al3c

al3c implements scalable Approximate Bayesian Computation using the ABC Sequential Monte Carlo (ABC-SMC) algorithm to perform parallel parameter inference from simulation-based models.

Key Features:

• C++ framework: Provided as a C++ framework for implementing ABC algorithms and application-specific code.
• ABC-SMC implementation: Implements the ABC Sequential Monte Carlo (ABC-SMC) algorithm for iterative parameter refinement.
• Parallel scalability: Distributes ABC computations across multiple processors to scale inference and reduce computation time.
• Addresses rejection-sampling inefficiencies: Targets computational inefficiencies of Monte Carlo rejection sampling by using ABC-SMC.
• Simulation-based, likelihood-free inference: Uses simulation of data and comparison to observed data for parameter estimation without explicit likelihoods.
• Model and prior specification: Requires user-specified simulation model and prior distribution functions.
• Configuration and extensibility: Configured via an XML file and extended through a C++ plug-in template for application-specific implementations.

Scientific Applications:

• Parameter inference for complex models: Inference for complex mechanistic or simulation-based models where traditional methods are computationally limited.
• Bayesian inference in bioinformatics: Applications in bioinformatics and other disciplines that rely on simulation-based, likelihood-free Bayesian inference.

Methodology:

Implements the ABC-SMC algorithm that iteratively refines parameters by simulating data, comparing simulated and observed data, and performing these steps in parallel; configured via an XML file and a C++ plug-in template.