Q-Chem fatal error occurred in module libmdc/newfileman.C, line 376

Dear Support
I was running a eom-sf-ccsd job using complex basis function using the below input file

$molecule
1 1
O 0.0000 0.0000 0.0000
N 0.0000 0.0000 1.1451
$end

$rem
METHOD hf
unrestricted false
BASIS aug-cc-pVTZ
SCF_CONVERGENCE 7
SCF_GUESS CORE
SCF_ALGORITHM DIIS
$end

@@@

$molecule
0 4
O 0.0000 0.0000 0.0000
N 0.0000 0.0000 1.1451
$end

$rem
BASIS gen
complex_theta = 10
METHOD eom-ccsd
SF_STATES [5,0,0,0]
SCF_CONVERGENCE = 8
CC_CONVERGENCE = 7
eom_davidson_convergence = 7
eom_davidson_maxvectors = 300
eom_davidson_max_iter = 350
complex_ccman 1
cs_ccsd 1
complex_exponents 1
complex_scf 3
complex_metscf 1
complex_basis gen
complex_scf_guess 1
cc_max_iter = 350
! thresh = 13
MAX_SCF_CYCLES 300
CC_TRANS_PROP = true
CC_EOM_PROP_TE = 1
CC_REF_PROP_TE = 1
unrestricted false
SCF_ALGORITHM DIIS
SCF_GUESS READ
! eom_shift = 1580
! mem_total 50000
! cc_memory 40000
! mem_static 2000
! MEM_TOTAL = 100000
! CC_MEMORY = 50000
$end

$complex_ccman
cs_alpha 1000
cs_theta 0
$end

I got this error just after ccman2 step

Q-Chem fatal error occurred in module libmdc/newfileman.C, line 376:

FileMan error: End of file reached prematurely reading (107648) bytes in file UNKNOWN FILE
Path: /state/partition1/qchem279851/2061.0

Please submit a crash report at Q-Chem Crash Reporter

I have submitted the job in a 40 core cpu with 172 GB RAM (TOT MEM). Can anyone have any idea regarding this problem.

Thanks
Abhisek

I can’t diagnose this without the complete input file; please use the preformatted text button, </>

It would also be useful if you could identify the minimal example that fails, rather than an input fiile with 50 different keywords.

Hi John
I attach a complete input file regarding this problem. This is a open shell system. When I used complex_scf 3 in input file for restricted open-shell calculation I found this error, but when I used complex_scf 2 in input file for unrestricted calculation then it completes successfully.

// input file
$molecule
0 4
O 0.0000 0.0000 0.0000
N 0.0000 0.0000 1.1451
$end

$rem
BASIS cc-pVTZ
complex_theta = 10
METHOD eom-ccsd
SF_STATES [1,0,0,0]
complex_ccman 1
cs_ccsd 1
complex_exponents 1
complex_scf 3
complex_metscf 1
complex_basis aug-cc-pVTZ
complex_scf_guess 1
$end

$complex_ccman
cs_alpha 1000
cs_theta 0
$end

Thanks
Abhisek

I can reproduce this error with the current Q-Chem trunk (pre-release v. 6.2), although interestingly, I had to turn down the convergence thresholds to get to the CCMAN2 step, otherwise the step

 -- obtaining guess coefficients from real R/ROHF calculation

does not converge in 50 iterations. I can also reproduce the fact that the job runs to completion with COMPLEX_SCF=2.

This is beyond my capabilities to pursue. I will create a ticket but you might also contact Q-Chem user support. I notice there is no example in the manual for COMPLEX_SCF=3, which should also get fixed.

Thanks John for the insight.