dysii Dynamic Systems Library

dysii is a C++ library for distributed probabilistic inference and learning in large-scale dynamical systems. It features the following:

Filtering and smoothing

• Kalman filter and smoother
• Rauch-Tung-Striebel (RTS) smoother
• Unscented Kalman filter and smoother
• Particle filter and forward-backward smoother
• Kernel forward-backward and two-filter smoothers
• Multiple resampling strategies for particle filters, including stratified, auxiliary and regularised resampling

Probability distributions

• Gaussian and Gaussian mixture distributions
• Dirac and Dirac mixture (weighted sample set) distributions
• Kernel density estimators

Ordinary and stochastic differential equations

• Adaptive numerical solvers for ordinary and stochastic differential equations, including Euler-Maruyama and stochastic Runge-Kutta
• Autocorrelator and equilibrium distribution sampler

Parallelisation using MPI

• Parallel particle filter and smoother
• Parallel kernel forward-backward and two-filter smoother
• Distributed storage of mixture densities
• Distibuted kd tree evaluations, including dual- and self-tree evaluations

Data management

• Serialization of vectors, matrices and probability distributions for fast and convenient data management, using Boost.Serialization
• Text file reader and writer

Performance

• Use of BLAS and LAPACK
• Template meta-programming
• Code profiling
• Compiler optimisation

The library has been optimised for performance, while maintaining a modularity and generality that makes it suitable for a wide range of applications.

I’ve posted the final version of my PhD thesis, "Bayesian Learning of Continuous-Time Dynamical Systems, with Applications in Functional Magnetic Resonance Imaging" to the research page. Now assessed, corrected and passed!
Note that this may serve as a useful manual for some of the detail behind the algorithms of the dysii Dynamic Systems Library.

Two example applications have been made available for dysii, a toy linear model and more elaborate double well stochastic differential model. The former of these will serve as a useful tutorial example for those wanting to learn to use the library. The latter is a more complex application with a number of features that make [...]

Version 1.4.0 of the dysii Dynamic Systems Library has been released. This is a major new release with a number of additional features and performance enhancements, as well as representing a consolidation of code and maturation of much of the API.
Particular new features include:

The kernel forward-backward and two-filter smoothers, suitable for fast, large-scale approximate inference [...]