Very slow ESP generation

jimkress · 2020-03-02 23:08:02

I am trying to use multiwfn to analyze a large QM calculation (see attached file RKS_PBE0-def2-SVP_18Na_no_wat_neut_nopts_plain.zip). Unfortunately, when I try to generate the ESP for this calculation (yes, I did import it as a Molden file), multiwfn tells me to be patient and then proceeds to calculate the ESP at the rate of 1% per day (see attached file slow_ESP.zip).

I am running multiwfn on a Window 7 pro SP1 workstation with 128GB of RAM using 22 cores from the dual Intel XEON CPUs ES-2680 v4 @ 2.40 GHz.

Is there a faster way to generate the ESP? I need it in a cube format.

Thanks for any help.

RKS_PBE0-def2-SVP_18Na_no_wat_neut_nopts_plain.zip

slow_ESP.zip

sobereva · 2020-03-03 02:23:47

The speed of ESP calculation code in current version of Multiwfn is indeed not fast. In fact, I have plan to significantly improve this part of code in very recent months.

Currently, if you have Gaussian, I strongly suggest you use subfunction 2 in main function 100 to convert the loaded .molden file to .fch file, then set "cubegenpath" in settings.ini to actual path of cubegen utility in Gaussian folder, and use the .fch file as input file, in this case Multiwfn will automatically invoke cubegen when ESP needs to be calculated, the cost will be significantly lowered. See Section 5.7 of Multiwfn manual for detail about this point.

If you do not have Gaussian and you only need cube file of ESP, you can try to use ORCA (or one of ORCA utilities, I am not familiar with this point) to generate the cube file.

jimkress · 2020-03-03 02:56:17

Thank you for your reply.

Unfortunately, I have been banned by Gaussian so that is not an option.

In addition, there is no ORCA function or utility that will generate a cube of ESP.

Jim

sobereva · 2020-03-03 05:43:37

Considering that orca_vpot is able to yield ESP, these days I will try to find a workaround based on orca_vpot. When it is finished, I will let you know here.

jimkress · 2020-03-05 17:44:13

Thank you.

sobereva · 2020-03-05 21:22:49

I have written a post to illustrate how to save ESP analysis cost on molecular surface for ORCA users, please check this post: http://sobereva.com/wfnbbs/viewtopic.php?pid=936

jimkress · 2020-03-06 04:59:40

Thanks. I will give it a try.

Multiwfn forum

#1 2020-03-02 23:08:02

Very slow ESP generation

#2 2020-03-03 02:23:47

Re: Very slow ESP generation

#3 2020-03-03 02:56:17

Re: Very slow ESP generation

#4 2020-03-03 05:43:37

Re: Very slow ESP generation

#5 2020-03-05 17:44:13

Re: Very slow ESP generation

#6 2020-03-05 21:22:49

Re: Very slow ESP generation

#7 2020-03-06 04:59:40

Re: Very slow ESP generation

Board footer