<html>
  <head>
    <meta content="text/html; charset=utf-8" http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    <div class="moz-cite-prefix">    Hello,<br>
              PETSC_USE_COMPLEX isn't a symbol in the shared library
      when Petsc is built with complex scalars, so I don't see a way to
      access it at runtime. I'll have to write a simple C program that
      uses sizeof() and write the value to a file.<br>
      <br>
              As for the MPI communicator, the julia MPI package uses a
      C int to store it, so I will typealias to that to ensure
      consistency.  If an MPI implementation uses an 8 byte pointer,
      MPI.jl will have to change too.<br>
      <br>
          Jared Crean<br>
      <br>
      On 7/14/2015 1:04 PM, Matthew Knepley wrote:<br>
    </div>
    <blockquote
cite="mid:CAMYG4GnwN=jm=k3w6jXHf2O0-017Pwy3oimKvO0Lg-erJHDUJA@mail.gmail.com"
      type="cite">
      <div dir="ltr">
        <div class="gmail_extra">
          <div class="gmail_quote">On Tue, Jul 14, 2015 at 10:56 AM,
            Jared Crean <span dir="ltr"><<a moz-do-not-send="true"
                href="mailto:jcrean01@gmail.com" target="_blank">jcrean01@gmail.com</a>></span>
            wrote:<br>
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex">
              <div bgcolor="#FFFFFF" text="#000000">
                <div>    Hello everyone,<br>
                          I got the package in a reasonably working
                  state and Travis testing setup, so I am putting the
                  package up on Github.<br>
                  <br>
                          <a moz-do-not-send="true"
                    href="https://github.com/JaredCrean2/PETSc.jl"
                    target="_blank">https://github.com/JaredCrean2/PETSc.jl</a><br>
                  <br>
                          There is still a lot more work to do, but its
                  a start.<br>
                  <br>
                          A couple questions:<br>
                          When looking though the code, I noticed the
                  MPI communicator is being passed as a 64 bit integer. 
                  mpi.h typedefs it as an int, so shouldn't it be a 32
                  bit integer?<br>
                </div>
              </div>
            </blockquote>
            <div><br>
            </div>
            <div>Some MPI implementations store the communicator as a
              pointer, which may be 64 bits. I think the only thing the
              standard says is</div>
            <div>that MPI_Comm should be defined.</div>
            <div> </div>
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex">
              <div bgcolor="#FFFFFF" text="#000000">
                <div>         Also, is there a way to find out at
                  runtime what datatype a PetscScalar is?  It appears
                  PetscDataTypeGetSize does not accept PetscScalar as an
                  argument.</div>
              </div>
            </blockquote>
            <div><br>
            </div>
            <div>If PETSC_USE_COMPLEX is defined its PETSC_COMPLEX,
              otherwise its PETSC_REAL. You can also just use
              sizeof(PetscScalar). What do you</div>
            <div>want to do?</div>
            <div><br>
            </div>
            <div>  Thanks,</div>
            <div><br>
            </div>
            <div>     Matt</div>
            <div> </div>
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex">
              <div bgcolor="#FFFFFF" text="#000000">
                <div><span class="HOEnZb"><font color="#888888"><br>
                          Jared Crean</font></span>
                  <div>
                    <div class="h5"><br>
                      <br>
                      <br>
                      On 07/06/2015 09:02 AM, Matthew Knepley wrote:<br>
                    </div>
                  </div>
                </div>
                <div>
                  <div class="h5">
                    <blockquote type="cite">
                      <div dir="ltr">
                        <div class="gmail_extra">
                          <div class="gmail_quote">On Mon, Jul 6, 2015
                            at 4:59 AM, Patrick Sanan <span dir="ltr"><<a
                                moz-do-not-send="true"
                                href="mailto:patrick.sanan@gmail.com"
                                target="_blank">patrick.sanan@gmail.com</a>></span>
                            wrote:<br>
                            <blockquote class="gmail_quote"
                              style="margin:0 0 0 .8ex;border-left:1px
                              #ccc solid;padding-left:1ex">
                              <div dir="ltr">I had a couple of brief
                                discussions about this at Juliacon as
                                well. I think it would be useful, but
                                there are a couple of things to think
                                about from the start of any new attempt
                                to do this:
                                <div>1. As Jack pointed out, one issue
                                  is that the PETSc library must be
                                  compiled for a particular precision.
                                  This raises some questions - should
                                  several versions of the library be
                                  built to allow for flexibility?</div>
                                <div>2. An issue with wrapping PETSc is
                                  always that the flexibility of using
                                  the PETSc options paradigm is reduced
                                  - how can this be addressed?
                                  Could/should an expert user be able to
                                  access the options database directly,
                                  or would this be too much violence to
                                  the wrapper abstraction?</div>
                              </div>
                            </blockquote>
                            <div><br>
                            </div>
                            <div>I have never understood why this is an
                              issue. Can't you just wrap our interface
                              level, and use the options just as we do?
                              That</div>
                            <div>is essentially what petsc4py does. What
                              is limiting in this methodology? On the
                              other hand, requiring specific types, ala
                              FEniCS,</div>
                            <div>is very limiting.</div>
                            <div><br>
                            </div>
                            <div>   Matt</div>
                            <div> </div>
                            <blockquote class="gmail_quote"
                              style="margin:0 0 0 .8ex;border-left:1px
                              #ccc solid;padding-left:1ex">
                              <div class="gmail_extra">
                                <div class="gmail_quote">On Sat, Jul 4,
                                  2015 at 11:00 PM, Jared Crean <span
                                    dir="ltr"><<a
                                      moz-do-not-send="true"
                                      href="mailto:jcrean01@gmail.com"
                                      target="_blank">jcrean01@gmail.com</a>></span>
                                  wrote:<br>
                                  <blockquote class="gmail_quote"
                                    style="margin:0 0 0
                                    .8ex;border-left:1px #ccc
                                    solid;padding-left:1ex">
                                    <div bgcolor="#FFFFFF"
                                      text="#000000"> <big>Hello,<br>
                                             I am a graduate student
                                        working on a CFD code written in
                                        Julia, and I am interested in
                                        using Petsc as a linear solver
                                        (and possibly for the non-linear
                                        solves as well) for the code.  I
                                        discovered the Julia wrapper
                                        file Petsc.jl in Petsc and have
                                        updated it to work with the
                                        current version of Julia and the
                                        MPI.jl package, using only MPI
                                        for communication (I don't think
                                        Julia's internal parallelism
                                        will scale well enough, at least
                                        not in the near future).<br>
                                        <br>
                                             I read the discussion on
                                        Github [<a
                                          moz-do-not-send="true"
                                          href="https://github.com/JuliaLang/julia/issues/2645"
                                          target="_blank">https://github.com/JuliaLang/julia/issues/2645</a>],

                                        and it looks like <br>
                                        there currently is not a
                                        complete package to access Petsc
                                        from Julia.  With your
                                        permission, I would like to use
                                        the Petsc.jl file as the basis
                                        for developing a package.  My
                                        plan is create a lower level
                                        interface that exactly wraps
                                        Petsc functions, and then
                                        construct a higher level
                                        interface, probably an object
                                        that is a subtype of Julia's
                                        AbstractArray, that allows users
                                        to store values into Petsc
                                        vectors and matrices.  I am less
                                        interested in integrating
                                        tightly with Julia's existing
                                        linear algebra capabilities than
                                        ensuring good scalability.  The
                                        purpose of the high level
                                        interface it simple to populate
                                        the vector or matrix.<br>
                                        <br>
                                             What do you think, both
                                        about using the Petsc.jl file
                                        and the  overall approach?<span><font
                                            color="#888888"><br>
                                            <br>
                                                 Jared Crean</font></span></big><br>
                                    </div>
                                  </blockquote>
                                </div>
                                <br>
                              </div>
                            </blockquote>
                          </div>
                          <br>
                          <br clear="all">
                          <div><br>
                          </div>
                          -- <br>
                          <div>What most experimenters take for granted
                            before they begin their experiments is
                            infinitely more interesting than any results
                            to which their experiments lead.<br>
                            -- Norbert Wiener</div>
                        </div>
                      </div>
                    </blockquote>
                    <br>
                  </div>
                </div>
              </div>
            </blockquote>
          </div>
          <br>
          <br clear="all">
          <div><br>
          </div>
          -- <br>
          <div class="gmail_signature">What most experimenters take for
            granted before they begin their experiments is infinitely
            more interesting than any results to which their experiments
            lead.<br>
            -- Norbert Wiener</div>
        </div>
      </div>
    </blockquote>
    <br>
  </body>
</html>