NEB Compiling and Calculation Issue!

[Trivanni Yadav]

Hello,

This is my VERY first time attempting a NEB calculation, and I must admit, I'm feeling quite confused. I would sincerely appreciate any feedback or suggestions. Here's the issue:

Following the instructions provided, I have two initial and final optimized geometries, represented by POSCAR1 and POSCAR2. Then, I used the nebmake.pl script to generate directories corresponding to the number of images. In my case, I used 3 images, resulting in the creation of 4 subdirectories numbered 00, 01, 02, 03, and 04. Each subdirectory contains the respective POSCAR file.

Currently, my parent directory looks like this:
directory_00
directory_01
directory_02
directory_03
directory_04
INCAR
KPOINTS
POSCAR1
POSCAR2
POTCAR
nebmake.pl

I also placed OUTCAR files in the initial and final structures' subdirectories, i.e., directory_00 and directory_04.

Now, ideally, I should proceed with running my NEB calculation, correct? However, after submitting the job on VASP, I kept encountering an error message stating "POSCAR: No such file or directory."

Then, I attempted to run VASP individually for each subdirectory. For instance, let's consider directory_01. For this step, I have all four input files in this directory. However, after running the VASP calculation, I encountered another error: "forrtl: No such file or directory."

Additionally, while I was going through the instructions for installing VTST codes, I noticed specific guidelines for compiling VTST code into VASP. Currently, I'm using VASP version 6.4.2. The instructions involved downloading the vtstcode-199. They included the following steps: "To build the code, the VASP .objects and makefile need to be changed. Find the variable SOURCE in the .objects file (a hidden file in src/), which defines which objects will be built, and add the following objects before chain.o

I cannot locate the .objects file anywhere!

This doesn't make sense to me at all. Have I misunderstood the process, or is there something else going wrong?

[graeme]

There are a few things that seem wrong. First, you need to build vasp with the vtstcode guided by https://theory.cm.utexas.edu/vtsttools/ ... ation.html ; or get help from a student in my group. The .objects file is in the vasp.6.4.0/src directory, but again, if you are new to this process, you can ask for help from someone in my group.

Let's solve this first before dealing with the NEB calculation - you can't run it separately in each directory; it has to be run as a single calculation from the main directory. It will spawn the specified number of images and relax them to the minimum energy path.

(and the subdirectories should be automatically created as 00 up to 0N; without any directory_ prefix. To help with subsequent problems, I usually ask that people just tar/gzip their directory so that I can see exactly what is going on to help debug it.

[Trivanni Yadav]

Thank you so much, Dr. Henkleman. I have contacted our HPC administrator and shared with him the instructions to compile vtsttools with VASP. However, I just want to make sure that before running the calculation, in my parent directory, I would have:

00
01
02
03
04
INCAR
POTCAR
KPOINTS
POSCAR1
POSCAR2
nebmake.pl

And in the directories 00 and 04, I must also have their respective OUTCAR files as well. That's the required format, right?

[Trivanni Yadav]

Hello Dr. Henkelman,

I wanted to update you on the progress regarding the compilation of VTST with VASP 6.4.2. I've received confirmation from my HPC administrator that VTST has been successfully compiled with VASP 6.4.2.

However, upon attempting to run one of your NEB calculation examples for practice, specifically the one for vacancy diffusion on Au(111) under the Force-based optimizer (I've attached the relevant files for your reference), I encountered several errors. The zipped folder contains the vasp.err and vasp.out files detailing these errors.

Strangely, when I ran a simple calculation without utilizing any NEB-specific tags to verify the compilation status of VTST, the job executed without issues. Upon inspecting the OUTCAR file, I noticed the following information:

```
VTST: version 4.1, (05/19/21)
CHAIN: initializing optimizer
OPT: Using VASP Dynamics algorithm
CHAIN: Read ICHAIN 0
```

Based on this output, it appears that VTST has been compiled correctly, right?. However, despite this, I'm encountering difficulties with the NEB calculation.

I apologize if I'm overlooking something obvious, as I'm relatively new to this and haven't performed NEB calculations before.

Any guidance or insight you could provide would be greatly appreciated.

[graeme]

I would suggesting starting simple and getting things working before extras (such as using a ML optimizer). I simplified your files in the attached, and it started fine.

[Trivanni Yadav]

I just used these files and simply ran them, but I got the exact same error. :(
Does this mean that the compilation is not done completely, or what? I can copy my HPC administrator here, since he is willing to make sure that all necessary configurations have been done to ensure the compatibility of all NEB tags with VTST compiled with VASP 6.4.2.

[graeme]

I'm happy to talk with anyone to get the vtstcode in vasp working. It's hard for me to debug problems that are related to a system that I don't have access to, in contrast to problems related just to input and output files. It was easier on older systems, but now with many possible combinations of modules, libraries, paths and ld_library_paths and so forth, it's hard to figure out what is going wrong in a specific user's system environment without being able to run tests.

[Trivanni Yadav]

Thank you so much, Dr. Henkleman. I understand what you are referring to, but I feel that if you and Dr. Wheat, my HPC administrator, can connect, we can work through the setup together. I'll send you an email and will copy Dr. Wheat on it. I really want to get this resolved, as I find your NEB method to be the easiest to understand.