I have more than 10000 files related to calculations done using Vienna Ab-initio Simulation Package. These files are stored in my University's server. I want to take these files in my personal drive before I graduate. I will be compressing these files before taking them in to my personal drive to save storage space.

Which files should I take so that I can revisit them in future for further analysis without re-doing those calculations?

  • $\begingroup$ Purely for analysis, I think the CONTCAR, OUTCAR and vasprun.xml files are the most important for general calculations, though other files might be important depending on the specific target quantities. $\endgroup$
    – CW Tan
    Jun 5, 2023 at 14:09
  • 1
    $\begingroup$ @CWTan It sounds like you can answer the question! We can also add the one-topic-per-answer tag if you'd like to write an answer only about the reason why the CONTCAR (or the OUTCAR or the vasprun.xml) files are important to save, and this gives others opportunities for others to write answers about why they feel the other files are important to save. $\endgroup$ Jun 13, 2023 at 17:49
  • $\begingroup$ @NikeDattani I just wrote a more complete answer, though I hope others can add on in case I missed anything crucial. $\endgroup$
    – CW Tan
    Jun 14, 2023 at 5:12

2 Answers 2


I'll separate the importance of the files to 2 categories.

  1. Purely for data analysis purposes

    • The CONTCAR files are definitely important for relaxed structural information.
    • The OUTCAR and vasprun.xml files would contain the calculation outputs conventionally used for analysis. The OUTCAR would also contain the INCAR settings in human-readable form so you can always refer back to what settings you used from the OUTCAR file (no need to save the INCAR files).
  2. Ease of redoing calculations

    • The CONTCAR files are also important so you don't have to re-relax the structures if you need to redo calculations later on.
    • The INCAR and KPOINTS are the general "settings" type files that would make it much easier to redo calculations and shouldn't be too memory intensive. But maybe you already have a set of scripts and made notes on certain parameter decisions that you could recreate the INCAR and KPOINTS files in line with the original calculations. Of course, if you forget these things, having the INCAR and KPOINTS around could serve as a good record for what calculations you did. The POTCAR file too has a similar role. I wouldn't save it unless you expect to update your set of POTCARs - then it would be important to have a reference of which POTCARs you used at the point when you performed those calculations.
    • I would personally not keep files like the WAVECAR that would be useful for pre-convergence but relatively memory-intensive. But of course, if getting the converged wavefunctions was especially difficult and computationally expensive, maybe it might be a good idea to keep it and spare you from the pain of repeating the full calculation.

Hope someone else can add on if I missed any crucial files.

  • $\begingroup$ Thanks, @CW Tan for the answer. I think WAVECAR and CHGCAR can be safely deleted, as I am only be using the final converged energy values in my future studies. For the sake of repeatability in the future, I just do not want to miss any important files. I think OUTCAR and the XML file should suffice. I shall wait for others to give their input before I mark this question as answered. $\endgroup$
    – Sufyan
    Jun 15, 2023 at 20:14
  • $\begingroup$ @Sufyan Sure, that sounds good for your use case. For keeping the WAVECAR, I was thinking more in the case of of very expensive and difficult to converge GW-BSE calculations for example, where a large number of bands (or all possible ones for a given plane wave grid) and dense kpoint grids are required. Also, if the WAVECAR is the important one, it isn't necessary to keep the CHGCAR because the electron density can be constructed from the wavefunctions. There are also other contexts in which you might want to keep the CHGCAR and not the WAVECAR. $\endgroup$
    – CW Tan
    Jun 16, 2023 at 4:09
  • $\begingroup$ I usually keep CHGCARs around since (at least for SCF calculations) converging the density is relatively expensive compared to solving the Hamiltonian (i.e. a single NSCF step) to get wave functions. This is disk space intensive though less than WAVECARs! $\endgroup$ Apr 8 at 4:24

CHGCAR files of the final run system may be needed, e.g. for CHGCAR visualization, charge density difference analysis, or Bader charge analysis.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .