This repository contains the development of a framework to support the creation and research of new Monte Carlo simulation algorithms. It also will contain these algorithms and their development.
A Python script is included to run batch jobs. A configuration file using JSON is used to specify parameters for the simulations
| Option | Help | Default | Type |
|---|---|---|---|
| beta | Beta, inverse temperature | None | Number |
| width | Simulation is square array, this is the width to use | None | Integer |
| percent_RECA | This is tuned to test RECA and Metropolis mixes | None | Number, 0 - 100 |
| n_states | Number of States to use | None | Integer, greater than or equal to 2 |
| seed | Seed for Random Number Generator | Random Integer 0 - 2^20 | Integer |
| cpu_seconds | Number of CPU seconds to evolve the system for | 30 | Number |
| samples | Number of times to simulate each set of parameters | None | Integer |
| root | The root level for data file storage | None | String |
usage: run_job.py [-h] -f JSON
arguments:
-h, --help show this help message and exit
-f JSON, --config JSON JSON config file
Several algorithms are implemented in a file. Note though that the idea is anyone using this structure can 'plug-in' their own algorithms.
Stores information about the system being simulated. It is attached closely to algorithm objects. Allows for the user to implement their own energy function (Hamiltonian).
Simulations generally evolve based on the interaction between 'nodes'. This class provides a tool to build a 2D periodic or non-periodic array of nodes.
Useful functionalities.