Learn how to ask questions correctly

Alex Granovsky

gran@classic.chem.msu.su

Dear Thom,

many thanks!

Kind regards,

Alex

On Fri Jun 28 '13 10:35am, Thomas wrote

---------------------------------------

>Dear Alex,

>Attached is the input file. Using FF 7.1.G, I got the same result as Xaiza (though the contribution of each term varied slightly -- the biggest variation is in the (D,4) term). Specifying COORD=UNIQUE seemed to solve the problem. Note that I did not test with FF 8.0.0.

>

>

>Kind regards,

>Thom

>

>

>

>

>On Fri Jun 28 '13 8:20am, Alex Granovsky wrote

>----------------------------------------------

>>Dear friends,

>>I have run MP2 and MP4 computations on this system using settings

>>which I believe are suitable in the case of strong linear dependence

>>in the used atomic basis set. Please find attached my input and

>>output files. The results I obtained are consistent and I do not

>>see any difference between coord=cart or coord=unique runs. This

>>is normal, this type of job isn't the one where one can expect

>>errors due to use of coord=cart except for expected differences

>>in rounding and truncation error accumulations.

>>Could you please provide me an input file where the use of

>>coord=cart results in sinnificant differences as compared with

>>coord=unique case? I'd like to look into this problem a bit deeper.

>>Most likely, the discrepancy is caused by a different accumulation

>>of errors in two different geometries but I want to double check.

>>At least the differences I see in my calculations all can be

>>explained this way.

>>The sensitivity of MP4(D) term of the MP4 energy to the linear

>>dependence in the AO basis set has been already discussed on this

>>forum several times. The MP4(D) term is the most sensitive one

>>because during its computation one needs to transform integrals and

>>other quantities from AO to MO basis, then back from MO to AO,

>>and finally from AO to MO again. Each transformation kills up to

>>several significant digits in the result (more precisely,

>>the number of digits lost is directly related to the condition

>>number of the overlap matrix) if all orbitals (including virtual

>>ones) are not computed very precisely. So in the worst case one

>>can lost all significant digits...

>>Kind regards,

>>Alex Granovsky

>>

>>

>>

>>

>>

>>On Thu Jun 27 '13 12:09pm, Thomas wrote

>>---------------------------------------

>>>Dear Xaiza,

>>>One additional comment: if you're going to rerun all of your calculations as Jim suggested, you should use one additional option:

>>>$CONTRL D5=.T. $END

>>>This selects the use of spherical functions, which is required for the 6-311G basis. Other sets included in Firefly (such as 6-31G and TZV) should not use this option.

>>>

>>>

>>>Kind regards,

>>>Thom

>>>

>>>

>>>

>>>On Wed Jun 26 '13 5:45pm, Xaiza wrote

>>>-------------------------------------

>>>>Dear Sir Thomas,

>>>>I am very thankful to your insights, suggestions and for the effort of re-running my calculations. I will apply your suggestions in my recalculations. But I have a question sir. Would it be fine if some of my MP4 single point calculations have that keywords and some don't? That is, I'll only apply the keywords you have suggested on the molecules in which I have some problem with. I am bit worried about the energy differences that I would be obtaining, most especially since in some of my reactions, only in one molecule that I encountered this MP4 problem.

>>>>Regards,

>>>>Xaiza

Fri Jun 28 '13 11:04am

This message read