Running MaxLEV

Run MaxLEV with the command:

maxlev <config.json>

where config.json is a JSON file that defines the optimization problem (see Configuration Format for details). The maxlev command is available after installation with pip install -e . (see Installation Guide).

Command-Line Options

--validate

Validate configuration without running optimization. Useful for checking that all parameters and outputs are valid before a long optimization run.

--verbose, -v

Enable verbose VPLanet output. By default, VPLanet output is suppressed during optimization.

--seed N

Override the random seed specified in the configuration file.

--maxiter N

Override the maximum number of iterations.

--workers N

Override the number of parallel workers:

  • 1: Serial execution (default)

  • -1: Use all available CPUs

  • N: Use N parallel workers

Examples

Validate a configuration file:

maxlev myconfig.json --validate

Run optimization with 4 parallel workers:

maxlev myconfig.json --workers 4

Run with a different random seed:

maxlev myconfig.json --seed 12345

Output Files

After a successful run, MaxLEV generates:

  1. Results file (default: maxlike_results.txt): Contains the best-fit parameters, negative log-likelihood, chi-squared, model predictions with residuals, and parameter bounds.

  2. MaxLEV input files (*_maxlev.in): For each VPLanet input file that contains optimized parameters, a new file is created with the maximum likelihood values substituted. These files are placed in the same directory as the template files.

Example output structure:

examples/<Method>/
    earth.in           # Original template
    earth_maxlev.in    # Generated with ML values
    sun.in
    vpl.in
<method>_results.txt

Timeout Handling

VPLanet simulations can sometimes hang or take extremely long for certain parameter combinations. MaxLEV includes timeout protection:

{
    "vplanet": {
        "timeout": 120
    }
}

If a simulation exceeds the timeout (in seconds), it is terminated and the parameter combination is assigned the failure penalty value.