Solntsev Pasha
solntsev@univ.kiev.ua
Pavel.
On Fri Dec 28 '12 10:03pm, Siddheshwar Chopra wrote
---------------------------------------------------
>Dear sir,
>I just ran exetyp=check, before running hessian calcs. And i got the message.....scf did not converge....perform more gradient runs.... so is this sufficient to conclude that global minimum is still far??? I mean do i save myself from running hessian calcs. to confirm it?
>n Fri Dec 28 '12 1:44am, Alex Granovsky wrote
>----------------------------------------------
>>Dear Siddheshwar,
>>strictly speaking, this statement means that a stationary point was
>>located (i.e. the point where gradient is zero). This statement does
>>not mean that the stationary point which was located is a minimum.
>>To check this, you need to perform Hessian computations as suggested
>>by Luca.
>>Kind regards,
>>Alex Granovsky
>>
>>
>>On Thu Dec 27 '12 8:49pm, Luca Maidich wrote
>>--------------------------------------------
>>>Dear Siddheshwar,
>>>the only way I know for being sure that you correctly found a stationary point is to run a frequency analysis job. It is a tedious job but it has to be done.
>>>Two considerations:
>>>a) if you loose too much OPTTOL value your optimization could lead to non-meaningful results;
>>>b) are you sure that what you are doing now is NOT more tedious than running optimization+frequency jobs? :)
>>>Kind Regards
>>>Luca