From owner-chemistry@ccl.net Tue Jan 24 12:27:00 2023 From: "POONAM PARKAR poonamparkar72#%#gmail.com" To: CCL Subject: CCL: Query regarding ADMP-MD calculation Message-Id: <-54841-230124102135-17169-RSIy9BvHlgQkPJxjt+G73g!^!server.ccl.net> X-Original-From: "POONAM PARKAR" Date: Tue, 24 Jan 2023 10:21:33 -0500 Sent to CCL by: "POONAM PARKAR" [poonamparkar72{}gmail.com] Dear All I am Poonam Parkar, a Ph.D. student. I have been studying DFT and involved in hydrogen storage. I am testing ADMP-MD simulation for C,B,N,Ti (up to 27 atoms). But I am unable to run ADMP as the job get terminated running for only one scf cycle giving an error message "GEWDen is stupid about Cholesky+NBsUse To: CCL Subject: CCL: Query regarding ADMP-MD calculation Message-Id: <-54842-230124151752-24282-gKt+0WZk73VhCp15rPCSNw*o*server.ccl.net> X-Original-From: Frank Jensen Content-Language: en-US Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset="us-ascii" Date: Tue, 24 Jan 2023 20:17:29 +0000 MIME-Version: 1.0 Sent to CCL by: Frank Jensen [frj^_^chem.au.dk] The error message suggests that the algorithm cannot handle the situation where the basis set (NBasis) is reduced to NBsUse due to near-linear dependency. The default is 10^-6. If the smallest eigenvalue in the overlap matrix is only slightly below this, you can turn off the check for small eigenvalues with iop(3/32=1). In that case, you may also want to increase the integral accuracy, by int(acc2e=15). If the smallest eigenvalue is below, say, 10^-8, you are probably too close to a linear dependent basis set to have any chance of getting the SCF to converge. However, the underlying reason is likely that you are using diffuse functions on all atoms (++). Do you really need this? And the 6-311G is really only a marginal better basis than 6-31G, and there are more modern basis sets that will give smaller basis set errors for the same computational cost..... Frank -----Original Message----- > From: owner-chemistry+frj==chem.au.dk.:.ccl.net On Behalf Of POONAM PARKAR poonamparkar72#%#gmail.com Sent: 24. januar 2023 16:22 To: Frank Jensen Subject: CCL: Query regarding ADMP-MD calculation Sent to CCL by: "POONAM PARKAR" [poonamparkar72{}gmail.com] Dear All I am Poonam Parkar, a Ph.D. student. I have been studying DFT and involved in hydrogen storage. I am testing ADMP-MD simulation for C,B,N,Ti (up to 27 atoms). But I am unable to run ADMP as the job get terminated running for only one scf cycle giving an error message "GEWDen is stupid about Cholesky+NBsUse