<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=us-ascii">
<meta name="Generator" content="Microsoft Exchange Server">
<!-- converted from rtf -->
<style><!-- .EmailQuote { margin-left: 1pt; padding-left: 4pt; border-left: #800000 2px solid; } --></style>
</head>
<body>
<font face="Tahoma, sans-serif" size="2">
<div>Dear Andreas (or anyone else),</div>
<div>Performing fully numerical derivatives with the dispersion corrected DFT is straightforward, however it seems like a huge waste; since there are analytical deratives for DFT, it would be nice to add the analytical DFT derivative to a numerical dispersion
correction. I tried input formats like the following:</div>
<div><font face="Consolas, monospace" size="2"> </font></div>
<div>(The system is just a dummy, so don’t worry that it isn’t one where you would normally want that correction)</div>
<div> </div>
<div>***,H2O</div>
<div>memory,400,M</div>
<div>basis=avdz</div>
<div>thrdscf=1e-12,thrdscf_max=1e-9</div>
<div>gthresh,throvl=1e-9,throrth=1e-7</div>
<div>gdirect</div>
<div>geomtyp=xyz</div>
<div>symmetry,nosym</div>
<div>orient,noorient</div>
<div>geometry={</div>
<div> O 0.0000000000 0.0417232258 0.0323055610</div>
<div> H 0.0000000000 0.0372295904 0.9974745462</div>
<div> H 0.0000000000 0.9750241937 -0.2137039261</div>
<div>}</div>
<div><font face="Consolas, monospace" size="2"> </font></div>
<div>label:</div>
<div><font face="Consolas, monospace" size="2"> </font></div>
<div>df-ks,b3lyp</div>
<div> </div>
<div>dispcorr</div>
<div>forces,numerical,variable=edisp,startcmd=dispcorr</div>
<div> </div>
<div>df-ks,b3lyp</div>
<div>eks=energy</div>
<div>forces;</div>
<div>add,1</div>
<div>energy=eks+edisp</div>
<div> </div>
<div>optg,startcmd=label:</div>
<div><font face="Consolas, monospace" size="2"> </font></div>
<div>A couple of notes:</div>
<div>(1) You need the initial DFT call to let dispcorr know what parameters to use</div>
<div>(2) You must calculate the numerical derivative first, because numerical derivatives don’t understand the “add” command, while analytical derivatives do</div>
<div>(3) This correctly calculates the force on the first step and optg takes what appears to be the correct first optimization step, however, after that it crashes. From what I can tell, it (for some reason) skips the numerical derivative of the dispersion
correction on the second iteration AND double calculates the ks derivative.</div>
<div> </div>
<div>Alternatively, one could imagine moving the first df-ks,b3lyp outside of the label: block, however then the code doesn’t seem to properly set the density fitting basis when it calls Alaska a second time (and it still skips the dispcorr gradient on the
second time through).</div>
<div> </div>
<div>Any ideas?</div>
<div>Thanks,</div>
<div>Andrew</div>
<div><font face="Consolas, monospace" size="2"> </font></div>
<div><font face="Consolas, monospace" size="2"> </font></div>
<div><font face="Consolas, monospace" size="2">Dear Andrew,</font></div>
<div><font face="Consolas, monospace" size="2"> </font></div>
<div><font face="Consolas, monospace" size="2">in the current user version of Molpro there are no gradients for the </font></div>
<div><font face="Consolas, monospace" size="2">dispersion correction implemented. But you can probably use the</font></div>
<div><font face="Consolas, monospace" size="2">scheme described in section 40.2 in the manual to calculate</font></div>
<div><font face="Consolas, monospace" size="2">numerical gradients (save the total sum of the DFT energy and the</font></div>
<div><font face="Consolas, monospace" size="2">dispersion correction in variable 'energy').</font></div>
<div><font face="Consolas, monospace" size="2"> </font></div>
<div><font face="Consolas, monospace" size="2">Wishes,</font></div>
<div><font face="Consolas, monospace" size="2">Andreas</font></div>
<div><font face="Consolas, monospace" size="2"> </font></div>
<div><font face="Consolas, monospace" size="2"> </font></div>
<div><font face="Consolas, monospace" size="2">On Monday 19 July 2010 21:36, Taube, Andrew wrote:</font></div>
<div><font face="Consolas, monospace" size="2">> Is there a way to use the empirical damped dispersion correction with DFT</font></div>
<div><font face="Consolas, monospace" size="2">> to do (analytical) geometry optimizations? Or does anyone have a slick way</font></div>
<div><font face="Consolas, monospace" size="2">> to calculate the numerical derivative of the dispersion correction and</font></div>
<div><font face="Consolas, monospace" size="2">> combine it with analytic DFT gradients? Thanks,</font></div>
<div><font face="Consolas, monospace" size="2">> Andrew</font></div>
<div><font face="Consolas, monospace" size="2">> ---</font></div>
<div><font face="Consolas, monospace" size="2">> Andrew G. Taube</font></div>
<div><font face="Consolas, monospace" size="2">> D. E. Shaw Research</font></div>
<div><font face="Consolas, monospace" size="2">> email: <a href="mailto:Andrew.Taube@DEShawResearch.com"><font color="#0000FF"><u>Andrew.Taube@DEShawResearch.com</u></font></a></font></div>
<div><font face="Consolas, monospace" size="2">> Tel: +1-212-478-0118</font></div>
<div><font face="Consolas, monospace" size="2">> Fax: +1-212-845-1118</font></div>
<div><font face="Calibri, sans-serif" size="2"> </font></div>
<div><font face="Calibri, sans-serif" size="2"> </font></div>
<div><font face="Calibri, sans-serif" size="2"> </font></div>
</font>
</body>
</html>