Problems in RAS-CI calculations of open shell radical system

Questions
#1

Hello,

I am new to Q-Chem and am trying to calculate some radical transition states using the RAS-CI methodology. I successfully obtained the transition states for a radical Benzyl using this input:

$rem
jobtype sp
exchange HF
correlation RASCI2
basis def2-svp
aux_basis RIMP2-def2-svp
unrestricted false
RAS_ACT_OCC 6 ! # alpha electrons
RAS_ACT_VIR 2 ! # virtuals in active space
RAS_ACT_DIFF 1 ! # set to 1 for odd # of e-s
ras_n_roots 19
$end

$molecule
0 2
…

However, I had several problems when I tried with a more complex Trityl radical. A segmentation fault error appears:

exported envs: QC QCAUX QCSCRATCH QCRUNNAME QCFILEPREF QCPROG QCPROG_S GUIFILE
/opt/cecisw/arch/manual/tis-2017.01/software/Q-Chem/6.1.0-SHMEM/bin/qchem: line 129: 22764 Segmentation fault ${QCPROG_S} ${inp} ${scr}
Error in Q-Chem run part 1

From what I understood, it seems linked with a memory issue. However, I did not manage to solve it, modifying the memory setup:

MEM_STATIC 2000
MEM_TOTAL 40000

Is there a keyword to impose dynamic memory allocation?
I will leave here my final full input file. Thank you in advance for the help.

$rem
jobtype sp
exchange HF
correlation RASCI2
basis def2-svp
aux_basis RIMP2-def2-svp
unrestricted false
RAS_ACT_OCC 7 ! # alpha electrons
RAS_ACT_VIR 4 ! # virtuals in active space
RAS_ACT_DIFF 1 ! # set to 1 for odd # of e-s
ras_n_roots 5
!MEMORY
MEM_STATIC 2000
MEM_TOTAL 40000
MAX_CIS_CYCLES 120
$end

$comment
C19H15
$end

$molecule
0 2
C 1.919721 -1.454914 0.686007
C 3.273779 -1.760037 0.674894
C 1.423488 -0.324283 0.008248
C 4.172842 -0.949979 -0.015516
C 2.349530 0.484097 -0.679483
C 3.702100 0.173096 -0.692340
C -0.004307 -0.001049 0.013031
C -0.997711 -1.074201 0.004062
C -2.233435 -0.932754 0.666452
C -0.753446 -2.287508 -0.669511
C -3.173331 -1.953829 0.655769
C -1.697613 -3.304333 -0.681301
C -2.912696 -3.145086 -0.018379
C -0.434250 1.397973 0.014333
C 0.299055 2.384625 0.702154
C -1.592420 1.806718 -0.675454
C -0.107949 3.711560 0.701692
C -1.993403 3.135563 -0.677879
C -1.255241 4.095752 0.010803
H 1.230322 -2.083781 1.241275
H 3.630970 -2.633023 1.213586
H 5.231597 -1.191510 -0.025488
H 1.990524 1.353550 -1.221901
H 4.391794 0.806360 -1.242875
H -2.440276 -0.015391 1.209277
H 0.183601 -2.415169 -1.202867
H -4.113419 -1.822922 1.183964
H -1.487168 -4.224780 -1.218309
H -3.649944 -3.942298 -0.027124
H 1.186434 2.093026 1.255842
H -2.165791 1.069683 -1.229567
H 0.470112 4.449395 1.250558
H -2.883827 3.425655 -1.228294
H -1.571202 5.134767 0.008790
$end

#2

Memory allocation is dynamic, with a max value set by MEM_TOTAL. (MEM_STATIC probably doesn’t need to be changed for this job.) I suggest you set MEM_TOTAL to the physical limit of your hardware and also try a minimal basis first, to make sure there aren’t other problems.

#3

Thank you for the answer. I tried to do what you told me, but the error remained the same regardless of the amount of MEM_TOTAL imposed. The calculation starts and ends the open shell SCF step calculation, but when the RAS-CI actual part starts after a while, it stops with the segmentation error. Could this be originated by another problem that isn’t the memory?

#4

Does it work in a minimal basis?

#5

Excuse me if I have not replied until now. I was on holiday. I tried with a basis set:
basis cc-pVDZ
aux_basis RIMP2-VDZ
but the results are the same.