Customization

I want to optimize my own function

See Tutorial: Adding a Custom Solver for a tutorial on defining your own objective function and minimizing it with ODAT-SE. It provides a complete, copy-paste-ready example.

I want to use an external program as a solver

You can call external programs using subprocess inside the evaluate method.

import subprocess

import odatse.solver

class MySolver(odatse.solver.SolverBase):
    def evaluate(self, x, args=()):
        # Write parameters to file
        with open("params.dat", "w") as f:
            for xi in x:
                f.write(f"{xi}\n")

        # Run external program
        subprocess.run(["./my_program", "params.dat"], check=True)

        # Read result
        with open("result.dat") as f:
            fx = float(f.read().strip())

        return fx

The evaluate method is called with proc_dir (per-process working directory) as the current directory, so there are no file conflicts during MPI parallel execution.

For developing a full-featured package, consider using the solver templates available in ODAT-SE Gallery (see Related Resources).

I want to see real analysis examples

ODAT-SE Gallery provides working examples for TRHEPD, SXRD, LEED, and XAFS analysis. Each example includes input files, execution scripts, and visualization scripts that can be run as-is. See Related Resources for details.