Welcome to libEnsemble’s documentation!

libEnsemble is a Python toolkit for coordinating asynchronous and dynamic ensembles of calculations.

• New to libEnsemble? Start here.

• Try out libEnsemble with a tutorial.

• Go in depth by reading the Overview.

• See the FAQ for common questions and answers, errors, and resolutions.

• Check us out on GitHub.

Users interested in updating to v0.9.0+ should read the Updating for libEnsemble v0.9.0 wiki.

User Guide:

• Quickstart
  • Dependencies
  • Installation
  • Testing
  • Basic Usage
  • Resources
• Advanced Installation
  • pip
  • Conda
  • Spack
• Understanding libEnsemble
  • Overview
  • Example Use Cases
  • Glossary
• Programming with libEnsemble
  • Main libE Module - Calling Script
  • Output Management
  • Data Structures
  • User Function API
  • Writing User Functions
  • Executors
  • Resource Manager
  • Convenience Tools and Functions
• Running libEnsemble
  • MPI Comms
  • Local Comms
  • TCP Comms
  • Further command line options
  • liberegister / libesubmit
  • Persistent Workers
  • Environment Variables
  • Further run information
• Running on HPC Systems
  • Central v Distributed
  • Configuring the Run
  • Systems with Launch/MOM nodes
  • Balsam - Externally managed applications
  • Mapping Tasks to Resources
  • Overriding Auto-detection
  • funcX - Remote User functions
  • Instructions for Specific Platforms

Tutorials:

• Multiprocessing with a Simple Sine
  • Getting started
  • Generator function
  • Simulator function
  • Calling Script
  • Next steps
• Executor with Electrostatic Forces
  • Getting Started
  • Calling Script
  • Simulation Function
• Executor - Assign GPUs
  • Simulation function
  • Running the example
  • Changing number of GPUs per worker
  • Varying resources
  • Checking GPU usage
  • Example submission script
• Parallel Optimization with APOSMM
  • Six-Hump Camel Simulation Function
  • APOSMM Operations
  • APOSMM Persistence
  • Calling Script
  • Final Setup, Run, and Output
  • Applications
• Borehole Calibration with Selective Simulation Cancellation
  • Introduction - Calibration with libEnsemble and a Regression Model
  • Overview of the Calibration Problem
  • Point Cancellation Requests and Dedicated Fields
  • Allocation function
  • Calling Script - Reading Results
  • Using cancellations to kill running simulations

Examples:

• Generation Functions
  • sampling
  • persistent_sampling
  • APOSMM
  • uniform_or_localopt
  • persistent_tasmanian
  • persistent_fd_param_finder
  • persistent_surmise
• Simulation Functions
  • six_hump_camel
  • chwirut
  • noisy_vector_mapping
  • periodic_func
  • borehole
  • executor_hworld
• Allocation Functions
  • give_sim_work_first
  • fast_alloc
  • fast_alloc_to_aposmm
  • start_only_persistent
  • start_persistent_local_opt_gens
• Calling Scripts
  • Local Sine Tutorial
  • Electrostatic Forces with Executor
  • Persistent APOSMM with Gradients

Additional References:

• Frequently Asked Questions
  • Common Errors
  • HPC Errors and Questions
  • libEnsemble Help
  • macOS and Windows Errors
• Known Issues
• Release Notes
  • Release 0.9.2
  • Release 0.9.1
  • Release 0.9.0
  • Release 0.8.0
  • Release 0.7.2
  • Release 0.7.1
  • Release 0.7.0
  • Release 0.6.0
  • Release 0.5.2
  • Release 0.5.1
  • Release 0.5.0
  • Release 0.4.1
  • Release 0.4.0
  • Release 0.3.0
  • Release 0.2.0
  • Release 0.1.0
• Contributing to libEnsemble
• Posters and Presentations
  • SciPy 2020 Virtual Poster
  • SciPy 2020 PDF Poster
  • CSE 2019 Poster

Developer Guide:

• Release Management for libEnsemble
  • Release Process
  • Release Platforms
• libEnsemble Internal Modules
  • Manager Module
  • Worker Module
  • History Module
  • Resources Module
  • RSET Resources Module
  • Worker Resources Module
  • Environment Resources Module
  • Node Resources Module
  • MPI Resources Module
  • Scheduler Module

Indices and tables

• Index

• Module Index