Module simd 

SIMD-accelerated operations for consciousness computation.

Provides vectorized KL-divergence, entropy, and matrix operations critical for Φ computation hot paths.

Functions

build_mi_edges

Build MI edge list (i, j, weight) with threshold pruning.

build_mi_matrix

Build full pairwise MI matrix (symmetric, zero diagonal). Returns flat n×n row-major matrix.

conditional_distribution

Extract conditional distribution P(future | state) from TPM row.

dense_matvec

Dense matrix-vector multiply y = A * x (row-major A). Used for TPM operations in Φ computation.

emd_l1

Earth Mover’s Distance (EMD) approximation for distribution comparison. Used in IIT 4.0 for comparing cause-effect structures.

entropy

Shannon entropy H(P) = -Σ p_i * ln(p_i).

entropy_scalar

kl_divergence

Compute KL divergence D_KL(P || Q) = Σ p_i * ln(p_i / q_i).

kl_divergence_scalar

Scalar KL divergence with branch-free clamping.

marginal_distribution

Compute marginal distribution by averaging over all rows.

pairwise_mi

Pairwise mutual information between elements i and j given marginals.