Alex Granovsky
gran@classic.chem.msu.su
I fixed missed $end in your input, generated orbitals for three-state
sa-casscf, and ran two-state and five-state jobs using these orbitals.
I cannot confirm that CASSCF energies are the same.
For two-state SA-CASSCF: FINAL MCSCF ENERGY IS -248.1545279418 AFTER 15 ITERATIONS -MCCI- BASED ON OPTIMIZED ORBITALS ---------------------------------- STATE # 1 ENERGY = -248.156145506 CSF COEF OCCUPANCY (IGNORING CORE) --- ---- --------- --------- ----- 51 -0.942722 221100 55 0.060463 121101 59 0.107271 121101 62 0.062682 022101 68 -0.100803 021102 71 -0.111556 211110 87 -0.194449 211110 109 -0.139984 201120 127 0.075022 211200 STATE # 2 ENERGY = -248.152910378 CSF COEF OCCUPANCY (IGNORING CORE) --- ---- --------- --------- ----- 1 -0.961798 222000 5 0.133538 220002 10 0.070634 022002 34 0.098154 202020 52 -0.063677 212100 89 -0.050022 202110 129 0.174534 202200 ----------------------------------------------------------------------- *** MC-XQDPT2 ENERGIES *** ----------------------------------------------------------------------- STATE 1ST ORDER 2ND ORDER 1 E(MCSCF)= -248.1561455060 E(MP2)= -248.9079919760 2 E(MCSCF)= -248.1529103775 E(MP2)= -248.9047150085 ----------------------------------------------------------------------- For averaging over five states: FINAL MCSCF ENERGY IS -248.0521602867 AFTER 12 ITERATIONS -MCCI- BASED ON OPTIMIZED ORBITALS ---------------------------------- STATE # 1 ENERGY = -248.160581722 CSF COEF OCCUPANCY (IGNORING CORE) --- ---- --------- --------- ----- 51 -0.938413 221100 55 0.078696 121101 59 0.106778 121101 62 0.067525 022101 68 -0.106713 021102 71 -0.122049 211110 87 -0.192858 211110 109 -0.141389 201120 127 0.083731 211200 STATE # 2 ENERGY = -248.142663331 CSF COEF OCCUPANCY (IGNORING CORE) --- ---- --------- --------- ----- 1 0.961168 222000 5 -0.140032 220002 10 -0.074236 022002 34 -0.096614 202020 52 0.052533 212100 129 -0.180044 202200 STATE # 3 ENERGY = -248.034961487 CSF COEF OCCUPANCY (IGNORING CORE) --- ---- --------- --------- ----- 11 0.594309 221010 15 -0.051170 121011 19 -0.068569 121011 28 0.068025 021012 32 -0.211827 211020 71 -0.069418 211110 127 0.715312 211200 136 -0.091056 111201 139 0.050888 012201 144 -0.080666 011202 149 -0.209935 201210 STATE # 4 ENERGY = -247.996749751 CSF COEF OCCUPANCY (IGNORING CORE) --- ---- --------- --------- ----- 53 -0.758161 122100 55 0.129638 121101 57 0.442667 220101 59 0.302257 121101 62 -0.106562 022101 64 0.158318 120102 73 0.098070 112110 75 0.057152 210111 90 -0.159442 112110 92 0.092872 210111 96 0.054112 111111 112 0.114868 102120 117 0.067192 200121 STATE # 5 ENERGY = -247.925845142 CSF COEF OCCUPANCY (IGNORING CORE) --- ---- --------- --------- ----- 2 -0.179547 221001 52 -0.888610 212100 63 0.125438 210102 70 -0.069855 012102 89 -0.359776 202110 100 0.051520 200112 129 -0.110300 202200 ----------------------------------------------------------------------- *** MC-XQDPT2 ENERGIES *** ----------------------------------------------------------------------- STATE 1ST ORDER 2ND ORDER 1 E(MCSCF)= -248.1605817219 E(MP2)= -248.9151665032 2 E(MCSCF)= -248.1426633312 E(MP2)= -248.9055013189 3 E(MCSCF)= -248.0349614872 E(MP2)= -248.7800596220 4 E(MCSCF)= -247.9967497511 E(MP2)= -248.7466441649 5 E(MCSCF)= -247.9258451419 E(MP2)= -248.7369833618 -----------------------------------------------------------------------
See attached file. Note, I optimized your input file a bit.
However, I checked that this does not affect CASSCF results.
This does slightly affected QDPT results as I removed CSF
selection as it is intended for use with very large active spaces.
Thus, now QDPT procedure gives exact numbers rather than
approximations to these exact numbers.
Kind regards,
Alex Granovsky
On Mon Jul 11 '11 0:22am, mark huntress wrote
---------------------------------------------
>Hi, thank you Dr. Granovsky.
>
>I think I should have been more specific in my explanation, because now I am not sure what you think I was saying I did.
>I was using WSTATE(1)=1, 1, 1 ( or WSTATE(1)=1, 1, 1, 1, 1) in the $GUGDM2 group, not in the $xmcqdpt group.
>I was using KSTATE(1)=1, 1, 1 in the $xmcqdpt group.
>In fact, here is my input:
> $contrl scftyp=mcscf
> runtyp=energy maxit=100
> icharg=1 mult=1
> MPLEVL=2
> fstint=.t. gencon=.t.
> exetyp=run
> $end
> $SYSTEM TIMLIM=60 MEMORY=204000000
> idle=.t. mklnp=2
> nojac=1 kdiag=0 $END
> $smp csmtx=.t. call64=1 $end
> $p2p p2p=1 dlb=1 $end
> $SCF DIRSCF=.T. DIIS=.T. SOSCF=.F. SHIFT=.F.
> $END
> $moorth nostf=1 nozero=1 tole=0 tolz=0 $end
> $MCSCF CISTEP=GUGA fullnr=.f. ISTATE=1 acurcy=5d-7 ENGTOL=5.0d-12
> soscf=.t. maxit=150 $END
> $TRANS dirtrf=.t. cuttrf=1d-12
> aoints=dist altpar=.t.
> mptran=2 mode=110 $end
> $DRT NMCC=19 NDOC=3 NALP=0 NVAL=3 FORS=.t. $end
> $GUGDIA NSTATE=3 ITERMX=100 cvgtol=1d-6 $end
> $GUGDM2 cutoff=1d-11 WSTATE(1)=1, 1, 1 $end
> $GUGEM pack2=.t. cutoff=1d-11 $end
> $XMCQDPT KSTATE(1)=1, 1, 1 MAXROW=10000
> MXTRFR=1000 iselct(1)=-3, 100000
> svdisk=.t.
> INORB=0
> edshft=0.02
> length=30000
> $basis GBASIS=N31 NGAUSS=6 NDFUNC=1 NPFUNC=0
> DIFFSP=.f. DIFFS=.f. $END
> $GUESS GUESS=moread NORB=106 norder=0
> $END
> $DATA
> title
> C1
> C1 6.0 -2.857095 -0.867132 0.802361
> C2 6.0 -1.477967 -0.898110 0.843796
> C3 6.0 -0.702403 -0.054721 0.046182
> C4 6.0 0.755594 -0.069250 0.031595
> C5 6.0 1.519947 0.760181 0.841547
> N6 7.0 2.836821 0.792683 0.846945
> H7 1.0 1.244644 -0.730447 -0.665136
> H8 1.0 -1.210120 0.611986 -0.633252
> H9 1.0 1.040969 1.434499 1.527664
> H10 1.0 -0.984194 -1.584551 1.508069
> H11 1.0 3.337133 1.405588 1.455677
> H12 1.0 3.383002 0.212329 0.243669
> H13 1.0 -3.445457 -1.511684 1.424304
> H14 1.0 -3.382177 -0.198160 0.147581
> $end
>--- OPTIMIZED MCSCF MO-S --- GENERATED AT 19:01:39 LT 6-JUL-2011
>.
>.
>.
>etc...
>So I am curious, does looking at the input change your answer at all?
>If not,
> Is there a difference between "constant state averaging" that you mention and the normal state averaging I am used to? If so, is there something I can read about that? In molcas and gaussian the energies of the states are severely affected by the number of roots included in the state averaging, so I am still trying to understand what is different.
>thank you so much,
>Mark
>
>
>
This message contains the 243 kb attachment [ different_averaging.zip ] |