Setting up your scripts

There are in general two types of scripts in a scientific paper: scripts for creating the figures and tables in the paper, and other scripts. You should strive to make a clear distinction here, especially if your analysis scripts take a long time to run.

Scripts for reproducing figures and tables

These scripts should preferably work with input data from file. Common formats for raw data are h5 (preferable for applications executed in parallel with MPI), csv if you want to use pandas.

Scripts for running the analysis

These scripts should have a clearly stated interface for input parameters, using for instance argparse or click. It is preferable to have default parameters for these scripts.

If you want to run such a script for a wide range of parameters, you could use argparse in the following way

from typing import Sequence, Optional
import pprint
import argparse


def analysis(argv: Optional[Sequence[str]]) -> int:
    parser = argparse.ArgumentParser(
        formatter_class=argparse.ArgumentDefaultsHelpFormatter)
    parser.add_argument(
        "-N", type=int, help="Optional input integer", default=0)
    parser.add_argument("-T", type=float, help="Input float")

    parser.add_argument("--run_code", dest="code", action="store_true",
                        help="Optional flag", default=False)
    args = vars(parser.parse_args(arguments))

    print(f"Arguments gotten: {pprint.pformat(args)}")
    return 0


if __name__ == "__main__":
    N = [1, 2, 5, 8]
    for run in [True, False]:
        for n in N:
            arguments = ["-N", f"{n}", "-T", "3.5"]
            if run:
                arguments.append("--run_code")
            print("-"*30)
            print(f"Input arguments {arguments}")
            analysis(arguments)