計算化学研究会

GAMESS/FMOワークショップ 事前準備、プログラム、tutorial


[事前準備]


2004.07.14 の時は Program の install と実行はクラスター計算機を使用したの
で、各自は端末としての Note-PC を持参することとなり、その事前準備を以下のように案内した。2004.10.13 の時は、Program の install と実行を各自持参の Note-PC で行った。(これは、上林世話人が Network 環境設定でミスをしたため。参加者にご迷惑をおかけして申し訳ありません。) そのための事前準備は以下の通りです。

-------2004.07.14----------事前準備--------------------------


Note-PC の OS が Linux でなく MS-Windows である時は TeraTermPro, TTSSH, 
WinSCP, PuTTY, FFFTP 等が必要になるかも知れません。これらの参考 web site を下記
に示します。

Guide for SSH: http://osksn2.hep.sci.osaka-u.ac.jp/~naga/miscellaneous/winsshb.html
TeraTerm: http://hp.vector.co.jp/authors/VA002416/
TTSSH: http://www.sakurachan.org/soft/teraterm-j/ttssh/
WinSCP: http://www.tab2.jp/~winscp/

FFFTP は SSH は suport していませんが.....
FFFTP: http://www2.biglobe.ne.jp/~sota/ffftp.html

GAMESS license 事前登録については各組織で行っておいてください。FreeWare です。
GAMESS Web Home Page URL http://www.msg.ameslab.gov/GAMESS/GAMESS.html 内の
"How to get GAMESS" から順にたどっていってください。一日位で down load の案
内が送られてきます。Down load Software の中に manual も入っていますが大部で
すので...... お時間があればお読み下さいとのこと。

-------2004.10.13----------事前準備-------------------------------


Note-PC の OS が MS-Windows OS の場合には cygwin を install しておく。Cygwin 
は "ALL install" を選択するか、開発環境を確実に install し、"user 設定" を済ま
せ、"tcsh (csh)", "gcc/g77" が確実に動作する事を確認しておく(後日追加)。
Linux や Macintosh OS-X の場合にはそのまま ( shell 環境 ) で結構。

参考 web site を下記に示します。
Cygwin: http://www.cygwin.com/
Cygwin-JE: http://www.sixnine.net/cygwin/
Cygwin/X: http://x.cygwin.com/

計算結果の表示に X-Window で gnuplot を使う場合: Cygwin に gnuplot があります。

参考 web site を下記に示します。
http://www.ifs.tohoku.ac.jp/cryogenic/gakusei/iwata/cygwin/
http://coconut.sys.eng.shizuoka.ac.jp/docs/gnuplot.html

GAMESS license 事前登録については "2004.07.14 事前準備" と同じ。


[プログラム]


Seminar on 2004.07.14-----------プログラム--------------------------


講習内容:
     講師: Dr. Dmitri G. Fedorov
     講師: Dr. 石田豊和 (午後の部 3. 6. 担当)

10:00-  GAMESS install 実習 --> *2
     linux 用 BLAS library の install 10分間
     若干の linux 環境設定のガイド 10分間
     GAMESS の install 20分間 --> *3 
     GAMESS の install 確認用試験 data の投入確認 20分間
昼休み
13:00- 
     1. GAMESS 入力 data の一般的なガイド 30分間 --> *4

     2. CH3CH2OH + H2O の系での GAMESS 実行
          最初の説明 30分間
          実行 数秒間
          計算結果出力の説明 30分間

     3. PDB から download した蛋白質の計算 1時間
      PDB から適当な file の download --> *5 
      PDB 形式から GAMESS 入力形式変換用 software: fmoutil の使用説明
      入力 data file 作成
      GAMESS 実行 --> *6 (代表実行)

     4. (FMO の説明を 3.のGAMESS 実行中に行う。) 1時間
      並列計算の説明 SMP
      fragment 分割
      基底関数
      精度比較

     5. 3.の GAMESS 実行の結果出力 data の説明 1時間

     6. 蛋白質モデリングについて 1時間
           低分子力場作成
           AMBER, QM/MM 等

18:00- 終了予定


[Tutorial]
----------------------------------Tutorial--------------

! How to make FMO input and how to look at the results.
! A brief FMO tutorial,
! written by D. G. Fedorov
! AIST, Japan
! July 5, 2004
!
! This tutorial is based on the Dr Sc. thesis ("PhD") of D. I. Mendeleev, that 
! was devoted to the one of the most micro and macro-economically important 
! problems in Russia: mixing of ethyl alcohol and water.
! You will learn how to mix C2H5OH and H2O using the FMO method.
!
! C2H5OH+H2O will be divided into 3 fragments:
! 1. CH3
! 2. CH2OH
! 3. H2O
! That is, CH3 | -CH2OH | H2O
!
! Start from this GAMESS file: 

! $CONTRL SCFTYP=RHF RUNTYP=ENERGY $END
! $SYSTEM TIMLIM=2 MEMORY=100000 $END
! $BASIS GBASIS=STO NGAUSS=3 $END
! $DATA
!C2H5OH+H2O
!C1
! C 6.0 2.3410689175 -0.2869692888 -0.0074194092
! H 1.0 3.0745859649 0.3772736987 0.4397744143
! H 1.0 2.5665310430 -0.3924000324 -1.0640918137 
! H 1.0 2.4261794556 -1.2632979826 0.4595623356
! C 6.0 0.9166901279 0.2761650904 0.1831975319
! H 1.0 0.7235442032 0.4041423414 1.2567611875 
! H 1.0 0.8641656999 1.2758468598 -0.2685095421
! O 8.0 -0.0215616632 -0.6201531625 -0.4156796115
! H 1.0 -0.9026816335 -0.1944297425 -0.2534321184
! O 8.0 -2.4493614824 0.5180105259 0.0102319306 
! H 1.0 -2.9309841137 0.6564728575 -0.8399969145 
! H 1.0 -3.0583517680 -0.1059613981 0.4726454459 
! $end
!
! 1. Keep common groups:
! $CONTRL SCFTYP=RHF RUNTYP=ENERGY $END
! $SYSTEM TIMLIM=2 MEMORY=100000 $END
! $BASIS GBASIS=STO NGAUSS=3 $END
!2. Add FMO-specific groups.
!First remember that $DATA in FMO is used to define the basis set, not 
!coordinates. If you define the basis set in $BASIS as above, then $DATA 
!becomes totally dummy. Nevertheless, put each sort of atom and append a 
!dummy to produce an even number of electrons. In this case you need C, O and H,
!add a dummy B to have 1+6+8+5=20 electrons. Cartesian coordinates are ignored,
!but make sure atoms do not occupy exactly the same place. 
!Advanced usage: you can define any non-standard basis set in $DATA. 
! $data
!C2H5OH+H2O
!c1
!h-1 1 1 0 0
!c-1 6 2 0 0
!o-1 8 3 0 0
!b-1 5 4 0 0 
! $end
!Now, define coordinates in $FMOXYZ.
! $FMOXYZ
! C 6 2.3410689175 -0.2869692888 -0.0074194092
! H 1 3.0745859649 0.3772736987 0.4397744143
! H 1 2.5665310430 -0.3924000324 -1.0640918137
! H 1 2.4261794556 -1.2632979826 0.4595623356
! C 6 0.9166901279 0.2761650904 0.1831975319
! H 1 0.7235442032 0.4041423414 1.2567611875
! H 1 0.8641656999 1.2758468598 -0.2685095421
! O 8 -0.0215616632 -0.6201531625 -0.4156796115
! H 1 -0.9026816335 -0.1944297425 -0.2534321184
! O 8 -2.4493614824 0.5180105259 0.0102319306
! H 1 -2.9309841137 0.6564728575 -0.8399969145
! H 1 -3.0583517680 -0.1059613981 0.4726454459
! $end
! Charges (second column) are expected to be integer, not real (as in $DATA).
! Then, create an $FMO group that defines fragmentation.
! You have three fragments (nfrag=3).
! For each atom assign its fragment number:
! indat(1)=1,1,1,1, 2,2,2,2,2, 3,3,3
! CH3 CH2OH H2O
!The total group then is:
! $fmo 
!nfrag=3 indat(1)=1,1,1,1, 2,2,2,2,2, 3,3,3 $end
!
! $fmo group name should not be followed by other keywords on the same line. 
!
!The atoms happen to be ordered but that is not required.
!Next, define fragmented bonds:
!A bond between atoms 1 and 5 is fractioned. (C-C), that is, CH3|-CH2OH.
!The other choice is CH3-|CH2OH, which differs from the former choice since 
!bonds are fractioned not between but at atoms. 
!The general rule is to avoid breaking at atoms involved in delocalisation. 
!In this case we break like this: CH3|-CH2OH, so the group becomes:
! $FMOBND
! -1 5 sto-3g mini
! $END
!If you wish to compare with the other choice, use -5 1 instead of -1 5.
!If you have more bonds to fraction, just add more lines.
!After the two numbers (-1 5) giving pairs of atoms between which a bond is
!fractioned, two other words appear: sto-3g mini. They refer to sets of LMOs,
!that you define in $FMOLMO. The first set (sto-3g) is for your basis set
!(STO-3G). The second is for the initial guess which is normally HUCKEL, using
!(always) MINI basis set. 
!LMOs are provided in ~/gamess/tools/fmo/LMOs.txt, or wherever your GAMESS be.
!From there you get:
! $FMOLMO
! sto-3g
! 5 5
! 1 0 -0.117784 0.542251 0.000000 0.000000 0.850774
! 0 1 -0.117787 0.542269 0.802107 0.000000 -0.283586
! 0 1 -0.117787 0.542269 -0.401054 -0.694646 -0.283586
! 0 1 -0.117787 0.542269 -0.401054 0.694646 -0.283586
! 0 1 1.003621 -0.015003 0.000000 0.000000 0.000000
! mini 
! 5 5
! 1 0 -0.109772 0.515046 0.000000 0.000000 0.864512
! 0 1 -0.109775 0.515062 0.815062 0.000000 -0.288166
! 0 1 -0.109775 0.515062 -0.407531 -0.705864 -0.288166
! 0 1 -0.109775 0.515062 -0.407531 0.705864 -0.288166
! 0 1 0.996474 0.015610 0.000000 0.000000 0.000000
! $END
!None of the numbers needs to be adjusted for a given molecule, once you get 
!the proper set for your basis set, just put them together and append 
!$FMOLMO/$END.
!
!3. Congratulations, you have just built your first FMO input, and now you can
!start playing with alcohol in water.
!Summarising, your input is:
$CONTRL SCFTYP=RHF RUNTYP=ENERGY $END
$SYSTEM TIMLIM=2 MEMORY=100000 $END
$BASIS GBASIS=STO NGAUSS=3 $END
$data
C2H5OH+H2O
c1
h-1 1 1 0 0
c-1 6 2 0 0
o-1 8 3 0 0
b-1 5 4 0 0
$end
$FMOXYZ
C 6 2.3410689175 -0.2869692888 -0.0074194092
H 1 3.0745859649 0.3772736987 0.4397744143
H 1 2.5665310430 -0.3924000324 -1.0640918137
H 1 2.4261794556 -1.2632979826 0.4595623356
C 6 0.9166901279 0.2761650904 0.1831975319
H 1 0.7235442032 0.4041423414 1.2567611875
H 1 0.8641656999 1.2758468598 -0.2685095421
O 8 -0.0215616632 -0.6201531625 -0.4156796115
H 1 -0.9026816335 -0.1944297425 -0.2534321184
O 8 -2.4493614824 0.5180105259 0.0102319306
H 1 -2.9309841137 0.6564728575 -0.8399969145
H 1 -3.0583517680 -0.1059613981 0.4726454459
$end
$fmo
nfrag=3 indat(1)=1,1,1,1, 2,2,2,2,2, 3,3,3 $end
$FMOBND
-1 5 sto-3g mini
$END
$FMOLMO
sto-3g 5 5
1 0 -0.117784 0.542251 0.000000 0.000000 0.850774
0 1 -0.117787 0.542269 0.802107 0.000000 -0.283586
0 1 -0.117787 0.542269 -0.401054 -0.694646 -0.283586
0 1 -0.117787 0.542269 -0.401054 0.694646 -0.283586
0 1 1.003621 -0.015003 0.000000 0.000000 0.000000
mini 5 5
1 0 -0.109772 0.515046 0.000000 0.000000 0.864512
0 1 -0.109775 0.515062 0.815062 0.000000 -0.288166
0 1 -0.109775 0.515062 -0.407531 -0.705864 -0.288166
0 1 -0.109775 0.515062 -0.407531 0.705864 -0.288166
0 1 0.996474 0.015610 0.000000 0.000000 0.000000
$END
!
!4. Finally, if your experiments have not lead to drowsiness and a mood to
!philosophise does not obstruct considering banal mundanities, let us look at
!the results. First of all, you may be surprised by the large output.
!This is good for understanding how the method works and what is happening,
!but normally you do not want to have as much output for real work
!(look at the ~/gamess/tools/fmo/water-16.inp sample for more options,
!in case of output these are the two nprint keywords).
!In the first half of the output 3 RHF calculations for each fragment are
!repeated self-consistently, which happens to be 13*3 times. Then each
!pair of fragments is computed once. Finally, the total properties are
!computed. Two major total properties are computed:
!
!Total energy of the molecule: Euncorr(2)= -227.107864787
!Dipole moment D(xyz),DA(2)= -2.7379615 0.7007414 0.1343559 2.8294034
!
!"2" in paretheses, Euncorr(2) or DA(2) refer to the 2-body method, that is what
!you most likely use for the final answers. These values can be compared with 
!the ones produced by the original (regular RHF) input:
!
!FINAL RHF ENERGY IS -227.1083045619
!Dipole moment: -2.766892 0.708072 0.135749 2.859280
!
!Thus you can observe that the error of the method in this case is 
! |-227.107864787-(-227.1083045619)|=.0004397749 a.u. or 0.28 kcal/mol.
! Another result worth of mentioning is the hydrogen bond between OH and OH2,
! which has the energy of -4.868 kcal/mol (interaction between fragments 2 and
! 3). Great number of other things can be learned, such as how much dipole
! moment of water and alcohol changes when they polarise each other etc.

! If you wish, you can add more water or alcohol, until satisfied. But please
! do not quite follow the grand maitre (Mendeleev) who is said to have used
! quantities measured in Russian units "buckets" (10 L) for this purpose. 
!
! 5. Normally alcohol is not subdivided into fragments (here it was done to
! llustrate the framentation process), and the fragment size is typically
! 10-30 atoms (the larger the size, the higher the accuracy, but slower the 
! method).
!
! 6. Advanced options
! indat has an alternative format that may be more appealing to you.
! FMOutil uses that alternative format.
! indat(1)=1,1,1,1, 2,2,2,2,2, 3,3,3 can be written as 
! indat(1)=0, 1, -4, 0, 5, -8, 9, 0, 10, -12, 0 
! Here, the first 0 indicates that the alternative form is used.
! Then atoms (possibly with intervals, such as 1-4) for each fragment are 
! entered, terminated by 0.

! 6. More information on Mendeleev can be found in Russian at:
! http://www.spbu.ru/History/275/Chronicle/spbu/Persons/M_endeleev.html 
! where the thesis title is given ("On the combination of the alcohol 
! with water") and "the" was used since there is only one alcohol in Russia.
!