Ceilidh ... Re: Conical Interaction Optimization Problem

Firefly and PC GAMESS-related discussion club

Learn how to ask questions correctly

We are NATO-free zone

Re: Conical Interaction Optimization Problem

Alex Granovsky
gran@classic.chem.msu.su

Dear Panwang Zhou,

> TOO LITTLE TIME TO DO ANOTHER POINT, TIME USED= 169114.3
> TIMLIM GIVEN= 172800.0 NEXT POINT NEEDS= 4026.5 SECONDS

The job has stopped as next geometry search step would exhausted
the allotted CPU timelimit (which can be set or changed using the
$system timlim variable). Unfortunately, the stops due to exhausted
time limit are not always graceful while running in parallel mode and
can result in hangs. This is a known problem with no elegant fix.

You need to restart this job. From what I can deduce looking
at the part of output below it is very close to convergence.
Taking this into account I'd recommend to set both shift and
shift0 to 0.001 and set stage2 to .true.

Kind regards,
Alex Granovsky

On Wed Nov 7 '12 8:42am, Panwang Zhou wrote
-------------------------------------------
>Dear all,

>When I using Firefly 8 rc33 to locate the conical interaction of retinal between S0 and S1 with MCSCF(12,12) using the following input:

>$CONTRL SCFTYP=MCSCF RUNTYP=OPTIMIZE EXETYP=RUN MAXIT=100 ICHARG=1
> MULT=1 FSTINT=.T. GENCON=.T. D5=.T. NOSYM=1 COORD=ZMT
> WIDE=.T. $END
>! Extra precision for SA gradients
>!-----------------------------------------------------
> $CONTRL INTTYP=HONDO ICUT=11 $END
> $MOORTH NOSTF=.T. NOZERO=.T. SYMS=.T. SYMDEN=.T.
> SYMVEC=.T. $END
> $DET CVGTOL=1D-7 $END
> $TRANS CUTTRF=1D-13 $END
> $MCSCF ACURCY=1D-7 ENGTOL=1.0D-12 $END
> $MCSCF NTRACK=2 $END
>!-----------------------------------------------------
> $SYSTEM MWORDS=300 KDIAG=0 NOJAC=100 $END
> $P2P P2P=.T. DLB=.T. $END
> $TRANS MPTRAN=2 DIRTRF=.T. AOINTS=DIST ALTPAR=.T. MODE=112 $END
> $MCSCF CISTEP=ALDET FULLNR=.F. SOSCF=.T. MAXIT=100 ISTATE=1 $END
> $DET NCORE=72 NACT=12 NELS=12 NSTATE=6 WSTATE(1)=1,1 $END
> $BASIS EXTFIL=.T. GBASIS=cc-pVDZ $END
> $GUESS GUESS=MOREAD NORB=465 $END
> $MCAVER JSTATE=2 CONIC=2 TARGET=MIXED $END
> $STATPT NSTEP=200 METHOD=CONIC $END
> $CONIC SHIFT0=1D-2 SHIFT=1D-3 $END
> $TRACK TOL=1.4 UPDATE=.T. $END

>After 62 cycles, the firefly hang for over 4 days, no update to output file and firefly is stilling running, due to large size of output file, I haven't attached it and only paste the last lines in the following. It seems that everything is OK but the program said "THE NEXT PREDICTED SET OF COORDINATES FOLLOWS. THEIR ENERGY AND GRADIENT IS UNKNOWN". What is the reason for this and how to fix it? Is it a bug for the hang of firefly?

>B.T.W. As the google cannot be used in these days in my country, I haven't search this in this discussion club.

> Norm(a2) = 0.1571709195 Norm(b2) = 0.0059238652
> Norm(a1) = 0.1577489750 Norm(b1) = 0.0056646555

> Alpha(a2,b2) = -0.7350671692 Alpha(a1,b1) = -0.7077329021
> Alpha(a1,a2) = 0.9998891396 Alpha(b1,b2) = 0.9366883833
> Alpha(a1,b2) = -0.7344442005 Alpha(a2,b1) = -0.7095897780

> Lambda 1 = 0.4723572074 Lambda 2 = 0.4778161661
> Lambda sc = 0.4778108999 Lambda = 0.4778108999
>
>
> DELTA ENERGY IS 0.000997824766
>
>
> NSERCH= 62 ENERGY= -829.1444838

> -----------------------
> GRADIENT (HARTREE/BOHR)
> -----------------------
> ATOM ZNUC DE/DX DE/DY DE/DZ
> --------------------------------------------------------------
> 1 C 6.0 -0.0005747 0.0000858 -0.0009105
> 2 C 6.0 -0.0001977 -0.0002411 0.0006072
> 3 C 6.0 0.0001430 -0.0002011 0.0005895
> 4 C 6.0 -0.0001807 -0.0000864 0.0004680
> 5 C 6.0 0.0005727 -0.0001515 -0.0003940
> 6 C 6.0 -0.0000348 -0.0000867 -0.0001496
> 7 C 6.0 0.0000853 -0.0005101 -0.0001875
> 8 C 6.0 0.0003814 0.0005652 -0.0004591
> 9 C 6.0 -0.0006678 -0.0002998 0.0003554
> 10 C 6.0 0.0004697 -0.0003144 -0.0002161
> 11 C 6.0 -0.0002373 0.0003476 0.0000629
> 12 C 6.0 0.0003191 0.0001739 0.0003986
> 13 C 6.0 -0.0000678 -0.0003032 -0.0005129
> 14 N 7.0 0.0005508 -0.0000767 0.0001967
> 15 H 1.0 0.0004238 -0.0003635 -0.0001561
> 16 H 1.0 -0.0000476 -0.0001445 0.0003316
> 17 H 1.0 0.0002894 0.0010020 -0.0006870
> 18 H 1.0 -0.0007163 0.0001561 0.0003434
> 19 H 1.0 0.0007289 0.0006126 0.0000210
> 20 H 1.0 -0.0001998 0.0000930 0.0001118
> 21 H 1.0 0.0008604 -0.0005630 0.0001195
> 22 H 1.0 -0.0001448 0.0005799 0.0000762
> 23 H 1.0 0.0002636 0.0000419 0.0001071
> 24 H 1.0 0.0000522 0.0001797 -0.0001029
> 25 H 1.0 -0.0001423 0.0006629 0.0003286
> 26 H 1.0 0.0004428 -0.0004590 0.0005445
> 27 H 1.0 0.0001466 0.0005788 0.0000796
> 28 H 1.0 -0.0000205 -0.0004508 0.0003209
> 29 C 6.0 -0.0001309 -0.0002108 0.0001337
> 30 C 6.0 -0.0001555 0.0000011 0.0001283
> 31 C 6.0 -0.0000465 -0.0000139 0.0001494
> 32 C 6.0 -0.0001558 -0.0001592 -0.0002548
> 33 C 6.0 -0.0000789 -0.0000450 -0.0002402
> 34 C 6.0 -0.0001309 -0.0000066 -0.0000784
> 35 C 6.0 -0.0000279 -0.0002730 -0.0002444
> 36 H 1.0 -0.0002264 0.0001539 -0.0003313
> 37 H 1.0 0.0000909 0.0000073 -0.0000024
> 38 H 1.0 0.0001453 0.0000052 -0.0000495
> 39 H 1.0 -0.0004683 -0.0000021 -0.0000317
> 40 H 1.0 -0.0002198 0.0001088 0.0002962
> 41 H 1.0 0.0001558 0.0001975 0.0000608
> 42 H 1.0 -0.0002074 0.0000074 -0.0000900
> 43 H 1.0 -0.0001414 0.0000865 -0.0000863
> 44 H 1.0 -0.0002656 0.0000265 0.0001567
> 45 H 1.0 -0.0003006 -0.0000732 -0.0001437
> 46 H 1.0 0.0002877 -0.0003071 -0.0000455
> 47 H 1.0 -0.0001605 0.0000000 -0.0000169
> 48 H 1.0 -0.0003048 0.0001017 0.0001348
> 49 H 1.0 0.0002342 -0.0001230 0.0002184
> 50 H 1.0 -0.0002848 0.0001651 -0.0001892
> 51 H 1.0 -0.0001054 -0.0004746 -0.0007606

> MAXIMUM GRADIENT = 0.0010020 RMS GRADIENT = 0.0003235
> HESSIAN UPDATED USING THE BFGS FORMULA
> RADIUS OF STEP TAKEN= 0.04026 CURRENT TRUST RADIUS= 0.10000

> TIMING STATISTICS ON NODE 0:
> CPU TIME: STEP = 0.05 , TOTAL = 169114.3 SECONDS ( 2818.6 MIN)
> WALL CLOCK TIME: STEP = 0.04 , TOTAL = 167074.8 SECONDS ( 2784.6 MIN)
> CPU UTILIZATION: STEP = 112.82%, TOTAL = 101.22%

>1 ***** FAILURE TO LOCATE STATIONARY POINT, NSERCH= 63 *****
> TOO LITTLE TIME TO DO ANOTHER POINT, TIME USED= 169114.3
> TIMLIM GIVEN= 172800.0 NEXT POINT NEEDS= 4026.5 SECONDS
> UPDATED HESSIAN, GEOMETRY, AND VECTORS WILL BE PUNCHED FOR RESTART

> retinal
> **** THE GEOMETRY SEARCH IS NOT CONVERGED! ****

> THE NEXT PREDICTED SET OF COORDINATES FOLLOWS. THEIR
> ENERGY AND GRADIENT IS UNKNOWN. YOU MAY PREFER TO RESTART
> WITH SOME OTHER COORDINATES THAN THESE.

> YOU SHOULD RESTART "OPTIMIZE" RUNS WITH THE COORDINATES
> WHOSE ENERGY IS LOWEST. RESTART "SADPOINT" RUNS WITH THE
> COORDINATES WHOSE RMS GRADIENT IS SMALLEST. THESE ARE NOT
> ALWAYS THE LAST POINT COMPUTED!

Wed Nov 7 '12 2:45pm

This message read 895 times