Multiwfn forum

koreyr

Member

Registered: 2019-03-12

Posts: 3

can we use gaussian cube files with resp?

Hello,

I'm excited to utilize the new resp functionality within 3.6dev and am wondering if there is a method for using Gaussian cube files either directly importing while in resp submodule (18) or a method to import and convert for use?

I produced  the cube data using the gaussian method cube=(medium, density/potential) and I would like to compare fitting with this method to using

iop(6/41=10,6/42=15,6/50=1) pop=(readradii,mk)

Fe=2.0

charges.esp

and importing to antechamber.
My goal is to switch to multiwfn for charge derivation of metal center complexes (Fe,Cu,etc.) over prior more automated methods.

Thank you!

With best wishes,
Korey

sobereva

Tian Lu (Multiwfn developer)

From: Beijing

Registered: 2017-09-11

Posts: 1,830

Website

Re: can we use gaussian cube files with resp?

Dear Korey,

I don't exactly understand your meaning. The easiest way of calculating RESP charge in Multiwfn is using .fch as input file, as shown in Section 4.7.7. You don't need any special keywords.

If you want to compare if the RESP charge generated by Multiwfn is identical to Antechamber, the answer is clear: Their results are completely identical. During developement of RESP module of Multiwfn, I have made such comparisons with Antechamber for some organic systems.

Multiwfn is unable to calculate RESP charges based on .cub file, because Multiwfn calculate ESP based on analytic wavefunction information, while .cub can only record value of certain real space functions at discrete points.

Best regards,

Tian

koreyr

Member

Registered: 2019-03-12

Posts: 3

Re: can we use gaussian cube files with resp?

Hello Tian,

Thank you for your reply, it helped me understand things a bit better. To better define my aim: I am migrating from the pyred server dev to Multiwfn and I am trying currently to follow the same recipe I employed using their interface for an apples to apples forcefield for my heme containing systems; I currently only have one system left which needs resp charge fitting and I would like to keep the esp point density to ~equal.

When I set the cubegen path and run a resp calculation from a gaussian .fchk file, Multiwfn hands off esp calculation to gaussian (for speed of course) and there is a  bash command being invoked:

cubegen   32 potential=SCF "Hem1.fchk" ESPresult.cub -5 h < cubegenpt.txt > nouseout

Is it possible for me to modify this so that I could change the npts variable to be -6 or -7 instead of -5 (as shown in the copied printout)?
Also, is there a limit to the number of MK-ESP points calculated with gaussian "pop=mk iop(6/41=10,6/42=15)" that Multiwfn can read in (assuming memory and allocation is infinite) does this even matter? * 6/41=Number of layers in esp charge fit. *6/42=Density of points per unit area in esp fit. *

Thank you very much for your guidance!

With best,
Korey

sobereva

Tian Lu (Multiwfn developer)

From: Beijing

Registered: 2017-09-11

Posts: 1,830

Website

Re: can we use gaussian cube files with resp?

Dear Korey

Please note that the ESPresult.cub in fact doesn't contain evenly distributed grid points as usual cube file, but contain the fitting points actually used for the ESP fitting, these points typically occur around van der Waals surface. Therefore, there is no any reason to modify the "-5" to "-6" or "-7", because this modification doesn't make sense.

Currently, in RESP module of Multiwfn, the fitting points can be generated in MK or CHELPG manners. If you want to change the density of the fitting points, you can select "3 Set method and parameters for distributing fitting points", and then choose corresponding options to define the point density.

Alternatively, you can let Multiwfn to load positions and ESP values of fitting points directly from Gaussian output file (e.g. # B3LYP/def2TZVP pop=mk iop(6/41=10,6/42=15,6/33=2)). In this case Multiwfn will not generate fitting points and calculate their ESP values. There is no internal limitation in the Multiwfn code about the number of fitting points.

Best,

Tian

koreyr

Member

Registered: 2019-03-12

Posts: 3

Re: can we use gaussian cube files with resp?

Hey Tian,

Thank you thank you thank you! I am so excited to have found your program. I would not have found it without the due diligence of you helping people on stackexchange! 

Thanks a million!

With best,
Korey