I am doing DFT calculation on sugar lipid system. The command I used is
"#n opt freq=noraman b3lyp/6-311++g(d,p) geom=connectivity scf=qc".
I get error message as below:
LinEq1: Iter=992 NonCon= 1 RMS=2.17D-10 Max=3.09D-09 NDo= 1
LinEq1: Iter=993 NonCon= 1 RMS=2.14D-10 Max=3.02D-09 NDo= 1
LinEq1: Iter=994 NonCon= 1 RMS=2.23D-10 Max=2.91D-09 NDo= 1
LinEq1: Iter=995 NonCon= 1 RMS=2.05D-10 Max=2.41D-09 NDo= 1
LinEq1: Iter=996 NonCon= 1 RMS=2.03D-10 Max=2.61D-09 NDo= 1
LinEq1: Iter=997 NonCon= 1 RMS=2.10D-10 Max=2.61D-09 NDo= 1
LinEq1: Iter=998 NonCon= 1 RMS=2.24D-10 Max=2.65D-09 NDo= 1
LinEq1: Iter=999 NonCon= 1 RMS=2.08D-10 Max=2.47D-09 NDo= 1
LinEq1: Iter=*** NonCon= 1 RMS=2.07D-10 Max=2.51D-09 NDo= 1
CPHF failed to converge in LinEq1.
Error termination via Lnk1e in /pkg/chem/gaussian/g09d01/g09/l508.exe at Mon Jan 5 17:54:15 2015.
Job cpu time: 10 days 13 hours 0 minutes 32.4 seconds.
File lengths (MBytes): RWF= 6033 Int= 0 D2E= 0 Chk= 98 Scr= 1
Error: segmentation violation
rax 0000000000000000, rbx ffffffffffffffff, rcx ffffffffffffffff
rdx 000000000000abd9, rsp 00007fffac218c08, rbp 00007fffac2191f0
rsi 000000000000000b, rdi 000000000000abd9, r8 0000000000000000
r9 00002afe95f83e10, r10 ffffffffffffffff, r11 0000000000000202
r12 0000000000000000, r13 0000000000000000, r14 00007fffac219238
r15 00000000000003e6
--- traceback not available
Can anyone give a suggestion on how to eradicate this error?
Thanks in advance.
This is convergence problem. Make new input file from this output and try to re-optimize. I think, your system is of large size so start with some small basis set like 6-31G(d,p).
First I have run with HF, 631G, 6311G and now 6311++G. Do I need to run from the last output file?
You may try. I think you should start with b3lyp/6-31G(d,p) first and then use last geometry for b3lyp/6-311++g(d,p) calculations.
How is the calculation terminate half the way? Should I generate the input file from the terminated output file?
Yes, generate the input file from the terminated output file using gaussview and do optimization.
I agree with @Mohommad. You can take your HF output coordinate as in put to the B3LYP. Yeah it is better to start with 6-31G(d) basis set if it run properly you can fed that out put to the B3LYP/6-311++g(d,p) calculation
In order to improve SCF/DFT convergence, especially when diffuse basis functions are used, I would recommend to use higher precision integrals:
Integral(Grid=Ultrafine,Acc2e=12)
Also, make sure that the SCF covnergence is tight enough. As you use SCF=QC, I guess you had convergence problems in your SCF calculation: try to add SCF(QC,Conver=9) and see whether you can get to convergence.
- Badges
- Science topic