Dear Dirac experts,
I’ve been running a few KR-MCSCF calculations and have run into some weird behavior.
1) Is there something special I need to do to restart these calculations from the previous orbitals besides just saving and using the CHECKPOINT.h5 file? I had assumed that everything that was needed was included there, but upon restart (and the output says that start orbitals are read from the checkpoint file) the energy is pretty close to what it was in the 1st iteration of the original run (which took 3 days before it quit due to too many macro iterations).
2) Also, as I was checking the progress of the original run, the total KR-MCSCF energies for the last few iterations were looking like this (obtained via grep):
Total KR-MCSCF energy : -86951.062349903353606
Total KR-MCSCF energy : -86951.064155763480812
Total KR-MCSCF energy : -86951.065897396329092
Total KR-MCSCF energy : -86951.066909556728206
Total KR-MCSCF energy : -86951.067753212540993
Total KR-MCSCF energy : -86951.067894136460382
Total KR-MCSCF energy : -86951.068276008518296
Total KR-MCSCF energy : -86951.068526746516000
But at the end when it prints out a summary, it only shows up to through the 19th iteration and not all 25:
15 -87235.446654255400 -23.861127488759 -86951.057781744166
16 -87235.453567853867 -23.854893104497 -86951.058460958360
17 -87235.461655292675 -23.847992106801 -86951.059647399481
18 -87235.464778236885 -23.846297288779 -86951.061075525664
19 -87235.462961806261 -23.849388097095 -86951.062349903354
Ideas? The first one is obviously the worst, but I’m concerned the last issue means the final orbitals that are displayed are not actually the converged ones.
Best regards,
-Kirk
Hi HJJ,
maybe this could be flagged as an issue ?
All the best,
Trond
To view this discussion visit https://groups.google.com/d/msgid/dirac-users/DB9PR08MB774629F08EA46F91503B18C9D11C2%40DB9PR08MB7746.eurprd08.prod.outlook.com.
Dear Hans Jørgen,
Aha, that definitely explains it. I wish I hadn’t cleaned up the run directory already…
Best regards,
-Kirk
Oh, and thanks for the quick reply!
Best regards, -Kirk
From: peterso...@gmail.com <peterso...@gmail.com>
Date: Friday, January 10, 2025 at 10:07 AM
To: <dirac...@googlegroups.com>
Subject: Re: [dirac-users] Re: KR-MCSCF questions
Dear Hans Jørgen,
Aha, that definitely explains it. I wish I hadn’t cleaned up the run directory already…
Best regards,
-Kirk
From: 'Hans Jørgen Aagaard Jensen' via dirac-users <dirac...@googlegroups.com>
Date: Friday, January 10, 2025 at 8:55 AM
To: dirac...@googlegroups.com <dirac...@googlegroups.com>
Subject: [dirac-users] Re: KR-MCSCF questions