gov wrote: > > Dear Sir > > Whenever I give reynolds number as 20000, my simulation stops running after few timesteps. The logfile shows some error related to Courrant number. Kindly help me out in solving this issue. > > T/R > Prashanth Mohanasundaram > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Feb 17 08:14:43 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 17 Feb 2016 15:14:43 +0100 Subject: [Nek5000-users] Mass flow rate and averages Message-ID: Hi neks, maybe my, will be noob questions, but i am a nek noob. I am running a DNS of channel with riblets, now i need to know the averages fields and the mass flow rate. For the averages fields i have seen in turbChannel example folder, and i have seen in turbchannel.usr that there is a subroutine that do it. But i didn't understand, if i set fintime(P10) and iotime(P14) and P68 in rea file, have i always the output?Because in turbchannel example, if i set how i already told, i have errors in compilation. And the second question is related to the mass flow rate. If i set forcing, can i know the relative mass flow rate? Or viceversa, if i set the mass flow rate in .rea file, can i know the forcing? Where can i read or compute it? Best regards From nek5000-users at lists.mcs.anl.gov Wed Feb 17 08:17:13 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 17 Feb 2016 15:17:13 +0100 Subject: [Nek5000-users] ERROR of compilating Message-ID: Hi all, I have a compilation issue with my example. I created a mesh 3D using n2to3 from the file w2dcyl020a.rea. I don't have any issue with compilation. Then I refined the mesh 2D using pretex and rebuilt the mesh 3D with n2to3 like before this and this time an error message occured: "obj/stenosis.o: In function `uservp_': stenosis.f:(.text+0xd): relocation truncated to fit: R_X86_64_PC32 against symbol `nekuse_' defined in COMMON section in obj/stenosis.o stenosis.f:(.text+0x1b): relocation truncated to fit: R_X86_64_PC32 against symbol `nekuse_' defined in COMMON section in obj/stenosis.o stenosis.f:(.text+0x29): relocation truncated to fit: R_X86_64_PC32 against symbol `nekuse_' defined in COMMON section in obj/stenosis.o obj/stenosis.o: In function `userf_': stenosis.f:(.text+0x35): relocation truncated to fit: R_X86_64_PC32 against symbol `nekuse_' defined in COMMON section in obj/stenosis.o stenosis.f:(.text+0x3c): relocation truncated to fit: R_X86_64_PC32 against symbol `nekuse_' defined in COMMON section in obj/stenosis.o stenosis.f:(.text+0x43): relocation truncated to fit: R_X86_64_PC32 against symbol `nekuse_' defined in COMMON section in obj/stenosis.o obj/stenosis.o: In function `userq_': stenosis.f:(.text+0x53): relocation truncated to fit: R_X86_64_PC32 against symbol `nekuse_' defined in COMMON section in obj/stenosis.o obj/stenosis.o: In function `userchk_': stenosis.f:(.text+0x73): relocation truncated to fit: R_X86_64_PC32 against symbol `dimn_' defined in COMMON section in obj/stenosis.o stenosis.f:(.text+0x189): relocation truncated to fit: R_X86_64_PC32 against symbol `dimn_' defined in COMMON section in obj/stenosis.o stenosis.f:(.text+0x1e7): relocation truncated to fit: R_X86_64_PC32 against symbol `dimn_' defined in COMMON section in obj/stenosis.o obj/stenosis.o: In function `userbc_': stenosis.f:(.text+0x2b4): additional relocation overflows omitted from the output collect2: ld a retourn? 1 code d'?tat d'ex?cution make: *** [nek5000] Erreur 1" The same issue occures if I change the parameters lx1 and lxd in SIZE-file. Does anybody have an idea on how to fix this? Thank you for any help or suggestion. *Andrew* -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Feb 17 09:22:34 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 17 Feb 2016 15:22:34 +0000 Subject: [Nek5000-users] Mass flow rate and averages In-Reply-To: References: Message-ID: If you set the mass flow (p54-55) in your periodic channel, the forcing as a function of time can be found via: grep vol logfile will show lines like: 1989 0.1989000E+01 3.72046E-02 3.99058E-05 1.99996E+00 2.00000E+00 volflow X Column 1 is step number, Column 2 is time. Column 3 is the mean (in space) pressure gradient, as function of time. This is the pressure gradient required to sustain a fix mass flow rate. If you wish to instead prescribe mean pressure gradient, set p54 and 55 = 0 in the .rea file and set ffx = 1 (say) in the .usr file. To get the volumetric flow rate, you can do the following in userchk: n = nx1*ny1*nz1*nelv volf = glsc3(vx,bm1,n)/xlength where xlength is the domain length. (This example assumes your flow is in the x direction. You can get xlength via: call domain_size(xmin,xmax,ymin,ymax,zmin,zmax) xlength = xmax-xmin Paul ________________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] Sent: Wednesday, February 17, 2016 8:14 AM To: nek5000-users at lists.mcs.anl.gov Subject: [Nek5000-users] Mass flow rate and averages Hi neks, maybe my, will be noob questions, but i am a nek noob. I am running a DNS of channel with riblets, now i need to know the averages fields and the mass flow rate. For the averages fields i have seen in turbChannel example folder, and i have seen in turbchannel.usr that there is a subroutine that do it. But i didn't understand, if i set fintime(P10) and iotime(P14) and P68 in rea file, have i always the output?Because in turbchannel example, if i set how i already told, i have errors in compilation. And the second question is related to the mass flow rate. If i set forcing, can i know the relative mass flow rate? Or viceversa, if i set the mass flow rate in .rea file, can i know the forcing? Where can i read or compute it? Best regards _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Wed Feb 17 09:26:37 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 17 Feb 2016 15:26:37 +0000 Subject: [Nek5000-users] ERROR of compilating In-Reply-To: References: Message-ID: Hi Andrew, It looks like you are trying to compile for too large a case. Two things to look for --- 1) Use the -mcmodel=medium flag in makenek Go to G= in makenek, uncomment, and change it to G="-mcmodel=medium" 2) What is your lelt value in SIZE ? If you have (say) 100 elements and are running on 8 processors, you would only need lelt=100/8 = 13 or 14 (Even numbers are preferred for byte-alignment reasons.) You would not need lelt=100. This change would reduce your memory footprint by 8x. Paul ________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] Sent: Wednesday, February 17, 2016 8:17 AM To: nek5000-users at lists.mcs.anl.gov Subject: [Nek5000-users] ERROR of compilating Hi all, I have a compilation issue with my example. I created a mesh 3D using n2to3 from the file w2dcyl020a.rea. I don't have any issue with compilation. Then I refined the mesh 2D using pretex and rebuilt the mesh 3D with n2to3 like before this and this time an error message occured: "obj/stenosis.o: In function `uservp_': stenosis.f:(.text+0xd): relocation truncated to fit: R_X86_64_PC32 against symbol `nekuse_' defined in COMMON section in obj/stenosis.o stenosis.f:(.text+0x1b): relocation truncated to fit: R_X86_64_PC32 against symbol `nekuse_' defined in COMMON section in obj/stenosis.o stenosis.f:(.text+0x29): relocation truncated to fit: R_X86_64_PC32 against symbol `nekuse_' defined in COMMON section in obj/stenosis.o obj/stenosis.o: In function `userf_': stenosis.f:(.text+0x35): relocation truncated to fit: R_X86_64_PC32 against symbol `nekuse_' defined in COMMON section in obj/stenosis.o stenosis.f:(.text+0x3c): relocation truncated to fit: R_X86_64_PC32 against symbol `nekuse_' defined in COMMON section in obj/stenosis.o stenosis.f:(.text+0x43): relocation truncated to fit: R_X86_64_PC32 against symbol `nekuse_' defined in COMMON section in obj/stenosis.o obj/stenosis.o: In function `userq_': stenosis.f:(.text+0x53): relocation truncated to fit: R_X86_64_PC32 against symbol `nekuse_' defined in COMMON section in obj/stenosis.o obj/stenosis.o: In function `userchk_': stenosis.f:(.text+0x73): relocation truncated to fit: R_X86_64_PC32 against symbol `dimn_' defined in COMMON section in obj/stenosis.o stenosis.f:(.text+0x189): relocation truncated to fit: R_X86_64_PC32 against symbol `dimn_' defined in COMMON section in obj/stenosis.o stenosis.f:(.text+0x1e7): relocation truncated to fit: R_X86_64_PC32 against symbol `dimn_' defined in COMMON section in obj/stenosis.o obj/stenosis.o: In function `userbc_': stenosis.f:(.text+0x2b4): additional relocation overflows omitted from the output collect2: ld a retourn? 1 code d'?tat d'ex?cution make: *** [nek5000] Erreur 1" The same issue occures if I change the parameters lx1 and lxd in SIZE-file. Does anybody have an idea on how to fix this? Thank you for any help or suggestion. Andrew -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Feb 17 09:52:42 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 17 Feb 2016 10:52:42 -0500 Subject: [Nek5000-users] gfortran version and issues In-Reply-To: References: Message-ID: Mansi, Hi... This is a VERY old version of gfortran... I'm using an old version and it's 4.4.x. The current 4.x.x. version is 4.9.1, and version 5 is out. You need to figure out what Linux distribution and version you are on; what rights you have on that machine; and how to get it upgraded to the latest version of Linux and gfortran. I don't remember the details, but I know when I started, I couldn't get it to compile with gfortran older than about 4.3.x... See Ya, Murph -- Murphy Leo O'Dea PhD Student Oakland University mlodea at oakland.edu ----------------------------------------------------------------------------------------------------------------------------------------------- Hello Murph, I tried doing the which gfortran and gfortran -v.I also tried the yum update and yum upgrade. All I got is nothing. I am so badly stuck nothing is working.[mansi at maea203-01 eddy]$ which gfortran /usr/bin/gfortran [mansi at maea203-01 eddy]$ gfortran v- gfortran: v-: No such file or directory [mansi at maea203-01 eddy]$ gfortran -v Using built-in specs. Target: x86_64-redhat-linux Configured with: ../configure --prefix=/usr --mandir=/usr/share/man --infodir=/usr/share/info --enable-shared --enable-threads=posix --enable-checking=release --with-system-zlib --enable-__cxa_atexit --disable-libunwind-exceptions --enable-libgcj-multifile --enable-languages=c,c++,objc,obj-c++,java,fortran,ada --enable-java-awt=gtk --disable-dssi --enable-plugin --with-java-home=/usr/lib/jvm/java-1.4.2-gcj-1.4.2.0/jre --with-cpu=generic --host=x86_64-redhat-linux Thread model: posix gcc version 4.1.2 20071124 (Red Hat 4.1.2-42) [mansi at maea203-01 eddy]$ From nek5000-users at lists.mcs.anl.gov Wed Feb 17 14:37:21 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 17 Feb 2016 15:37:21 -0500 Subject: [Nek5000-users] Clarification on Adjoint solver In-Reply-To: References: Message-ID: Hi Philipp , Many thanks for your replies. I think what we have in mind is very similar to what has been done in the paper you provided for the JCF, with one major difference. In our problem an optimal disturbance with a finite amplitude is sought after, which results in a time-dependent U_b. >From previous comments looks like it can't be done automatically in Nek (and the user needs to implement this manually). Does this manual way mean the user has to load velocity at each time step (say in a loop that executes the Nek solver)? Is there any advice on how can one implement the classical check-pointing approach? My second question is boundary condition. From Nek literature, it seems line "inhomogeneous" boundary conditions can't be implemented (which is usually the case for the outlet). Do you confirm that? In JCF paper you provided I see $P_{a}n+1/Re \nabla u_{a}.n=(U_{b}.n)u_{a}$ with u_{a} and p_{a} as adjoint variables is considered as inhomogeneous. Neglecting the pressure term, this BC more looks like a Robin type to me. So what's the jargon here for inhomogeneous BC? Last question, is equation (7) of that paper the kind of equation solved by Nek? Or one should change the terms slightly in the source code? Many thanks, Saleh On 2/10/2016 4:10 PM, nek5000-users at lists.mcs.anl.gov wrote: > Dear all, > > Let me add some answers below. > > On 2016-02-10 18:58, nek5000-users at lists.mcs.anl.gov wrote: >> Adam and Oana, >> Thanks again for your responses. >> Let me summarize what I understood. >> 1) We may have either evolving (unsteady) or non-evolving (steady) base >> flow. As for the latter, we may use restart option in rea file. > > The repo version is for linear adjoint, i.e. for a non-evolving base > flows. > >> 2) For non-evolving case (which is what I have in mind in the end) Nek >> "is able to" handle and to do so parameter #31 in rea file needs to be >> modified. Can you please clarify how. Suppose that there is velocity and >> a passive scalar or even temperature, so there are three adjoint fields >> (velocity, pressure and the scalar). > > The repo version only has the adjoint equations implemented for the > velocity. The temperature adjoint has first been implemented at ANL, > and we have further validated it. We hope that we can commit it back > to the repo within a few weeks. > Parameter 31 gives the number of disturbance fields, negative for a > fixed base flow. For typical adjoint runs it should be -1 (one adjoint > variable for fixed base flow). > >> Also, is what Oana mentioned, "one has to do the change of the >> velocities himself" related to remark number 2 above? > > Exactly, for a changing base flow, one had to do that manually. > >> 3) There's also a confusion in userchk: First Adam mentioned that we >> need to include 'ADJOINT' plus if(ISTEP.eq.0) ADJOINT=.TRUE. for adjoint >> solver in order to i) changing the time and pressure sign and 2) >> modification of advection terms. But in the second example, while he >> still uses include 'ADJOINT' in userchk but then there is this statement >> IFADJ = .FALSE. Can you please clarify what's the difference here? > > It was a typo, the relevant variable is IFADJ and nothing else. This > one had to be true or false depending on the equation to be solved. > >> I'm really grateful that you take the time to answer these questions. >> Please let me know if we can take this conversation off the mailing >> list. >> >> Brief summary of the problem we have in mind: we'd like to optimize the >> flow with respect to some parameters (either BCs or ICs) in terms of a >> defined cost function. So this can be regarded as either design or >> optimal control problem. Calculus of variations and Lagrange multipliers >> lead to adjoint equations which are very similar to NS equations. We >> already solved the problem using NEK and we'd like to use this again for >> adjoint equations. All the results will be in an outer loop for >> optimization (based on, say, steepest gradient). In general, the problem >> is unsteady, but we can start off with steady and non-evolving cases at >> early stages. > > We have calculated optimal initial conditions etc. using Nek, see e.g. > the results of the jet in crossflow, > http://www.sciencedirect.com/science/article/pii/S0997754614001009. Is > your intended optimisation similar to that? > > Best regards, > Philipp > > > >> >> Best, >> Saleh >> >> On 2/8/2016 4:26 AM, nek5000-users at lists.mcs.anl.gov wrote: >>> Hi Saleh >>> >>> If I understand you correctly, you ask about a base flow, i.e. the >>> velocity field you do your linearisation about (the one you use to >>> advect prturbation). If this is the case, than it is possible to do >>> with nek. Otherwise you have to wait for the code Oana mentioned. What >>> we do in our stability calculations, we take the solution of >>> non-linear NS (steady state reached with e.g. SFD) and next we load it >>> into nek as non-linear filed. One can do it by setting restart options >>> in ###.rea file. It should look like: >>> 3.33333 3.33333 -0.833333 -1.16667 >>> XFAC,YFAC,XZERO,YZERO >>> **MESH DATA** 1st line is X of corner 1,2,3,4. 2nd line is Y. >>> -1472 2 1472 NELT,NDIM,NELV >>> 1 PRESOLVE/RESTART OPTIONS ***** >>> bfext_cyl?.f00000 >>> 7 INITIAL CONDITIONS ***** >>> C Default >>> C Default >>> >>> >>> Where bfext_cyl0.f00000 is our restart file (we usually save this >>> velocity field in double precision). Next you have to set linear >>> solver with non-evolving base flow (nek is able to evolve both so be >>> careful). You freeze baseflow by setting negative value of parameter >>> 31 in ###.rea: >>> 0.00000 p030 > 0 ==> properties set in uservp() >>> -1.00000 p031 NPERT: #perturbation modes >>> 0.00000 p032 #BCs in re2 file, if > 0 >>> >>> Nek is able of evolving number of perturbation fields set by >>> abs(param(31)) (in ###.rea) and lpert in SIZE. I have to mention that >>> nek by itself outputs the non-linear fields only, so to see >>> perturbation you should add to your userchk something like this >>> >>> >>> subroutine userchk >>> >>> include 'SIZE' ! NX1, NY1, NZ1, NELV, NIO >>> include 'TSTEP' ! ISTEP, IOSTEP >>> include 'SOLN' ! V[XYZ]P, TP >>> include 'INPUT' ! IF3D, PARAM >>> include 'ADJOINT' ! IFADJ >>> >>> ! direct or adjoint run >>> IFADJ = .FALSE. >>> >>> ! write perturbation field >>> if (mod(ISTEP,int(PARAM(15))).eq.0) then >>> call outpost2(VXP,VYP,VZP,PRP,TP,0,'prt') >>> endif >>> >>> I hope it is an answer to your question. >>> regards >>> adam >>> >>> On 2016-02-06 19:15, nek5000-users at lists.mcs.anl.gov wrote: >>>> Hi Saleh, >>>> >>>> There is no easily accessible implementation for this, one has to do >>>> the change of the velocities himself. In fact we have ongoing work on >>>> this and once we are done I guess we can share it. >>>> What kind of case are you looking at, long time horizons are quite >>>> tricky to handle. We can take this conversation off the mailing list >>>> if needed. >>>> >>>> Oana >>>> >>>> >>>> >>>> ________________________________________ >>>> From: nek5000-users-bounces at lists.mcs.anl.gov >>>> [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of >>>> nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >>>> Sent: Friday, February 05, 2016 2:15 PM >>>> To: nek5000-users at lists.mcs.anl.gov >>>> Subject: Re: [Nek5000-users] Clarification on Adjoint solver >>>> >>>> Adams, >>>> >>>> I really appreciate your detailed answer. I definitely clarified >>>> the Nek >>>> functionality for adjoint solvers. One question remains is how to >>>> include the velocity of the forward solver into adjoint equations. I >>>> assume this is also the case when you do linear stability analysis >>>> with >>>> Nek, since the velocity in convective terms is not completely part of >>>> the solution. Mathematically, my question is: >>>> >>>> in $V. \nabla u_a$ term, as the advection, how can we substitute >>>> $V$ (as >>>> the forward velocity which is already solved for in a regular NS >>>> solution) in adjoint equations. suppose that the velocity, along with >>>> other quantities are saved in afield out put such as fld.f0000X for X >>>> time-step. So we have access to that velocity but how can we >>>> include it >>>> in adjoint equations? >>>> >>>> Many thanks again, >>>> Saleh >>>> >>>> On 2/2/2016 3:09 AM, nek5000-users at lists.mcs.anl.gov wrote: >>>>> Hi >>>>> >>>>> The adjoint implementation in nek assumes change of the direction of >>>>> time evolution. The goal is to use the standard nek solver with >>>>> minimal modifications (only advection terms have to be changed). You >>>>> get these equations from you standard one by changing the time and >>>>> pressure sign (t -> -t; p-> -p). This of course changes your time >>>>> integration limits, but it shouldn't be a problem. So when you have >>>>> your linear direct simulation, to turn it into dual one it is enough >>>>> to include in your userchk include files ADJOINT and TSTEP, and for >>>>> ISTEP.eq.0 set IFADJOINT to true. Something like: >>>>> subroutine userchk >>>>> >>>>> include 'SIZE' >>>>> include 'TSTEP' ! ISTEP >>>>> include 'ADJOINT' ! IFADJ >>>>> >>>>> if(ISTEP.eq.0) ADJOINT=.TRUE. >>>>> >>>>> There is no need to modify ###.rea or ###.map files. The only problem >>>>> are boundary conditions for open flows, as they would require >>>>> non-homogeneous bc. In this case to avoid problems we usually use >>>>> sufficiently large domain together with zero Dirichlet bc. The last >>>>> issue is sufficient resolution for both direct and adjoin runs. I >>>>> hope >>>>> this gives you some idea about nek implementation. >>>>> Regards >>>>> Adam >>>>> >>>>> On 2016-02-01 22:17, nek5000-users at lists.mcs.anl.gov wrote: >>>>>> Hi Neks, >>>>>> >>>>>> To my knowledge, Nek5000 is able to solve adjoint equations of the >>>>>> form: >>>>>> $\partial u_a/\partial t + V.\nabla u_a - u_a.(\nabla V)^T + \nabla >>>>>> \p_a + 1/Re \nabla^2 u_a=0$ and $\nabla.u_a=0$ >>>>>> where u_a and p_a are adjoint velocity and pressure. (Slightly >>>>>> different formulation may be seen in literature). >>>>>> >>>>>> Let's assume the forward problem is already solved so that $V$, i.e. >>>>>> velocity, (and possibly $p$ as the pressure) is (are) already know. >>>>>> Thus, the initial and boundary conditions for adjoint NS are also >>>>>> known. How can we now solve the dual/adjoint NS problem with >>>>>> Nek5000? >>>>>> Is there any example on how to modify .rea files? Specailly, the >>>>>> "convection term of $V.\nabla u_a - u_a.(\nabla V)^T$ needs to be >>>>>> modified compared to forward NS but am not sure how. >>>>>> My search of mailing list entails some modifications in "perturb.f" >>>>>> subroutine, but a little bit more clarification on that would be >>>>>> really appreciated. >>>>>> >>>>>> Yours, >>>>>> Saleh >>>>>> _______________________________________________ >>>>>> Nek5000-users mailing list >>>>>> Nek5000-users at lists.mcs.anl.gov >>>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>>>> _______________________________________________ >>>>> Nek5000-users mailing list >>>>> Nek5000-users at lists.mcs.anl.gov >>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>>> _______________________________________________ >>>> Nek5000-users mailing list >>>> Nek5000-users at lists.mcs.anl.gov >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>>> _______________________________________________ >>>> Nek5000-users mailing list >>>> Nek5000-users at lists.mcs.anl.gov >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Wed Feb 17 14:52:24 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 18 Feb 2016 07:52:24 +1100 Subject: [Nek5000-users] Please unsubscribe me Message-ID: Please unsubscribe me from the list. The 10 or so emails received each day are of no value and have become a form of spam Thanks -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Feb 17 15:00:52 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 17 Feb 2016 22:00:52 +0100 Subject: [Nek5000-users] Please unsubscribe me In-Reply-To: References: Message-ID: Hi, you will need to do that yourself at the address given below. Best regards, Philipp On 2016-02-17 21:52, nek5000-users at lists.mcs.anl.gov wrote: > Please unsubscribe me from the list. > The 10 or so emails received each day are of no value and have become a > form of spam > Thanks > > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > From nek5000-users at lists.mcs.anl.gov Thu Feb 18 03:51:15 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 18 Feb 2016 10:51:15 +0100 Subject: [Nek5000-users] ERROR of compilating In-Reply-To: References: Message-ID: Hi Paul, Thanks a lot for replying so soon. It indeed solved my problems. Regards, Andrew 2016-02-17 16:26 GMT+01:00 : > > Hi Andrew, > > It looks like you are trying to compile for too large a case. > > Two things to look for --- > > 1) Use the -mcmodel=medium flag in makenek > > Go to G= in makenek, uncomment, and change it to > > > G="-mcmodel=medium" > > 2) What is your lelt value in SIZE ? If you have (say) 100 elements > and are running on 8 > processors, you would only need > > lelt=100/8 = 13 or 14 > > (Even numbers are preferred for byte-alignment reasons.) > > You would not need lelt=100. > > This change would reduce your memory footprint by 8x. > > Paul > > > ------------------------------ > *From:* nek5000-users-bounces at lists.mcs.anl.gov [ > nek5000-users-bounces at lists.mcs.anl.gov] on behalf of > nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > *Sent:* Wednesday, February 17, 2016 8:17 AM > *To:* nek5000-users at lists.mcs.anl.gov > *Subject:* [Nek5000-users] ERROR of compilating > > Hi all, > > I have a compilation issue with my example. > I created a mesh 3D using n2to3 from the file w2dcyl020a.rea. I don't > have any issue with compilation. Then I refined the mesh 2D using pretex > and rebuilt the mesh 3D with n2to3 like before this and this time an error > message occured: > > "obj/stenosis.o: In function `uservp_': > stenosis.f:(.text+0xd): relocation truncated to fit: R_X86_64_PC32 against > symbol `nekuse_' defined in COMMON section in obj/stenosis.o > stenosis.f:(.text+0x1b): relocation truncated to fit: R_X86_64_PC32 > against symbol `nekuse_' defined in COMMON section in obj/stenosis.o > stenosis.f:(.text+0x29): relocation truncated to fit: R_X86_64_PC32 > against symbol `nekuse_' defined in COMMON section in obj/stenosis.o > obj/stenosis.o: In function `userf_': > stenosis.f:(.text+0x35): relocation truncated to fit: R_X86_64_PC32 > against symbol `nekuse_' defined in COMMON section in obj/stenosis.o > stenosis.f:(.text+0x3c): relocation truncated to fit: R_X86_64_PC32 > against symbol `nekuse_' defined in COMMON section in obj/stenosis.o > stenosis.f:(.text+0x43): relocation truncated to fit: R_X86_64_PC32 > against symbol `nekuse_' defined in COMMON section in obj/stenosis.o > obj/stenosis.o: In function `userq_': > stenosis.f:(.text+0x53): relocation truncated to fit: R_X86_64_PC32 > against symbol `nekuse_' defined in COMMON section in obj/stenosis.o > obj/stenosis.o: In function `userchk_': > stenosis.f:(.text+0x73): relocation truncated to fit: R_X86_64_PC32 > against symbol `dimn_' defined in COMMON section in obj/stenosis.o > stenosis.f:(.text+0x189): relocation truncated to fit: R_X86_64_PC32 > against symbol `dimn_' defined in COMMON section in obj/stenosis.o > stenosis.f:(.text+0x1e7): relocation truncated to fit: R_X86_64_PC32 > against symbol `dimn_' defined in COMMON section in obj/stenosis.o > obj/stenosis.o: In function `userbc_': > stenosis.f:(.text+0x2b4): additional relocation overflows omitted from the > output > collect2: ld a retourn? 1 code d'?tat d'ex?cution > make: *** [nek5000] Erreur 1" > > The same issue occures if I change the parameters lx1 and lxd in SIZE-file. > > Does anybody have an idea on how to fix this? > Thank you for any help or suggestion. > > *Andrew* > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Feb 18 09:56:51 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 18 Feb 2016 16:56:51 +0100 Subject: [Nek5000-users] Please unsubscribe me In-Reply-To: References: Message-ID: For me it?s ok, when a user decides to unsubscribe from this list. What worries me more, is that the content is considered as spam. This implies that the questions and/or replies aren?t helpful. We (the community) invest all of work into this list. So let?s think about how we can increase the quality. I would appreciate your input to the following two questions: 1. What is good? 2. What needs to change? Cheers, Stefan Am 2/17/16, 10:00 PM schrieb "nek5000-users-bounces at lists.mcs.anl.gov on behalf of nek5000-users at lists.mcs.anl.gov" unter : >Hi, >you will need to do that yourself at the address given below. >Best regards, >Philipp > >On 2016-02-17 21:52, nek5000-users at lists.mcs.anl.gov wrote: >> Please unsubscribe me from the list. >> The 10 or so emails received each day are of no value and have become a >> form of spam >> Thanks >> >> >> >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >_______________________________________________ >Nek5000-users mailing list >Nek5000-users at lists.mcs.anl.gov >https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Thu Feb 18 10:23:44 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 18 Feb 2016 17:23:44 +0100 Subject: [Nek5000-users] Mass flow rate and averages In-Reply-To: References: Message-ID: Hi Pual, thanks for your answer. I did some tests, i added your code in my userchk subroutine. This is what i have wrote: c----------------------------------------------------------------------- subroutine userchk include 'SIZE' include 'TOTAL' include 'NEKUSE' c integer mynxo c----------------------------------------------------------------------- integer istatus call domain_size(xmin,xmax,ymin,ymax,zmin,zmax) !calcolo lunghezza dominio xlenght = xmax-xmin !mass flow rate n = nx1*ny1*nz1*nelv volf = glsc3(vx,bm1,n)/xlenght open (UNIT=15,FILE= 'mfr.txt', STATUS ='replace',ACTION= 'write' & ,iostat=istatus) write (15,'(A15)') '#time= ',time write (15,'(2ES15.7)') ,volf return end c------------------------------------------------------------------------- now, compilation is ok, but when i run the sim, it exit with the following errors: call userchk Program received signal SIGSEGV: Segmentation fault - invalid memory reference. Program received signal SIGSEGV: Segmentation fault - invalid memory reference. Backtrace for this error: Program received signal SIGSEGV: Segmentation fault - invalid memory reference. Backtrace for this error: Program received signal SIGSEGV: Segmentation fault - invalid memory reference. Backtrace for this error: Backtrace for this error: #0 0x7F45C00F7777 #1 0x7F45C00F7D7E #2 0x7F45BF833D3F #0 0x7F1BDED46777 #1 0x7F1BDED46D7E #2 0x7F1BDE482D3F #0 0x7FCA82F99777 #1 0x7FCA82F99D7E #2 0x7FCA826D5D3F #0 0x7FE467B7B777 #1 0x7FE467B7BD7E #2 0x7FE4672B7D3F #3 0x480C3F in glsc3_ at math.f:1079 #3 0x480C3F in glsc3_ at math.f:1079 #3 0x480C3F in glsc3_ at math.f:1079 #4 0x402A72 in userchk_ at curve.f:53 #4 0x402A72 in userchk_ at curve.f:53 #4 0x402A72 in userchk_ at curve.f:53 #5 0x4035BF in nek_init_ at drive1.f:136 #6 0x402893 in nekton at drive.f:26 #7 0x7F45BF81EEC4 #5 0x4035BF in nek_init_ at drive1.f:136 #6 0x402893 in nekton at drive.f:26 #7 0x7F1BDE46DEC4 #5 0x4035BF in nek_init_ at drive1.f:136 #6 0x402893 in nekton at drive.f:26 #7 0x7FCA826C0EC4 #3 0x480C3F in glsc3_ at math.f:1079 #4 0x402A72 in userchk_ at curve.f:53 #5 0x4035BF in nek_init_ at drive1.f:136 #6 0x402893 in nekton at drive.f:26 #7 0x7FE4672A2EC4 =================================================================================== = BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES = EXIT CODE: 139 = CLEANING UP REMAINING PROCESSES = YOU CAN IGNORE THE BELOW CLEANUP MESSAGES =================================================================================== YOUR APPLICATION TERMINATED WITH THE EXIT STRING: Segmentation fault (signal 11) This typically refers to a problem with your application. Please see the FAQ page for debugging suggestions Obviously i tried to comment the code suggested and it run normally. Thanks. Il 17/02/2016 16:22, nek5000-users at lists.mcs.anl.gov ha scritto: > If you set the mass flow (p54-55) in your periodic channel, the forcing as a function of time > can be found via: > > grep vol logfile > > will show lines like: > > > 1989 0.1989000E+01 3.72046E-02 3.99058E-05 1.99996E+00 2.00000E+00 volflow X > > Column 1 is step number, > Column 2 is time. > > Column 3 is the mean (in space) pressure gradient, as function of time. > > This is the pressure gradient required to sustain a fix mass flow rate. > > If you wish to instead prescribe mean pressure gradient, set p54 and 55 = 0 in the .rea > file and set ffx = 1 (say) in the .usr file. > > To get the volumetric flow rate, you can do the following in userchk: > > n = nx1*ny1*nz1*nelv > volf = glsc3(vx,bm1,n)/xlength > > where xlength is the domain length. (This example assumes your > flow is in the x direction. > > You can get xlength via: > > call domain_size(xmin,xmax,ymin,ymax,zmin,zmax) > > xlength = xmax-xmin > > Paul > > ________________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > Sent: Wednesday, February 17, 2016 8:14 AM > To: nek5000-users at lists.mcs.anl.gov > Subject: [Nek5000-users] Mass flow rate and averages > > Hi neks, > maybe my, will be noob questions, but i am a nek noob. > I am running a DNS of channel with riblets, now i need to know the > averages fields and the mass flow rate. > For the averages fields i have seen in turbChannel example folder, and i > have seen in turbchannel.usr that there is a subroutine that do it. > But i didn't understand, if i set fintime(P10) and iotime(P14) and P68 > in rea file, have i always the output?Because in turbchannel example, if > i set how i already told, i have errors in compilation. > And the second question is related to the mass flow rate. > If i set forcing, can i know the relative mass flow rate? Or viceversa, > if i set the mass flow rate in .rea file, can i know the forcing? Where > can i read or compute it? > > Best regards > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Thu Feb 18 11:11:25 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 18 Feb 2016 12:11:25 -0500 Subject: [Nek5000-users] Please unsubscribe me In-Reply-To: References: Message-ID: Hi Stefan, I have two comments in this regard: 1- I noticed many of the question asked by members have been previously discussed and if they search the list archive, they can find the answers. So the questions and/or replies are in many cases redundant. 2- There are email clients (such as Thunderbird) in which you can ignore a thread on a topic in a forum without going through or deleting each email of the thread. Cheers, Mohsen On 18-Feb-16, at 10:56 AM, nek5000-users at lists.mcs.anl.gov wrote: > For me it?s ok, when a user decides to unsubscribe from this list. > What worries me more, is that the content is considered as spam. This > implies that the questions and/or replies aren?t helpful. We (the > community) invest all of work into this list. So let?s think about > how we > can increase the quality. I would appreciate your input to the > following > two questions: > > 1. What is good? > 2. What needs to change? > > Cheers, > Stefan > > Am 2/17/16, 10:00 PM schrieb "nek5000-users- > bounces at lists.mcs.anl.gov on > behalf of nek5000-users at lists.mcs.anl.gov" unter > nek5000-users at lists.mcs.anl.gov>: > >> Hi, >> you will need to do that yourself at the address given below. >> Best regards, >> Philipp >> >> On 2016-02-17 21:52, nek5000-users at lists.mcs.anl.gov wrote: >>> Please unsubscribe me from the list. >>> The 10 or so emails received each day are of no value and have >>> become a >>> form of spam >>> Thanks >>> >>> >>> >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Thu Feb 18 13:02:37 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 18 Feb 2016 19:02:37 +0000 Subject: [Nek5000-users] Please unsubscribe me In-Reply-To: References: Message-ID: Hi Mohsen and Stefan, To help deal with redundancy, I know other projects that strive to respond to user queries with links to the documentation. If the answer doesn't exist in the documentation, it is added. In that way, the docs grow naturally in a user-driven way. Obviously, this isn't always appropriate for edge cases or overly specific questions, but the added cost to add to the docs is generally amortized over a large part of the documentation readers. Max On Thu, Feb 18, 2016 at 11:11 AM wrote: > Hi Stefan, > > I have two comments in this regard: > > 1- I noticed many of the question asked by members have been > previously discussed and if they search the list archive, they can > find the answers. So the questions and/or replies are in many cases > redundant. > > 2- There are email clients (such as Thunderbird) in which you can > ignore a thread on a topic in a forum without going through or > deleting each email of the thread. > > Cheers, > Mohsen > > On 18-Feb-16, at 10:56 AM, nek5000-users at lists.mcs.anl.gov wrote: > > > For me it?s ok, when a user decides to unsubscribe from this list. > > What worries me more, is that the content is considered as spam. This > > implies that the questions and/or replies aren?t helpful. We (the > > community) invest all of work into this list. So let?s think about > > how we > > can increase the quality. I would appreciate your input to the > > following > > two questions: > > > > 1. What is good? > > 2. What needs to change? > > > > Cheers, > > Stefan > > > > Am 2/17/16, 10:00 PM schrieb "nek5000-users- > > bounces at lists.mcs.anl.gov on > > behalf of nek5000-users at lists.mcs.anl.gov" unter > > > nek5000-users at lists.mcs.anl.gov>: > > > >> Hi, > >> you will need to do that yourself at the address given below. > >> Best regards, > >> Philipp > >> > >> On 2016-02-17 21:52, nek5000-users at lists.mcs.anl.gov wrote: > >>> Please unsubscribe me from the list. > >>> The 10 or so emails received each day are of no value and have > >>> become a > >>> form of spam > >>> Thanks > >>> > >>> > >>> > >>> _______________________________________________ > >>> Nek5000-users mailing list > >>> Nek5000-users at lists.mcs.anl.gov > >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > >>> > >> _______________________________________________ > >> Nek5000-users mailing list > >> Nek5000-users at lists.mcs.anl.gov > >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > > > > _______________________________________________ > > Nek5000-users mailing list > > Nek5000-users at lists.mcs.anl.gov > > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Fri Feb 19 04:57:29 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 19 Feb 2016 18:57:29 +0800 (GMT+08:00) Subject: [Nek5000-users] temperature boundary Message-ID: Hi Nek, I want to run a cavity Rayleigh-Benard case and I find in the user guide that we can set constant temperature Dirichlet BC directly in .rea file. I need hot wall BC on the bottom of the square cavity and cold one on the top of it, so how can I set them in .rea file? Thank you for your help. Kind regards Xu B.L. -- Xu Bolun, University of Science and Techonology of China, Hefei,Anhui,China -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Fri Feb 19 08:08:27 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 19 Feb 2016 14:08:27 +0000 Subject: [Nek5000-users] temperature boundary In-Reply-To: References: Message-ID: Hi Xu, Take a look at, e.g., the examples/benard/ case and in particular, in ray_dd.rea & ray_cr.usr files Aleks ________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] Sent: Friday, February 19, 2016 4:57 AM To: nek5000-users at lists.mcs.anl.gov Subject: [Nek5000-users] temperature boundary Hi Nek, I want to run a cavity Rayleigh-Benard case and I find in the user guide that we can set constant temperature Dirichlet BC directly in .rea file. I need hot wall BC on the bottom of the square cavity and cold one on the top of it, so how can I set them in .rea file? Thank you for your help. Kind regards Xu B.L. -- Xu Bolun, University of Science and Techonology of China, Hefei,Anhui,China -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Feb 17 08:21:48 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 17 Feb 2016 15:21:48 +0100 Subject: [Nek5000-users] Mass flow rate and averages Message-ID: Hi neks, maybe mine, will be noob questions, but i am a nek noob. I am running a DNS of channel with riblets, now i? need to know the averages fields and the mass flow rate. For the averages fields i have seen in turbChannel example folder, and i have seen in?turbchannel.usr?that there is a subroutine that do it. But i didn't understand, if i set fintime(P10) and iotime(P14) and P68 in rea file, have i always the output?Because in turbchannel example, if i set how i already told, i have errors in compilation. And the second question is related to the mass flow rate. If i set forcing, can i know the relative mass flow rate? Or viceversa, if i set the mass flow rate in .rea file, can i know the forcing? Where can i read or compute it? From nek5000-users at lists.mcs.anl.gov Mon Feb 22 09:30:48 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 22 Feb 2016 16:30:48 +0100 Subject: [Nek5000-users] User defined boundary conditions from internal plane Message-ID: Dear All, I would like to impose boundary conditions (in userbc) with velocity values coming from a 2D plane in the same 3D domain being simulated (a channel). For example, having two channels on top of each other, how can I extract planar/cross-section values from the top channel (set with periodic boundary conditions in the streamwise direction) and impose them as "v" bc conditions in the bottom channel (set with v as inlet and O (outflow))? What I cannot understand is how to extract the values from a certain y-z plane to impose them as bc in userbc (assuming the element distribution is the same in both cases). Regards, jp -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Feb 22 10:07:48 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 22 Feb 2016 11:07:48 -0500 Subject: [Nek5000-users] flow thru a pipe Message-ID: hello NEKS i am trying to simulate steady state turbulent flow through a finite pipe. is there a example that can help me with the geometry and initial steps? i am new to nek5000. thank you -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Feb 23 02:28:13 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 23 Feb 2016 16:28:13 +0800 Subject: [Nek5000-users] Can nek do this? Message-ID: Hi developers and users of nek5000, I have been trying for a long time now to rotate the whole computational domain to simulate a rotating wing. I have scoured the nek forums but was unable to get any information on this. Is this even possible? I initially wanted to add a Coriolis force but coriolis in the Navier stokes is a cross product between angular velocity vector and velocity vector and in the userf, when tried this I received a compilation error saying that the ranks are different. This implied that i must use ux, uy and uz instead of vx,vy and vz. But as i understand, ux uy and uz are given values and not the velocity field taken from the calculations. Also for a rotating wing, it don't think it is advisable to apply a coriolis force to simulate rotation as the geometry is fixed and the angle of attack is not changing. This is why I'm trying to rotate the whole computational domain. In that sense, i can avoid using the ALE formulation for deforming meshes and therefore avoid the vanishing Jacobian errors which I have encountered. Can anybody guide me in the right direction? Rif -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Feb 23 05:45:34 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 23 Feb 2016 11:45:34 +0000 Subject: [Nek5000-users] Can nek do this? In-Reply-To: References: Message-ID: Hi Rif, You want to use ux, uy and uz (if 3D). For example, if Omega = (0 , 0 , omega), you would set omega = whatever ffx = 2*uy*omega ffy = -2*ux*omega ffz = 0 in userf. ux,uy,uz are user interface values. In userbc and useric, they are set by the user. They may also be _read_ by the user, since they are assigned the vx(i,j,k,e), vy(i,j,k,e), and vz(i,j,k,e) values prior to the call, point by point. The same holds for userq and userf --- values are set prior and read after. If the routine/call under consideration is looking for values of vx (say, on an inflow boundary in the case of userbc), then the user sets ux and this value is loaded into vx by Nek. hth, Paul ________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] Sent: Tuesday, February 23, 2016 2:28 AM To: nek5000-users at lists.mcs.anl.gov Subject: [Nek5000-users] Can nek do this? Hi developers and users of nek5000, I have been trying for a long time now to rotate the whole computational domain to simulate a rotating wing. I have scoured the nek forums but was unable to get any information on this. Is this even possible? I initially wanted to add a Coriolis force but coriolis in the Navier stokes is a cross product between angular velocity vector and velocity vector and in the userf, when tried this I received a compilation error saying that the ranks are different. This implied that i must use ux, uy and uz instead of vx,vy and vz. But as i understand, ux uy and uz are given values and not the velocity field taken from the calculations. Also for a rotating wing, it don't think it is advisable to apply a coriolis force to simulate rotation as the geometry is fixed and the angle of attack is not changing. This is why I'm trying to rotate the whole computational domain. In that sense, i can avoid using the ALE formulation for deforming meshes and therefore avoid the vanishing Jacobian errors which I have encountered. Can anybody guide me in the right direction? Rif -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Feb 23 12:16:49 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 23 Feb 2016 19:16:49 +0100 Subject: [Nek5000-users] output filed Message-ID: Hi neks, i need to print the output field u,v,w for every x coordinate. Now in userchk i wrote: do i=1,ntot write(16,*) istep,time, xm1(i,1,1,1), & vx(i,1,1,1),vy(i,1,1,1),vz(i,1,1,1) enddo and i receive the following output: 1 7.1530856036040695E-003 0.0000000000000000 0.37293630591777449 -0.32084904392637759 0.33230910496406607 1 7.1530856036040695E-003 0.17357493101301319 0.21130775186288545 -0.27768404661864132 0.20739454454471290 1 7.1530856036040695E-003 0.45442506898698676 0.46084398518785818 -0.27156542355496571 0.18149139586311763 1 7.1530856036040695E-003 0.62800000000000000 0.46110890335130450 -3.9277174842836750E-002 8.5610114286700789E-002 1 7.1530856036040695E-003 0.0000000000000000 0.74352058860894332 -7.5686819189858645E-002 1.3289264769474618E-002 1 7.1530856036040695E-003 0.17357493101301319 0.40009961054319920 0.13480828683699964 0.44478023194716509 1 7.1530856036040695E-003 0.45442506898698676 0.81382696494932794 -0.11005654233044537 0.17652132504468790 1 7.1530856036040695E-003 0.62799999999999989 0.67943618184656285 -9.6081837856840727E-002 -0.14185885411564772 1 7.1530856036040695E-003 0.0000000000000000 0.32147262016787648 -0.24804263078583569 -0.29171129659235678 1 7.1530856036040695E-003 0.17357493101301319 0.37769256515663086 -0.15125228365962307 -3.2429247313227494E-002 1 7.1530856036040695E-003 0.45442506898698676 0.24361827129204860 -3.1253044882220637E-002 0.19706743377184774 1 7.1530856036040695E-003 0.62799999999999989 0.63567408933529579 0.14261690852412279 -0.10888850075602638 1 7.1530856036040695E-003 0.0000000000000000 0.97344927314055640 0.24670700538197837 0.11490941870543755 1 7.1530856036040695E-003 0.17357493101301319 0.70722038909668339 0.30861285610239220 0.13262927045873465 1 7.1530856036040695E-003 0.45442506898698676 -0.26663222342418402 8.8968243007683773E-002 -5.8482352542457344E-002 1 7.1530856036040695E-003 0.62800000000000000 0.68703349900478228 5.2212760358220950E-002 -0.20495777585812078 I see that in x i have, every 4 rows, the numbers are repeated. Can someone tell me how should i read this numbers? I think they are vertices of every cell... it's all right? How can i have only x vx vy vz ordered by x from low to high? From nek5000-users at lists.mcs.anl.gov Wed Feb 24 12:28:11 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 24 Feb 2016 13:28:11 -0500 Subject: [Nek5000-users] nek capablities Message-ID: hey NEKS i have what may be a silly question. can NEK5000 solve for only the momentum equation and not the energy equation. Basically, i dont want the temperature term coming in the picture, just the velocity. thank you AVR -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Feb 25 09:55:41 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 25 Feb 2016 16:55:41 +0100 Subject: [Nek5000-users] Pressure calculation issue Message-ID: Hi all, I'm trying to realize the calculating of flow in a pipe. It works good In dimensionless 3D case but in dimensional case the solution is divergent. I have found that there is a problem with calculation of pressure. In dimensionless case there is a pressure difference between inlet and outlet while in dimensional case the pressure values at inlet and outlet are the same. Is there any explication on my issue? What is th difference between pressure calculation in both cases? Is there any tricky thing for dimensional case? Thanks in advance for any suggestion, Andrew -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Feb 25 16:09:36 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 25 Feb 2016 22:09:36 +0000 Subject: [Nek5000-users] help error in compiling nek 5000 Message-ID: Hello, I am trying to compile nek 5000 using gfortran and gcc compilers this Is the error that I am getting as below: Warning: Named COMMON block 'scruz' at (1) shall be of the same size gfortran -c -O2 -fdefault-real-8 -fdefault-double-8 -x f77-cpp-input -DPTRSIZE8 -DLONGINT8 -DUNDERSCORE -DGLOBAL_LONG_LONG -I/home/users/mansi/eddy -I/home/users/mansi/nek5_svn/trunk/nek -I/home/users/mansi/nek5_svn/trunk/nek/cmt -I./ /home/users/mansi/nek5_svn/trunk/nek/cmt/intpdiff.f -o obj/intpdiff.o /home/users/mansi/nek5_svn/trunk/nek/navier8.f: In function 'setvert3d': /home/users/mansi/nek5_svn/trunk/nek/navier8.f:2354: internal compiler error: in expand_expr_real_1, at expr.c:6846 Please submit a full bug report, with preprocessed source if appropriate. See for instructions. make: *** [obj/navier8.o] Error 1 make: *** Waiting for unfinished jobs.... In file /home/users/mansi/nek5_svn/trunk/nek/cmt/intpdiff.f:108 Thanks, mansi Patel Ph: 3527459651 -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Fri Feb 26 01:49:53 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 26 Feb 2016 08:49:53 +0100 Subject: [Nek5000-users] User defined boundary conditions from internal plane Message-ID: Any help/ideas regarding the issue please? jp -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Sat Feb 27 00:12:52 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Sat, 27 Feb 2016 11:42:52 +0530 Subject: [Nek5000-users] termination of simulation at high reynolds number Message-ID: Dear Sir I am currently running a simulation related to oscillating airfoil. Whenever I increase the reynolds number (>10000), the simulation stops running after few time steps showing the statement as shown below: Step 1652, t= 4.4604000E-02, DT= 2.7000000E-05, C= 2.938 4.2779E+03 3.4404E+00 Solving for fluid F T T 1652 Hmholtz VELX: 53 6.3582E-10 1.2153E+03 1.0000E-09 1652 Hmholtz VELY: 53 6.4018E-10 9.5397E+00 1.0000E-09 1652 U-PRES gmres: 64 9.9886E-08 1.0000E-07 4.8269E-01 8.7161E-01 2.3652E+00 1652 4.4604E-02 3.4916E+00 Fluid done CFL, Ctarg! 5.4135660200985729 0.50000000000000000 call outfld: ifpsco: F 1653 4.4604E-02 Write checkpoint: 1653 4.4604E-02 OPEN: naca0012.fld05 Emergency exit: 1653 time = 4.46039999999989986E-002 Latest solution and data are dumped for post-processing. *** STOP *** Kindly guide me on how to resolve this issue. T/R Prashanth Mohanasundaram -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Feb 29 02:58:35 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 29 Feb 2016 16:58:35 +0800 (GMT+08:00) Subject: [Nek5000-users] How to calculate Nusselt number in RB convection case Message-ID: Hi Nek, How should I calculate the Nusselt number(the average heat flux over the boundary) on the hot or cold wall of a 2D RB convection case? Nu=-Integral(tempy),that tempy means the partial difference of temprature in y direction, and Integral means the integral of tempy from x=0 to x=1. Is there any functions can realize that calculation? Thanks. Kind regards B.L.Xu -- Xu Bolun, University of Science and Techonology of China, Hefei,Anhui,China -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Feb 29 20:49:18 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 29 Feb 2016 19:49:18 -0700 Subject: [Nek5000-users] termination of simulation at high reynolds number In-Reply-To: References: Message-ID: Try decreasing DT (timestep), as it seems like a CFL issue. Best Regards, Tanmoy On Fri, Feb 26, 2016 at 11:12 PM, wrote: > Dear Sir > > I am currently running a simulation related to oscillating airfoil. > Whenever I increase the reynolds number (>10000), the simulation stops > running after few time steps showing the statement as shown below: > > Step 1652, t= 4.4604000E-02, DT= 2.7000000E-05, C= 2.938 4.2779E+03 > 3.4404E+00 > Solving for fluid F T T > 1652 Hmholtz VELX: 53 6.3582E-10 1.2153E+03 1.0000E-09 > 1652 Hmholtz VELY: 53 6.4018E-10 9.5397E+00 1.0000E-09 > 1652 U-PRES gmres: 64 9.9886E-08 1.0000E-07 4.8269E-01 > 8.7161E-01 2.3652E+00 > 1652 4.4604E-02 3.4916E+00 Fluid done > CFL, Ctarg! 5.4135660200985729 0.50000000000000000 > call outfld: ifpsco: F > > 1653 4.4604E-02 Write checkpoint: > > 1653 4.4604E-02 OPEN: > naca0012.fld05 > > > Emergency exit: 1653 time = 4.46039999999989986E-002 > Latest solution and data are dumped for post-processing. > *** STOP *** > > Kindly guide me on how to resolve this issue. > > T/R > Prashanth Mohanasundaram > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Feb 29 21:34:19 2016 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 01 Mar 2016 14:34:19 +1100 Subject: [Nek5000-users] Error when creating own mesh In-Reply-To: References: Message-ID: Hello, finally, did you solve this problem? I am having the same problem. First I generate a 2D mesh using gmsh, and after that I am using gmsh2nek to create the .rea file. This file seems to be ok. However, when I run genmap, I find the following error: genmap Input (.rea) file name: Cylinder Input mesh tolerance (default 0.2): NOTE: smaller is better, but generous is more forgiving for bad meshes. 0.2 reading .rea file data ... ERROR: error reading 1 1 782 aborting 510 in routine rdbdry. 1 quit What is the problem? Could you help me, please? Thanks SL El 09-07-2014 23:29, nek5000-users at lists.mcs.anl.gov escribi?: > Dear Friedrich, > > I recommend running "diff" on the example file and your file. > Assuming that the example file works you should then be close > to understanding the source of difficulty. > > Paul > > ------------------------- > > FROM: nek5000-users-bounces at lists.mcs.anl.gov > [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of > nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > SENT: Wednesday, July 09, 2014 5:57 AM > TO: nek5000-users at lists.mcs.anl.gov > SUBJECT: [Nek5000-users] Error when creating own mesh > > Dear Neks, > > I am have attempted to use nek5000, firstly by following some of the > examples etc. > > Now I am trying to create my own simulation, which is based upon the > ocyl examples. > > My efforts simply include creating my own grid points in the .rea file > to create a slightly different dimension... I believe I have > understood how to define the problem however I continually receive > errors when creating the genmap it reads as follows.... > > Input (.rea) file name: > ocyl > Input mesh tolerance (default 0.2): > NOTE: smaller is better, but generous is more forgiving for bad > meshes. > 0.2 > reading .rea file data ... > ERROR: error reading 1 11 24 > aborting 510 in routine rdbdry. > > 1 quit > > All I have done again is change the dimensions of the elements in the > .rea file so I do not see why it would now not work? > > Thank you for any help in advance, > > Friedrich. > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users