v0.4.3

Pathogen v0.4.3

• Visualization improvements
• Example update

Diff since v0.4.2

v0.4.2

Pathogen v0.4.2

• Use BitArrays for transmission networks
• Update dependency compatibility
• Drop various unnecessary struct mutability
• Visualization improvements

Diff since v0.4.1

v0.4.1

Pathogen v0.4.1

• Address Distributions.jl deprecations for bounded distributions
• Improve Base.show for Simulations

Diff since v0.4.0

v0.4.0

Pathogen v0.4.0

• Compatibility improvements
• Add AbstractRisks abstract type, improve code reuse
• Embrace nothing in structs through type unions
• Drop unnecessary mutability from various structs
• Change some log message levels
• Other small bugfixes

Diff since v0.3.1

v0.3.1

Pathogen v0.3.1

• Likelihood bugfix
• Generalization of distance matrix

Diff since v0.3.0

v0.3.0

Pathogen v0.3.0

• Simulations and inference can now be optionally initialized with non-susceptible individuals

v0.2.0

Pathogen v0.2.0

• Perform event time data augmentation in batches, with an O(n) increase in computation time with number of event_batches. More batches are generally required for higher dimensional and/or difficult to explore event time-spaces.
• Options for network restricted and unrestricted event time data augmentation. The network restricted method is more conservative in the proposals it generates, which is necessary for some more complicated model extensions. The unrestricted data augmentation process is the default.
• Loglikelihood shortcircuiting. When performing MCMC the loglikelihood that would be required for acceptance by the Metropolis-Hastings algorithm is pregenerated/calculated. If the loglikelihood calculation goes below this threshold it will conclude immediately.
• Implementation of Adaptive MCMC using an O(1) online covariance matrix calculation from the OnlineStats.jl package.
• Updates to examples.

v0.1.0

Pathogen v0.1.0

• Initial release of Pathogen.jl
• Pathogen.jl provides simulation and Bayesian inference methods for individual level infectious disease models. SEIR, SEI, SIR, and SI models are supported.