scientific ensemble simulation parameter space exploration

This small utility generates parameter-space files for ensemble simulations. It writes two files for each parameter specification:

• .txt — the raw parameter combinations (space-separated floats, one combination per line)
• _norm.txt — the same combinations but each parameter axis normalized to [-1, 1]

The generator performs a Cartesian product across parameter axes defined as [min, max, count]. Values along each axis are evenly spaced using numpy.linspace.

• app.py — main generator. Call generate_params(param_specs, prefix) or run the script directly to generate castro.txt and castro_norm.txt (example included in the script).
• castro.txt, castro_norm.txt, clover.txt, clover_norm.txt, gadget.txt, gadget_norm.txt — example outputs (committed here for reference).

Requirements:

• Python 3.7+ (tested with 3.8+)
• numpy

Install dependencies (recommended in a virtualenv):

python3 -m pip install --user numpy

Run the script directly to generate the included example:

python3 app.py

Or import and call from Python:

from app import generate_params

param_specs = [
	[0.80, 0.95, 20],  # e.g. MP
	[0.80, 0.95, 20],  # e.g. MS
]
generate_params(param_specs, prefix="castro")

Each output file is plain text. Each line is a single parameter combination with values separated by spaces. Numeric values are formatted to 6 decimal places.

Example line from castro.txt:

0.800000 0.800000

Example corresponding normalized line from castro_norm.txt:

-1.000000 -1.000000

• The normalisation maps each axis independently to [-1, 1]. If any axis has identical min and max (zero range), normalization will produce a division-by-zero — avoid degenerate axis ranges.
• The number of combinations equals the product of the count values. Be careful with large counts which can create very large files and high memory usage during Cartesian product construction.

• Stream the Cartesian product to file instead of building full lists to reduce memory usage for large grids.
• Add CLI argument parsing (argparse) to accept parameter specifications from command line or configuration files.
• Support other normalization schemes (z-score, min-max without [-1,1], log-scales).

MIT