Dear Tian Lu,

I already asked a question about generating a promolucle .wfn with atomic files defined in atomwfn folder. However, I faced another difficulty, and maybe creating a new topic for a separate question is better.

The atomic wfn files you are so helpfully providing with MultiWfn were generated using the 6-31G* basis. However, it is not defined for the elements below the 4th row (some of which are present in my systems as well). I have recalculated all atomic wfns for stable nuclei in orca. However substituting them into atomwfn folder makes MultiWFN try to generate the "missing" wfns in Gaussian (which i do not have on my laptop). This makes me think that there is something about my wfns that MultiWFN is unhappy about. I did a visual comparison, aside from the header "Generated by Multiwfn" and "GAUSSIAN" in the next line, the file structure is very similar. One other difference is that the virial ratio in my orca wfns is 0.0 (although the corresponding output files have the value close to 2, so I assume it is a bug in orca). In connection to this, two questions:

1. Do you have an idea of what could be the problem?
2. Regarding sphericalization. There is no specific option for that in MultiWFN as far as I can see. Is it then done on the fly with specific instructions in the config? How is MultiWFN with sphericalization of external files? I read somewhere here that I does not do it (or was not for that time). Are you aware of any alternatives?

Thank you and best regards,
Serhii

PS. Just in case I will send my wfns to you via email (sorry if too spammy).

Ah, thank you, Tian Lu! Now i get it why it would crash.

As illustrated in the example in Multiwfn manual, you should load .fch/fchk file when Multiwfn boots up, and load .out/log file only after entering subfunction 1 (hole-electron analysis) of main function 18.

Many thanks!

Thank you very much for your helpful and detailed response. I truly appreciate your guidance and the references you shared. I plan to follow your advice and will proceed with the IRI analysis in Multiwfn to investigate the presence of the intramolecular hydrogen bond and assess the zwitterionic nature of the molecule.

The thing you are talking about is not "to combine alpha and beta occupations and exponents". It is probably related to two different kinds of natural orbitals: 1) spatial natural orbitals; 2) natural spin orbitals.

If you perform a ground state triplet CASSCF calculation, OpenMolcas will generate two .molden files, one is xxx.rasscf.molden and the other one is xxx.rasscf.molden.1. The spatial natural orbitals are stored in the former file, and the natural spin orbitals are stored in the latter file (which corresponds to what you called "always split into alpha and beta"). So, what you need is in xxx.rasscf.molden, and you can run

molden2fch xxx.rasscf.molden -molcas

to generate file the xxx.rasscf.fch, which can be opened and visualized by GaussView/Multiwfn. Here `molden2fch` is a utility in the open source package MOKIT.

But if you perform a triplet CASSCF calculation with multiple triple states, xxx.rasscf.molden contains the CASSCF average pseudo-natural orbitals which are often useless since they are not NOs of any electronic state. In this case, one needs to add densities in xxx.rasscf.molden.N to obtain the total density for state N, then generate NOs for this state.

Dear Sayan,

I am not sure if your TrESP charges have been correctly generated. Just taking the example in Section 4.A.9 of Multiwfn manual, S0-S0 transition of 4-nitroaniline. By following the step in the manual (I'm using Gaussian 16), you should have .chg file with following content:

C -0.003279 0.022698 1.208952 0.2134106035 C -0.003279 -1.356850 1.208468 -0.1679471876 C -0.002220 -2.074407 0.000000 0.1841565861 C -0.003279 -1.356850 -1.208468 -0.1681325143 C -0.003279 0.022698 -1.208952 0.2136075560 C -0.002956 0.708053 0.000000 -0.1725671768 H -0.002297 0.584212 2.134941 -0.0164467259 H -0.008330 -1.894941 2.152080 0.0062724543 H -0.008330 -1.894941 -2.152080 0.0063059790 H -0.002297 0.584212 -2.134941 -0.0165045142 N -0.047761 -3.443794 0.000000 0.1716754368 H 0.214094 -3.921723 0.844971 0.0002801698 H 0.214094 -3.921723 -0.844971 0.0003282958 N 0.000588 2.162345 0.000000 -0.0646299850 O 0.002065 2.727687 -1.077690 -0.0949102566 O 0.002065 2.727687 1.077690 -0.0948987209

If you load this file into Multiwfn, you will see

 Component of electric dipole moment: X= -0.016488 a.u. ( -0.041908 Debye ) Y= -2.518398 a.u. ( -6.401129 Debye ) Z= 0.000017 a.u. ( 0.000043 Debye ) Total electric dipole moment: 2.518452 a.u. ( 6.401267 Debye )

As mentioned in Multiwfn manual, the TrESP calculated based on Gaussian .wfn file should be divided by sqrt(2), and thus the transition dipole moment should also be fixed in this way. So, for example, Y component of transition dipole moment of S0-S2 should be -2.518398/sqrt(2)=-1.78077 a.u.

Now, look at Gaussian output file (4-Nitroaniline_IOp.out in the TrEsp.zip), you can find

 Ground to excited state transition electric dipole moments (Au): state X Y Z Dip. S. Osc. 1 -0.0000 -0.0000 0.0001 0.0000 0.0000 2 -0.0165 -1.7911 0.0000 3.2083 0.3408 3 0.0210 0.0188 0.0000 0.0008 0.0001

Clearly, the corresponding data is -1.7911, which is very close to the -1.78077 calculated based on TrESP. So, there should be no sign issue.

As you are an ORCA user, I would like to note that if you are using TDDFT, currently Multiwfn is unable to accurately calculate TrESP, since ORCA doesn't print excitation and de-excitation configuration coefficients separately. You have to use TDA-DFT instead. (PS: The situation will be changed in the next release ORCA).

Best,

Tian

Hello,

This decomposition currently is not feasible for ORCA.

However, there are alternative approaches to realize this aim. Mayer bond order from sigma LMO and pi LMO are usually (at least approximately) additive, so you can gain the pi component of Mayer bond order (i.e. pi-bond order) based on LMOs, see Section 4.100.22 of Multiwfn manual on how to do. Also please check Theoretical Chemistry Accounts, 139, 25 (2020) https://doi.org/10.1007/s00214-019-2541-z to gain more information about this idea.

Hello, I don't have experience in this feature of ORCA, I always calculate enthalpy of formation of solids using CP2K, which is easiest for this purpose. I think setting up a proper cluster and calculate this quantity using ORCA should be evidently more cumbersome and challenging.

Because pseudopotential is used for Pt in the HF-3c method, you need to properly modify the [Atoms] section of the molden file produced by orca_2mkl, so that Multiwfn can recognize the actual effective nuclear charge of Pt. Please check "Molden input file" paragraph in Section 2.5 of Multiwfn manual, and //www.umsyar.com/wfnbbs/viewtopic.php?pid=721

Alternatively, you can simply use ORCA 6.0 or later, then you do not need to manually modify the molden file.

Rookie mistake: I set radpot and sphpot in settings.ini but Multiwfn kept using defaults. Turns out iautointgrid= 1 was silently overriding them. Switched it to 0, problem solved.

There is no existing function for this purpose. Manually calculating it is easy: Using Multiwfn to simulate spectrum, export the two curves as text files,  import them into e.g. Origin, define a new column as their overlap, then perform numerical integration.

I prefer to display all surface extrema, but only manually label some of them, namely the relative important ones (i.e. the ones with relative large magnitude of ESP)

1 sobEDA.sh script only supports Gaussian. Perhaps there are some ways to realize sobEDA in the framework of ORCA, but there is no script available currently.

2 "sum of pair energies" fully corresponds to orbital interaction energy, while Coulomb, exchange, dispersion effects are irrelevant to it.

Dear Prof Tian Lu,

Thank you so much for your reply, it helps me a lot !

My best

Alessio