<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <div class="moz-cite-prefix">Dear Matt,<br>
          Thanks for your reply on the 1st question and I'm sorry for
      the confusion on the 2nd one. <br>
          In order to create a non-uniform Cartesian grid, I expected
      that the way I was programming can let me freely control the grid
      size. In the code attached in the previous E-mail, the
      'DMDASetUniformCoordinate' generates a uniform coordinate with
      100*48*48 grid number and 0.1*0.1*0.1 grid size. I was trying to
      manually double the grid size in the y-direction for the whole
      mesh with operations of DMDA in the 'SetNonUniformGrid()', which
      follows the method that /src/ksp/ksp/examples/tutorials/ex49.c
      used. However, as you can see in the figure attached in the
      previous E-mail, only the mesh on z = 0 plane and x = 0 plane has
      been changed to dy = 0.2; while I expected all mesh should have dy
      = 0.2 after I call 'SetNonUniformGrid()'. <br>
          Any idea about this?<br>
      <br>
      thanks,<br>
      Alan<br>
          <br>
      On 2/28/2013 6:31 PM, Matthew Knepley wrote:<br>
    </div>
    <blockquote
cite="mid:CAMYG4GmruJOFOeU5znsBD_W_EXzmFXmQGmyJ9pR3vy3zsJBPug@mail.gmail.com"
      type="cite">
      <div dir="ltr">On Thu, Feb 28, 2013 at 7:19 PM, Zhenglun (Alan)
        Wei <span dir="ltr"><<a moz-do-not-send="true"
            href="mailto:zhenglun.wei@gmail.com" target="_blank">zhenglun.wei@gmail.com</a>></span>
        wrote:<br>
        <div class="gmail_extra">
          <div class="gmail_quote">
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex">Dear
              folks,<br>
                  Here are several questions following up my previous
              ones:<br>
              1, I modified my code based on ex49.c. However, I found
              that if I don't call 'DMDASetUniformCoordinates' before
              calling 'DMDAVecGetArray', then an error comes up:<br>
              [0]PETSC ERROR: Null argument, when expecting valid
              pointer!<br>
              [0]PETSC ERROR: Null Object: Parameter # 2!<br>
                  The code do works if I add a call for
              'DMDASetUniformCoordinates' in it. However, I feel it is a
              waste of calling 'DMDASetUniformCoordinates' since what
              follows this calling is to change the coordinate to a
              non-uniform gird. Does it have any way to bypass this call
              or you think it is fine to keep it here.<br>
            </blockquote>
            <div><br>
            </div>
            <div style="">If you do not call it, then you have to create
              the coordinate vector yourself. I would just call it.</div>
            <div> </div>
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex">
              2, Also, if I do by this way, the outputted mesh looks
              weird, as shown in the figure.<br>
                  I was trying to double the dy value manually in the
              SetNonUniformGrid(). However, the only mesh changed is
              just on the z=0 and x=0 plane. Here I attached my code. Do
              I need to do anything else except DMDAVecRestoreArray()
              after I modified the coordinate information?<br>
            </blockquote>
            <div><br>
            </div>
            <div style="">I do not understand what you want here.</div>
            <div style=""><br>
            </div>
            <div style="">  Matt</div>
            <div> </div>
            <blockquote class="gmail_quote" style="margin:0 0 0
              .8ex;border-left:1px #ccc solid;padding-left:1ex">
              thanks,<br>
              Alan<br>
              <br>
              <br>
              On 2/26/2013 8:46 PM, Barry Smith wrote:<br>
              <blockquote class="gmail_quote" style="margin:0 0 0
                .8ex;border-left:1px #ccc solid;padding-left:1ex">
                On Feb 26, 2013, at 4:49 PM, "Zhenglun (Alan) Wei"<<a
                  moz-do-not-send="true"
                  href="mailto:zhenglun.wei@gmail.com" target="_blank">zhenglun.wei@gmail.com</a>>
                 wrote:<br>
                <br>
                <blockquote class="gmail_quote" style="margin:0 0 0
                  .8ex;border-left:1px #ccc solid;padding-left:1ex">
                  Dear folks,<br>
                        I have one question for the ex49.<br>
                        It do show a way to 'get' the 'coord' by the
                  'DMDAVecGetArray'. Should I just modified the
                  coordinate on the 'coord' in order to obtain a
                  non-uniform grid? and then, use DMDAVecRestoreArray to
                  link the modified 'coord' back to 'cda' so that the
                  'coord' will be updated to the mesh I want?<br>
                </blockquote>
                    Yes, you can do this.<br>
                <br>
                <br>
                <blockquote class="gmail_quote" style="margin:0 0 0
                  .8ex;border-left:1px #ccc solid;padding-left:1ex">
                  thanks,<br>
                  Alan<br>
                  <br>
                  On 2/22/2013 3:20 PM, Matthew Knepley wrote:<br>
                  <blockquote class="gmail_quote" style="margin:0 0 0
                    .8ex;border-left:1px #ccc solid;padding-left:1ex">
                    On Fri, Feb 22, 2013 at 3:57 PM, Zhenglun (Alan)
                    Wei<<a moz-do-not-send="true"
                      href="mailto:zhenglun.wei@gmail.com"
                      target="_blank">zhenglun.wei@gmail.com</a>>
                     wrote:<br>
                    Dear all,<br>
                    I hope you're having a nice day.<br>
                    I wonder if anyone could point out a example for me
                    about the<br>
                    DMDASetCoordinates(). I found this<br>
                    (<a moz-do-not-send="true"
href="http://www.mcs.anl.gov/petsc/petsc-current/src/dm/impls/da/dacorn.c.html#DMDASetCoordinates"
                      target="_blank">http://www.mcs.anl.gov/petsc/petsc-current/src/dm/impls/da/dacorn.c.html#DMDASetCoordinates</a>)<br>
                    online; however it does not really demonstrate how
                    to create the<br>
                    'coordinate vector - c'.<br>
                    Also, I found that in the petsc sub-folder:<br>
                    /petsc-dev/src/dm/impls/patch has a patch.c, which I
                    suppose that it<br>
                    does a local refinement for mesh. Does this a mature
                    model that we can<br>
                    use or just a testing code?<br>
                    <br>
                    Look at KSP ex49. In the routine
                    DMDACoordViewGnuplot2d() it reads out the
                    coordinates.<br>
                    You could just as easily set them in this fashion.<br>
                    <br>
                        Matt<br>
                      thanks,<br>
                    Alan<br>
                    <br>
                    On 11/7/2012 6:55 PM, Barry Smith wrote:<br>
                    <blockquote class="gmail_quote" style="margin:0 0 0
                      .8ex;border-left:1px #ccc solid;padding-left:1ex">
                      On Nov 7, 2012, at 6:50 PM, Alan<<a
                        moz-do-not-send="true"
                        href="mailto:zhenglun.wei@gmail.com"
                        target="_blank">zhenglun.wei@gmail.com</a>>
                       wrote:<br>
                      <br>
                      <blockquote class="gmail_quote" style="margin:0 0
                        0 .8ex;border-left:1px #ccc
                        solid;padding-left:1ex">
                        Dear folks,<br>
                        I hope you're having a nice day.<br>
                        I'm testing the Poisson solver in
                        /src/ksp/ksp/examples/tutorials/ex45.<br>
                        It uses the 'DMDACreate3d' to generate a
                        Cartisian structured grid and<br>
                        solve it with uniform grid size. However, I
                        attempt to modify it to<br>
                        obtain solution with a Cartsian structured grid
                        and non-uniform grid (It<br>
                        is not unstructured local refined mesh). Should
                        I still use the<br>
                        'DMDACreate3d' and implement different grid size
                        in 'ComputeMatrix' and<br>
                        'ComputeRHS'? In other words, does
                        'DMDACreate3d' itself include any<br>
                        mesh size information?<br>
                      </blockquote>
                          You can use DMDASetCoordinates() and
                      DMDAGetCoordinates() (called DMSetCoordinates()
                      and DMGetCoordinates() in petsc-dev) to keep track
                      of the coordinates but you are responsible for
                      generating the correctly scaled entries in the
                      matrix etc.<br>
                      <br>
                          Barry<br>
                      <br>
                      <blockquote class="gmail_quote" style="margin:0 0
                        0 .8ex;border-left:1px #ccc
                        solid;padding-left:1ex">
                        thanks,<br>
                        Alan<br>
                      </blockquote>
                    </blockquote>
                    <br>
                    <span class="HOEnZb"><font color="#888888">
                        <br>
                        -- <br>
                        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<br>
                      </font></span></blockquote>
                </blockquote>
              </blockquote>
              <br>
            </blockquote>
          </div>
          <br>
          <br clear="all">
          <div><br>
          </div>
          -- <br>
          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>
    </blockquote>
    <br>
  </body>
</html>