<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <div class="moz-cite-prefix">Hello,<br>
      <br>
      thanks as well from me for your contribution!<br>
      <br>
      Concerning what Erik said, I would suggest that you open an issue
      on our GitLab (gitlab.com/gromacs/gromacs) so we can have the
      proper discussion there.<br>
      <br>
      We are also having major ongoing effort for improving the free
      energy code.</div>
    <div class="moz-cite-prefix">So it would be great if you could join
      the next developer call on the 24th so that we can discuss things
      there with the rest of the developer community.<br>
      <br>
      Please don't hesitate to contact me for further discussions in any
      case.<br>
      <br>
      Cheers<br>
      <br>
      Paul<br>
    </div>
    <div class="moz-cite-prefix"><br>
    </div>
    <div class="moz-cite-prefix">On 15/03/2021 12:34, Erik Lindahl
      wrote:<br>
    </div>
    <blockquote type="cite"
      cite="mid:1E10A755-006A-41CB-B8BB-C497A6F6EFAC@gmail.com">
      <meta http-equiv="content-type" content="text/html; charset=UTF-8">
      Hi Yuzhi (&amp;Junhan)!
      <div><br>
      </div>
      <div>First, thanks for getting involved! We love to have more
        people involved, and will do our best to welcome you to the
        extended team. free energy is definitely an area we want to
        improve further, and contributions like the ones below sound
        great.</div>
      <div><br>
      </div>
      <div>Why don't you joint our next open developer call every second
        Wednesday? Or, if the timezone makes things inconvenient we
        could schedule a separate meeting.</div>
      <div><br>
      </div>
      <div>Some of the things we will like discuss would be:</div>
      <div><br>
      </div>
      <div>1. We are gradually moving to support both CUDA, SYCL and Hip
        so we run well on all accelerators - but this also means we need
        to avoid hacking things just for CUDA, and rather have a clean
        hierarchical implementation where different platforms might
        initially support different accelerated features. Much of this
        stuff has already started to happen in gmx 2021.</div>
      <div><br>
      </div>
      <div>2. For new features and functional forms we tend to be
        restrictive, partly because it means a collective undertaking
        for the entire team the next 5-10 years to ensure that feature
        works with all other features. We have not quite settled things
        on free energy yet, and we might consider modifying the
        interactions, but we don't want 4-5 different implementations
        from different teams - instead we should try to settle on which
        one will serve us all best.</div>
      <div><br>
      </div>
      <div>3. Documentation and Unit tests are critical to ensure new
        code keeps working ;-)</div>
      <div><br>
      </div>
      <div>Cheers,</div>
      <div><br>
      </div>
      <div>Erik</div>
      <div><br>
        <div dir="ltr">--
          <div>Erik Lindahl <a class="moz-txt-link-rfc2396E" href="mailto:erik.lindahl@scilifelab.se">&lt;erik.lindahl@scilifelab.se&gt;</a></div>
          <div>Professor of Biophysics</div>
          <div>Science for Life Laboratory</div>
          <div>Stockholm University &amp; KTH</div>
          <div>Office (SciLifeLab): +46 8 524 81567</div>
          <div>Cell (Sweden): +46 73 4618050 </div>
          <div>Cell (US): 1 267 307 8746</div>
          <div><br>
          </div>
        </div>
        <div dir="ltr"><br>
          <blockquote type="cite">On Mar 15, 2021, at 12:13, Yuzhi Zhang
            <a class="moz-txt-link-rfc2396E" href="mailto:352@pku.edu.cn">&lt;352@pku.edu.cn&gt;</a> wrote:<br>
            <br>
          </blockquote>
        </div>
        <blockquote type="cite">
          <div dir="ltr">
            <div class=" __aliyun_node_has_color"
style="margin:0px;padding:0px;border:0px;outline:0px;line-height:1.7;font-family:Tahoma,
              Arial, STHeiti, SimSun;font-size:14px;">
              <div
                style="margin:0px;padding:0px;border:0px;outline:0px;clear:both;">
                Dear Gromacs developers, 
                <div
                  style="margin:0px;padding:0px;border:0px;outline:0px;clear:both;">
    Based on Gromacs-2020.2, we've done a series of development related to FEP in following aspects . We would like to contribute our codes to Gromacs and we're looking forward to hearing
                  your opinions first. </div>
                <div
                  style="margin:0px;padding:0px;border:0px;outline:0px;clear:both;">
                      High-performance implementations:  </div>
                <div
                  style="margin:0px;padding:0px;border:0px;outline:0px;clear:both;">
                      1.  Offload FEP-PME to GPU.  Our status: finished.
We've also noticed that this function has been implemented in commit f7be07e3cc901eb03700d93248fc09b573370282 by M. Lundborg et. al. and has been merged in Gromacs-2021. So we won't commit this. 
                </div>
                <div
                  style="margin:0px;padding:0px;border:0px;outline:0px;clear:both;">
                      2.  Offload non-bonded free energy
                  kernel to GPU. Our status: finished.  </div>
                <div
                  style="margin:0px;padding:0px;border:0px;outline:0px;clear:both;">
                      3.  Offload bonded free energy
                  kernel to GPU. Our status: finished. </div>
                <div
                  style="margin:0px;padding:0px;border:0px;outline:0px;clear:both;">
                      4.  Support GPU updates for
                  FEP. Our status: finished md-integrator.
We've noticed that this has also been implemented in Gromacs-2021, but we'are still interested in supporting GPU updates of sd-integrator.  Similar discussion can be found in issue  #3258 on gitlab. 
                </div>
                <div
                  style="margin:0px;padding:0px;border:0px;outline:0px;clear:both;">
                      5.  Support GPU replica exchange. Our status:
                  finished. </div>
                <div
                  style="margin:0px;padding:0px;border:0px;outline:0px;clear:both;">
                  <br>
                </div>
                <div
                  style="margin:0px;padding:0px;border:0px;outline:0px;clear:both;">
                      New features:  </div>
                <div
                  style="margin:0px;padding:0px;border:0px;outline:0px;clear:both;">
    1.  Support "Soft Bond Potential" in FEP, which is proposed in  DOI: 10.1021/acs.jctc.6b00991 and allows a more smooth change of bond topology when there is bond formation or breaking in a FEP transformation, such as 5-member ring to 6-member ring.  Our status: finished. 
                </div>
                <div
                  style="margin:0px;padding:0px;border:0px;outline:0px;clear:both;">
    2.  Split softcore parameters for VDW and Coulomb interactions.  We noticed that currently Gromacs supports only one parameter "sc-alpha" to modify soft-core interactions.  After our tests, this will cause a phase-transition-like phenomenon at middle lambdas in some cases.  Our status: finished.  We also noticed recently there
have been some new forms of softcores  like gapsys softcore (DOI:  dx.doi.org/10.1021/ct300220p and discussed in <a
href="https://gitlab.com/gromacs/gromacs/-/merge_requests/882"
                    target="_blank" moz-do-not-send="true">https://gitlab.com/gromacs/gromacs/-/merge_requests/882</a> ) and Amber SSC2 softcore (DOI: 10.1021/acs.jctc.0c00237.)  And we would like to ask if you consider it's necessary to support such new softcore potentials. 
                </div>
                <div
                  style="margin:0px;padding:0px;border:0px;outline:0px;clear:both;">
    These terms are all that we want to contribute to Gromacs. But we have to admit that there is still a lot of work to test and standardize our codes to meet all requirements of Gromacs.  Is there any suggestion? 
                </div>
                <div
                  style="margin:0px;padding:0px;border:0px;outline:0px;clear:both;">
                  <br>
                </div>
                <div
                  style="margin:0px;padding:0px;border:0px;outline:0px;clear:both;">
                      Best wishes! </div>
                <div
                  style="margin:0px;padding:0px;border:0px;outline:0px;clear:both;">
                      Happy Simulating!<br>
                </div>
                <div
                  style="margin:0px;padding:0px;border:0px;outline:0px;clear:both;">
                      Yuzhi &amp; Junhan  </div>
              </div>
            </div>
            <div
style="margin:0px;padding:0px;border:0px;outline:0px;clear:both;font-size:0px;height:1px;overflow:hidden;"></div>
            <div
style="margin:0px;padding:0px;border:0px;outline:0px;line-height:20px;clear:both;">
              <br style="font-family:Helvetica, Tahoma,
                Arial;font-size:14px;white-space:normal;">
            </div>
            <span>-- </span><br>
            <span>Gromacs Developers mailing list</span><br>
            <span></span><br>
            <span>* Please search the archive at
              <a class="moz-txt-link-freetext" href="http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List">http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List</a>
              before posting!</span><br>
            <span></span><br>
            <span>* Can't post? Read
              <a class="moz-txt-link-freetext" href="http://www.gromacs.org/Support/Mailing_Lists">http://www.gromacs.org/Support/Mailing_Lists</a></span><br>
            <span></span><br>
            <span>* For (un)subscribe requests visit</span><br>
            <span><a class="moz-txt-link-freetext" href="https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers">https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers</a>
              or send a mail to <a class="moz-txt-link-abbreviated" href="mailto:gmx-developers-request@gromacs.org">gmx-developers-request@gromacs.org</a>.</span></div>
        </blockquote>
      </div>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
    </blockquote>
    <p><br>
    </p>
    <pre class="moz-signature" cols="72">-- 
Paul Bauer, PhD
GROMACS Development Manager
KTH Stockholm, SciLifeLab
0046737308594</pre>
  </body>
</html>