From owner-chemistry#* at *#ccl.net Wed Apr 15 14:38:01 2009 From: "Tim Dransfield timothy.dransfield!=!umb.edu" To: CCL Subject: CCL:G: Trouble with G98 PCM caculations, "Increase the MxTs"? Message-Id: <-39084-090415143607-25007-t9c+wMkFtzPS/dQisBe5vw]^[server.ccl.net> X-Original-From: "Tim Dransfield" Date: Wed, 15 Apr 2009 14:36:02 -0400 Sent to CCL by: "Tim Dransfield" [timothy.dransfield^-^umb.edu] I received no responses from the list, but I DID get a response from Gaussian's help desk. I'm posting it here so that other people searching for a solution to their trouble will at least see that the question has been answered, albeit not in a particularly useful way: Dr. Dransfield, Thank you for giving us a chance to comment. The short answer here is that you are out of luck. G98W had fixed arrays for holding the tiles for the solvent cavities. These could not be changed at run time and you comment that you are trying to increase TSNum to get good results. This is a collision course. We completely rewrote this code for Gaussian03 and have improved it further for Gaussian 09. There is no option for changing G98W. You can try decreasing the number of tiles but that probably has other drawbacks. Do you have access to Gaussian03? Possibly with a colleague at another university who could run a calculation for you? --------------------------- Sent to CCL by: "Tim Dransfield" [timothy.dransfield#,#umb.edu] Hello, everyone I've got an undergraduate struggling with a fairly ambitious computational project for his senior thesis, trying to model some transformations of nicotine in water solutions. He's running with the PCM model in B3LYP/6-31G**, and has a recurring problem that I can't find anywhere in the Gaussian manual. Unfortunately, all of my expertise is in gas phase calculations, and I'm at a loss. It's also frustrating, because a quick search of the CCL archives shows that this problem shows up annually... but there are no posted solutions! The problem begins when the PCM calculation fails due to insufficient tesserae. We use TSNUM to increase that value, and the job will often converge. However, in some recent cases with the larger basis set, we run into an error that we can't explain. A sample: Solvent: WATER Model : PCM/UAHF, Icomp = 2 Version: MATRIX INVERSION Cavity : ICOSAHEDRA with 320 initial tesserae ------------------------------------------------------------------------------ Nord Group Hybr Charge Alpha Radius Bonded to 1 CH sp2 0.08 1.20 1.815 C2 [s] N6 [d] 2 C sp3 0.08 1.20 1.500 C1 [s] C3 [s] C11 [s] 3 CH sp2 0.08 1.20 1.905 C2 [s] C4 [d] 4 CH sp2 0.08 1.20 1.815 C3 [d] C5 [s] 5 CH sp3 0.08 1.20 1.635 C4 [s] N6 [s] 6 N sp2 0.08 1.20 1.590 C1 [d] C5 [s] 11 CH sp3 0.08 1.20 1.680 C2 [s] C13 [s] 13 CH2 sp3 0.08 1.20 1.860 C11 [s] C16 [s] 16 CH2 sp3 0.08 1.20 1.680 C13 [s] C19 [s] 19 CH2 sp3 0.08 1.20 1.770 C16 [s] N22 [s] 22 NH sp3 0.08 1.20 1.590 C19 [s] C24 [s] 24 CH3 sp3 0.08 1.20 1.860 N22 [s] ------------------------------------------------------------------------------ Too many tesserae. Increase the MxTs. Error termination via Lnk1e in c:\program files\g98w\l502.exe. I do not know HOW to increase the MxTs, nor what a reasonable number might be. (This job was run with TSNUM=300) We would both very much appreciate any input from the list! Thanks in advance. Tim Timothy Dransfield Department of Chemistry UMass Boston timothy.dransfield-$-umb.edu