GA ERROR fehler?
H. -J. Werner
werner at theochem.uni-stuttgart.de
Wed Oct 8 16:39:15 BST 2003
I ran your job using the current mpp production version (2002.6) using
2 processors on one of our itanium2 machines (hp-ux). It did the
integrals and sort without any problems and finished the first 2 scf
iterations:
ITERATION DDIFF GRAD ENERGY 2-EL.EN. DIPOLE MOMENTS DIIS
1 0.000D+00 0.000D+00 -2612.08207685 2224.622343 -1.007284 -0.000055 -0.000023 0
2 0.000D+00 0.591D-03 -2612.09687629 2224.359070 -0.733499 -0.000033 -0.000018 1
I think we fixed the problem you had some time ago, and I see changes in this part
of the program relative to 2002.3. Therefore, the solution for you is to
get 2002.6. We cannot make any fixes for 2002.3 any more.
The question remains if you will be able to do the CCSD(T). This
will be very expensive and might require an enormous amount
of memory. The integral file is about 65 GB, and during the sort
it needed 134 GB in total.
Good luck!
Joachim Werner
On Die, 07 Okt 2003, Scott Yockel wrote:
>This job was ran on molpro2002.3 i8 on SGI Origin 2000 at NCSA.
>
>See attached input file.
>
>Kirk Peterson sudgested that I "try increasing the size of the parameter mxbuck
>in src/sort/aosortv.f and re-make. This is not a GA error, but just a
>dimensioning problem in the integral sort routine."
>
>But I want to make sure of this before I ask the people at NCSA if I can get
>them to recompile a special version of molpro for me to fix this.
>
>Quoting "H. -J. Werner" <werner at theochem.uni-stuttgart.de>:
>
>> This should certainly not happen. Are you using an i4 or i8
>> version? Which molpro version? If 2002.6, could you please send
>> your input so we can check it out?
>>
>> The max number of basis functions can be modified using configure.
>> The default in 2002.6 is 2000 contractions, and so this should not be the
>> problem.
>>
>> Best regards
>> Joachim Werner
>>
>> On Die, 07 Okt 2003, Scott Yockel wrote:
>>
>> >I received the following error on a rather large coupled cluster single
>> point
>> >calculation before the the first SCF procedure had even started. What type
>> of
>> >error is this? What does it mean?
>> >
>> >? Error
>> > ? mxbuck too small
>> > ? The problem occurs in outbck(sort1)
>> >
>> > GA ERROR fehler
>> >
>> > GA ERROR
>> >0:0:fehler:: 0
>> >
>> >Also is there a maximum basis function limit for SCF or CC calculations
>> because
>> >I'm at:
>> >
>> >NUMBER OF PRIMITIVE AOS: 815
>> >NUMBER OF SYMMETRY AOS: 621
>> >NUMBER OF CONTRACTIONS: 507 ( 507A )
>> >
>> >
>> >Scott Yockel
>> >University of North Texas
>> >Department of Chemistry
>> >Masters Hall Room 12
>> >
>> >
>> >
>>
>> --
>> Prof. Hans-Joachim Werner
>> Institute for Theoretical Chemistry
>> University of Stuttgart
>> Pfaffenwaldring 55
>> D-70569 Stuttgart, Germany
>> Tel.: (0049) 711 / 685 4400
>> Fax.: (0049) 711 / 685 4442
>> e-mail: werner at theochem.uni-stuttgart.de
>>
>text,CH3Br neutral
>mem,450,m;
>
>geom={C;
> H,1,ch1;
> H,1,ch2,2,hch;
> H,1,ch3,2,hch,3,d1;
> Br,1,cb,2,bch,3,d2;}
>
>ch1=1.08442975 angstrom
>ch2=1.08444276 angstrom
>ch3=1.08449843 angstrom
>cb=1.94391722 angstrom
>bch=107.79092162
>hch=111.10385214
>d1=124.22961872
>d2=-117.89427265
>
>b=[AV5Z]
>
>basis={spdfgh,c,av5z;c;spdfg,h,av5z;c;spdfgh,br,av5z;c;}
>
>int;
>rhf;save,2102.2;escf(1)=energy
>rccsd(t);e2(1)=emp2;eccsdt(1)=energt(1)
>
>table,b
>head,Basis_Set
>table,escf
>head,HF
>table,eccsdt
>head,CCSD(T)
>table,escf-eccsdt
>head,Corr_Energy
>title,CH3Br cc-sets
>
--
Prof. Hans-Joachim Werner
Institute for Theoretical Chemistry
University of Stuttgart
Pfaffenwaldring 55
D-70569 Stuttgart, Germany
Tel.: (0049) 711 / 685 4400
Fax.: (0049) 711 / 685 4442
e-mail: werner at theochem.uni-stuttgart.de
More information about the Molpro-user
mailing list