From: Mayne, Christopher G (cmayne2_at_illinois.edu)
Date: Thu May 18 2017 - 21:46:03 CDT

Glad to hear it worked out. I will make a note to add a check for irregularly terminated gaussian files.

Regards,
Christopher Mayne

> On May 18, 2017, at 9:30 PM, Siladitya Padhi <siladitya.padhi_at_research.iiit.ac.in> wrote:
>
> Dear Christopher,
>
> Thank you very, very much indeed for your suggestions. I followed all three of your suggestions, and the charge optimisation now runs to completion in a few seconds.
>
> Indeed, there was a QM calculation that had terminated with an error. About the charges on non-polar hydrogens: I made the charge assignment in the input PSF and omitted these hydrogens from the optimisation.
>
> Best regards,
> Siladitya
>
> ----- Original Message -----
> From: "Mayne, Christopher G" <cmayne2_at_illinois.edu>
> To: "siladitya" <siladitya.padhi_at_research.iiit.ac.in>
> Cc: "vmd-l" <vmd-l_at_ks.uiuc.edu>
> Sent: Wednesday, May 17, 2017 12:29:52 AM
> Subject: Re: vmd-l: FFTK charge optimisation does not converge
>
> Siladitya,
>
> Indeed the charge optimization should be much, much faster than that. Typically, charge optimization within ffTK only takes a couple of seconds, but this time is dependent on the termination criteria.
>
> Based on the information you have provided, there’s no way to determine what the problem is. However, I do have a couple of thoughts to try:
>
> 1) make sure that all of the water interaction QM calculations have terminated normally by checking the end of the Gaussian log file.
>
> 2) visually inspect the optimized position of the QM-optimized water interaction to make sure that a reasonable interaction distance and geometry was achieved. exclude any instances where the waters do not find a local minimum, i.e., either fly off into the distance, or settle into areas that are greater than a few angstroms from the interaction site.
>
> 3) all non-polar hydrogens should be removed from the optimization. as stated in our paper and in all tutorials, non-polar hydrogens should be fixed to +0.09 by convention. Make sure that you are making these fixed charge assignments in the input PSF, or by using the advanced options using the Override ReChargeFromPSF; although the former is the preferred method. After you have removed the non-polar hydrogens from the Charge Constrains box, make sure that you recalculate the net charge values using the “Calculate from PSF” button.
>
> Hopefully something in here will correct whatever issue you are running into.
>
> Regards,
> Christopher Mayne
>
>
>
>> On May 14, 2017, at 3:55 PM, Siladitya Padhi <siladitya.padhi_at_research.iiit.ac.in> wrote:
>>
>> Dear All,
>>
>> I am employing the FFTK module to parameterise a molecule that includes a homocyclic ring and a heterocyclic ring, and is 21 atoms big. The quantum mechanical calculations for characterising water interaction sites run fine. However, when I read in the log files from these calculations on FFTK and try to optimise the charges, the optimisation keeps on running without any hint of convergence (I let it run for 14-15 hours, and then eventually terminate it, because I believe the optimisation should take much less time than that).
>>
>> Here are the initial charges and upper/lower bounds that I am using.
>> Carbons: initial charge of 0.0, lower bound of -0.5, upper bound of +0.5
>> Nitrogens: initial charge of 0.0, lower bound of -1.0, upper bound of 0.0
>> Hydrogens: initial charge of +0.09, lower bound of +0.09, upper bound of +0.09 (all hydrogens being non-polar)
>>
>> I tried changing the tolerance from 0.005 to 0.01, but that didn't help. I also tried omitting the hydrogens from the optimisation, but that resulted in the error "domain error: argument not in valid range". Could anyone suggest modifications that would lead to proper completion of the optimisation?
>>
>> Many thanks indeed in advance.
>>
>> Best regards,
>> Siladitya Padhi
>>