wb97m-d3bj/def2-tzvppd: still stuck at 3468 with 950+ incomplete jobs. Submitted fat node jobs (48 cores/240 GB/ 500GB scratch) on UCI cluster to get this out but all are QCEngine unknown failures.
PB – I’ll try one of these failing jobs locally and watch it closely. Not sure whether it’s resource requirements or something else.
SPICE sets with openff-default spec
MBIS errors with Iodine containing molecules seems suspicious
In code:
/* Updated parameters for initial MBIS params Nai and 1/Sai
(from https://github.com/theochem/denspart/blob/740449c375f7e1c7b14cc8a978a8d0b1ed4617e3/denspart/mbis.py#L82 by Toon
Verstraelen) */
// Attention: MBIS is not supported for elements with atomic number greater than 36
Fatal Error: Atomic Number 53 unsupported by MBIS
PB – I’m confused, since we’ve previously done sets using Iodine.
DD – Do we not ask for MBIS charges in our default level of theory?
PB – It looks like we DON’T have MBIS set in our default, but we ARE requesting MBIS in SPICE. But MBIS doesn’t work for Iodine.
DD – Does method and basis interact with MBIS?
PB – Maybe.
DD – We’d worked with the psi4 devs before about this, right?
PB – Yeah, they has auxilliary basis for stuff up to Bromine, and after talking to us they added it up to iodine.
PB – I’ll bring this us with psi4 devs.
Complete
SPICE Solvated Amino acids
Moved to end of life:
SPICE DES370K Single Points Dataset v1.0
SPICE DES Monomers Single Points Dataset v1.1
SPICE Ion Pair
SPICE Dipeptides Single Points Dataset v1.2
SPICE Pubchem sets 1,2, (3)
In queue (being computed):
SPICE Pubchem set 4,5,6
User questions/issues
DD: Node allotment?
PB – The current mix looks good. We have the SPICE sets going through and the current number+size of nodes look good.
CC: Geometric release?
JW – I’ll be running a forcebalance release soon. So I’ll be in contact with LPW. Could push on this. It’s not at the top of my to-do list so it may be >1 week out. Is that OK?
PB + CC – Yes.
DD – Currently can’t submit jobs on MPI cluster. I’ll work with sysadmins to figure out why. They were of limited use anyway because the nodes didn’t have a ton of scratch space. These were consuming jobs from openff-uci and openff-tscc labels. 3mers are on the TSCC tag, uci tag doesn’t have any active sets right now.
Add Comment