

TEVars

A TEVars instance maps variables of a custom delay reconstruction to the correct marginals during TE computation. In practice, this translates to using the correct columns of the Dataset furnishing the delay reconstruction when computing the different marginals during transfer entropy computation.

Map reconstruction variables to marginals using keywords

# TransferEntropy.TEVars — Method.

TEVars(;Tf = Int[], Tpp = Int[], Spp = Int[], Cpp = Int[])

Which axes of the state space correspond to the future of the target (Tf), the present/past of the target (Tpp), the present/past of the source (Spp), and the present/past of any conditioned variables (Cpp)? Indices correspond to variables of the embedding or colums of a Dataset.

This information is used by the transfer entropy estimators to ensure the marginal distributions are computed correctly.

Infer the mapping between reconstruction variables and marginals

Alternatively, the mapping is inferred from the order of the inputs. Here, target_future is the $T_{f}$ component, target_presentpast is the $T_{pp}$ component, and source_presentpast is the $S_{pp}$ component of the delay reconstruction. For conditional analyses, the additional component conditioned_presentpast (the $C_{pp}$ component) is needed.

# TransferEntropy.TEVars — Method.

TEVars(target_future::Vector{Int},
        target_presentpast::Vector{Int},
        source_presentpast::Vector{Int})

Which axes of the state space correspond to the future of the target, the present/past of the target, and the present/past of the source? Indices correspond to variables of the embedding or colums of a Dataset.

This information is used by the transfer entropy estimators to ensure the marginal distributions are computed correctly.

This three-argument constructor assumes there will be no conditional variables.

# TransferEntropy.TEVars — Type.

TEVars(target_future::Vector{Int}
    target_presentpast::Vector{Int}
    source_presentpast::Vector{Int}
    conditioned_presentpast::Vector{Int})

Which axes of the state space correspond to the future of the target, the present/past of the target, the present/past of the source, and the present/past of any conditioned variables? Indices correspond to variables of the embedding or colums of a Dataset.

This information is used by the transfer entropy estimators to ensure the marginal distributions are computed correctly.