Multiwfn forum

Multiwfn official website: http://sobereva.com/multiwfn. Multiwfn forum in Chinese: http://bbs.keinsci.com/wfn

You are not logged in.

#1 Re: Multiwfn and wavefunction analysis » Automatic selection of ELF basins » 2024-11-02 19:16:24

Dear Prof. Lu,

Thank you for the tips. I will try them out and share what happened so that other users may benefit from this knowledge.

Best,

Marcos

#2 Multiwfn and wavefunction analysis » Automatic selection of ELF basins » 2024-11-02 02:26:49

mverissi
Replies: 2

Dear Prof. Lu,

I intend to use Multiwfn to plot ELF functions to try to understand how catalysis takes place in the active site of a DNA polymerase. I have performed QM/MM calculations with ORCA and I have successfully plotted ELF isosurfaces in VMD.

I will plot the ELF functions for each image in an NEB calculation with many images (about 60!) and hopefully attempt to detect changes in the basins along the reaction coordinate determined by the NEB calculation.

Looking in the manual, I saw how one can plot monosynaptic and disynaptic basins in different colors. It might be a good idea to do this, since perhaps I could detect changes in these basins visually.

I have 197 atoms in my quantum region, and I performed a calculation with the coarsest grid, just to see the sequence of options that I would input in a submission script to automate the execution. However, I was a bit disappointed when I realized I'd have to input the numbers of the basins, and that there's no option to select all basins for output.

With the current workflow, I suppose that, to automate the calculation, I would have to perform the basin calculation redirecting the MWFN output to a file. Then I'd have to interrupt it to get the indices of the N-synaptic basins from the MWFN run, and input them in a second run. That would, of course, be a waste of computer time. Is there any other way to tell MWFN to save a cube file with all the basins?

If there's no way to do this, although my Fortran is a bit rusty, I suppose I could try to look into the code and attempt to adapt it if necessary. Could you kindly point me to which files I should look into?

Best regards,

Marcos Verissimo Alves

#3 Re: Multiwfn and wavefunction analysis » Bug in conversion from molden input format to .wfx? » 2023-12-08 08:27:43

Dear Prof. Lu,

Thanks for the kind answer. Took me a long time to thank you because I never again touched the project in which I was having this trouble and I also forgot to check for the answer :-)

Best,

Marcos

#4 Multiwfn and wavefunction analysis » Bug in conversion from molden input format to .wfx? » 2023-03-24 14:55:06

mverissi
Replies: 2

Dear all,

I am converting Orca (5.0.3) gbw files from unrestricted DFT calculations in molecules to Gaussian wfx, as an input to calculations of partial charges using the DDEC6 method. Orca produces wfx files, but then the DDEC6 code complains about accuracy issues. Therefore, I converted the ORCA gbw file to the molden input format using orca_2mkl, and subsequently I converted the resulting molden input file to Gaussian wfx using Multiwfn. I used both my own compiled version for Mac, about which I have posted earlier in the forum, and the Multiwfn 3.7 pre-compiled binary for Linux.

The conversion works fine, except for one section of the .wfx file, the one containing the charge of the molecule:

<Net Charge>
******
</Net Charge>

for both my own compiled Mac version and version 3.7's pre-compiled binary for Linux. When I changed the asterisks to 1 (the net charge of the molecule), the DDEC6 calculation ran fine.

Now, I checked the molden input file, and I could not find any entries regarding the total charge of the molecule (I grepped for charge, case-insensitive, and got nothing). I am not totally familiar with the molden format, so I don't know if it would have a field specifying the total charge of the molecule. Could it be a bug in Multiwfn, or would this perhaps be a bug from orca_2mkl?

Best,

Marcos

#5 Multiwfn and wavefunction analysis » Compilation of Multiwfn for Mac with Apple Silicon chips » 2023-03-24 12:34:13

mverissi
Replies: 1

Hi all,

This isn't really a question, but something to help people who are using Macs with Apple Silicon chips. I have been using one lately and I wanted to perform some test calculations outside the cluster where I will run heavier stuff. Since the GUI depends on Motif and I couldn't find either Motif or the executable without the GUI for Mac, I set out to compile it myself. So I thought I could contribute with a very brief guide  that worked for compiling the no GUI version of Multiwfn on my MacBook Air with M2 chip.

1) Download the latest Intel compiler for Mac at https://www.intel.com/content/www/us/en … ml#fortran - without it, you'll need to “get your hands dirty” with changing some of the code. My Fortran skills are only good enough to write simple code, so after having to change 10 lines of code and still getting errors from gfortran, I decided that downloading and installing ifort was a wiser choice.

I downloaded the online installer mistakenly (26 KB; there's a link for downloading the offline installer, 96 MB), but it worked fine for me. Double click on the dmg and drag the installer file to the applications folder, then start it. After the installation, add the line

source /opt/intel/oneapi/setvars.sh

to .zshrc and source the .zshrc file.

2) Unzip the Multiwfn_3.7_src_Mac.zip file and enter the directory. Before anything, edit the Makefile_noGUI file. In the “OPT = (...)" line, change the -openmp and -openmp-link=static options to -qopenmp and -qopenmp-link=static, which are the corresponding ifort options in the latest versions. For a first compilation, keep the libreta_slow module. On my MacBook Air, the compilation with libreta_fast took over two hours to complete using 8 cores.

3) For the no GUI compilation, there's a bash script called “noGUI.sh”. Set it to executable with

chmod +x noGUI.sh

and execute it. It will create a noGUI directory; enter it and type make -j8. With the libreta_slow module, the compilation took less than 10 minutes to complete.

4) Edit your .zshrc file and add the line

export KMP_STACKSIZE=64000000

Source your .zshrc with

source ˜/.zshrc

5) Section 2.1.3 of the manual says we should set

kern.sysv.shmmax = 512000000

in the /etc/sysctl.conf file. I couldn't find it, so I created one (you'll need sudo for this) and added the line above, but without spaces:

kern.sysv.shmmax=512000000

This is because when I tried setting it through the command line (sudo sysctl kern.sysv.shmmax = 512000000), it didn't work. When I issued the command without the spaces (sudo sysctl kern.sysv.shmmax=512000000), it worked fine. However, setting it through the command line will only work for your current session and will require sudo every time; if you reboot the system, this setting will be lost.

6) The file settings.ini is not included in Multiwfn_3.7_src_Mac.zip . The workaround is to download Multiwfn_3.7_bin_Mac.zip, unzip it and copy the settings.ini file to the Multiwfn_3.7_src_Mac/noGUI directory.

7) When you compile the no GUI version, no libdislin_d.11.dylib is created. At that point, I decided to test if Multiwfn had compiled correctly simply by executing it. It appears to be fine, but I didn't run any real tests; the executable just starts without errors.

8) Setting the Multiwfnpath environment variable can be done as explained in section 2.1.2 of the manual.

I hope this will help people who work on Macs to compile Multiwfn and use it. I would very much rather use a Linux system instead but, while my Linux computer isn't repaired, I'll have to work on the Mac.

If Prof Lu finds this worth of inclusion in the Multiwfn manual, I can update this after having performed a few tests. So far, I have converted a file in molden format to Gaussian wfx without crashes.

Best regards,

Marcos

Universidade Federal Fluminense, Volta Redonda, Brazil / Lund University, Lund, Sweden (until September 2024)

#6 Re: Multiwfn and wavefunction analysis » AIM analysis from PDB electron density maps » 2022-02-19 15:10:48

Hi Tian,

Once again, thanks for the reply. My intent is to see if AIM and/or NCI analysis, applied to experimental electronic densities, would give any information about the protonation states of charged residues such as histidine in active sites.

It's good to know that Multiwfn does AIM on grid-based densities. I guess I'll give it a try and see what happens.

Best,

Marcos

#7 Re: Multiwfn and wavefunction analysis » AIM analysis from PDB electron density maps » 2022-02-19 05:08:53

Dear Prof. Lu,

Thanks for the reply. I have indeed been performing exploratory DFT and xTB calculations for active sites of proteins with MultiWFN over the last few months, and now I’m about to start DLPNO-CCSD(T) calculations with ORCA, aiming to perform AIM analysis with them. It’s a pity that MultiWFN doesn’t support AIM analysis for experimental charge densities. Is there a fundamental limitation, or is it just not implemented for reasons, say, of priorities?

Would you know of any code that would perform AIM analysis on experimental charge densities? It would be fundamental for the project I intend to undertake with those calculations. I would still use MultiWFN for the theoretical calculations, since it’s a great code and I found it quite easy to work with.

Best,

Marcos

#8 Multiwfn and wavefunction analysis » AIM analysis from PDB electron density maps » 2022-02-19 01:29:20

mverissi
Replies: 4

Dear all,

I am interested in comparing AIM results from wave function / DFT calculations with those performed directly on electron density maps from PDB structures. Is it possible to do it using MultiWFN? I suppose MultiWFN cannot read PDB files and calculate the corresponding electron density maps, so does anyone know of a code that could read the maps and save the electron density in a format that MultiWFN could read?

Best regards,

Marcos Verissimo Alves
Physics Department, ICEx
Universidade Federal Fluminense
Volta Redonda, Brazil

Board footer

Powered by FluxBB