PC GAMESS/Firefly-related discussion club


Re^4: non-gradient triplet optimization problem

Denis Zavelev
denis.zavelev@gmail.com


Hello, Alex!

>indeed there is the funny bug in the old-style UMP2
>energy code which does not allow parallel execution
>to be performed correctly more than once per job
>execution.
>The workaround is just to use PC GAMESS/Firefly
>specific code, which is not affected by this bug
>and is faster.
>Just add:
>

 
 $mp2 method=1 $end 

>to your input.
Thanks!

>I'd also suggest you to add
>

 $contrl inttyp=hondo $end

>as well, as you use basis with diffuse basis functions.
Exactly! I don't know why did I forget this...

>However, closer examination of your system shows that
>1. The single-determinant wavefunction does not describe this system correctly, and
>2. There is strong spin-contamination already at SCF level

>Taking this into account I'd suggest you to use CASSCF+MCQDPT2
>rather than UHF+UMP2 approach.
I should use SCFTYP=MCSCF for the geometry optimization, right?


>>>I believe this exception may be caused by some errors
>>>in your input file (HINT cordinates are really tricky),
>>>as you specified ten parameters (NPAR=10) to optimize,
>>>while the number of internal coordinates is only 9 (3*5-6=9)
>>If so, why this computation breaks after ~2 minutes while the same computation on single CPU PC still goes from yesterday's evening?
BTW, that computation also crashed (after about a day)...
So it seems that I really put a mistake in $DATA section.

But I can't understand these HINT coordinates (a really stupid beginner, oh yeah!).
I found that article and read it thoroughly but it's real hard to find the difference between NPCC and CCPA type connections (as they are shown strictly the same!), furthermore, I can't understand what is "beta" angle in CCPA case. HINT coordinates have some common with well-known spherical coordinates, but seem to be the perverted version of the last ones. I couldn't find anything in the web that can help me to understand...

In this case it seems we have the following (using C1 symmetry just in case, though Cs will not solve the problem):
we are putting H atom in the centre and draw I,J,K (and 3 zeros for H atom, LC, "-").
Then we place O2 onto x axis, so we have LC "-" connection with H.
Placing O3 onto xz plane, we have PCC, "-", with bond length and angle.
But how to place the rest 2 oxygen atoms?
I suppose I should use CCPA, but what is beta?
Just in case: cartesian coordinates go below.
HYDROGEN    1.0      0.000000000         0.000000000         0.000000000
OXYGEN      8.0      1.223961499         0.000000000         0.000000000
OXYGEN      8.0      0.521949141         0.000000000        -1.107091165
OXYGEN      8.0     -0.875913455         0.655452553         0.548852733
OXYGEN      8.0     -0.875913455        -0.655452553         0.548852733


[ Previous ] [ Next ] [ Index ]           Tue Nov 11 '08 3:36pm
[ Reply ] [ Edit ] [ Delete ]           This message read 1168 times