Utilities

gettransmat(seq; relabel = false) -> (Dict, Matrix)

Return the transition matrix associated to the label sequence seq. The labels must be positive integer.

Arguments

• seq::Vector{<:Integer}: positive label sequence.

Keyword Arguments

• relabel::Bool = false: if set to true the sequence will be made contiguous. E.g. [7,7,9,9,1,1] will become [2,2,3,3,1,1].

Output

• Dict{Integer,Integer}: the mapping between the original and the new labels.
• Matrix{Float64}: the transition matrix.

randtransmat([rng,] prior) -> Matrix{Float64}

Generate a transition matrix where each row is sampled from prior. The prior must be a multivariate probability distribution, such as a Dirichlet distribution.

Arguments

• prior::MultivariateDistribution: distribution over the transition matrix rows.

Example

A = randtransmat(Dirichlet([0.1, 0.1, 0.1]))

randtransmat([rng, ]K, α = 1.0) -> Matrix{Float64}

Generate a transition matrix where each row is sampled from a Dirichlet distribution of dimension K and concentration parameter α.

Arguments

• K::Integer: number of states.
• α::Float64 = 1.0: concentration parameter of the Dirichlet distribution.

Example

A = randtransmat(4)

remapseq(seq, ref) -> Vector{Integer}

Find the permutations of seq indices that maximize the overlap with ref.

Arguments

• seq::Vector{Integer}: sequence to be remapped.
• ref::Vector{Integer}: reference sequence.

Example

ref = [1,1,2,2,3,3]
seq = [2,2,3,3,1,1]
remapseq(seq, ref)
# [1,1,2,2,3,3]