[petsc-users] TS initial guess

[Jed Brown]

Some methods have extrapolation options. For THETA (which includes backward Euler as theta=1), you can use -ts_theta_initial_guess_extrapolate.

This type of extrapolation is sometimes unstable or may produce an invalid state, such as negative density.

I'm assuming in your question that you're concerned about the number of Newton iterations rather than number of Krylov iterations.

Barry Smith <bsmith at petsc.dev> writes:

>   For TSBEULER (the theta method) see https://petsc.org/release/docs/manualpages/TS/TSTHETA.html <https://petsc.org/release/docs/manualpages/TS/TSTHETA.html> and look at the source code src/ts/impls/implicit/theta/teta.c for TSStep_Theta. You can use -snes_monitor_solution OPTIONS to see what the solutions are the nonlinear system look like as it solves the system.
>
>
>
>   Barry
>
>
>
>> On Oct 11, 2021, at 12:26 PM, Alfredo J Duarte Gomez <aduarteg at utexas.edu> wrote:
>> 
>> Good morning PETSC team,
>> 
>> I have  a working algorithm for my implicit TS integrator with a system of ODE/DAE's, but I am observing a rather high number of iterations
>> 
>> I am currently using the simplest settings of a TSBEULER and setting a constant time step.
>> 
>> My question  right now is whether the default settings use any sort of initial guess algorithm before every time step. 
>> 
>> Since I have seen that the time step adapter calculates the Local Truncation Error, it should be possible to use an extrapolation of arbitrary order of accuracy as an initial guess for every time step right? Can someone indicate how I would be able to use that?
>> 
>> Additionally, it would be very helpful to take a look at that initial guess, is it possible to use any existing function to calculate it either in the PreStep or PostStep function to visualize it?
>> 
>> Thank you,
>> 
>> -- 
>> Alfredo Duarte
>> Graduate Research Assistant
>> The University of Texas at Austin