correlation()

double MayaFlux::Kinesis::correlation ( std::span< const double >  a, std::span< const double >  b  )

inlinenoexcept

Pearson correlation between two equal-length windows.

Parameters

a	First span
b	Second span, same length as a

Returns

Correlation in -1..1, or zero if either span has near-zero variance or the lengths differ

The base relation measure: do these two streams move together, apart, or independently, over this window, with no notion of one leading the other. Scale-free, so a pressure axis in 0..1 and a tilt axis in degrees compare meaningfully without the caller normalizing first.

Definition at line 45 of file Relation.hpp.

{
    if (a.size() != b.size() || a.size() < 2)
        return 0.0;
 
    const size_t n = a.size();
    double mean_a = 0.0;
    double mean_b = 0.0;
    for (size_t i = 0; i < n; ++i) {
        mean_a += a[i];
        mean_b += b[i];
    }
    mean_a /= static_cast<double>(n);
    mean_b /= static_cast<double>(n);
 
    double cov = 0.0;
    double var_a = 0.0;
    double var_b = 0.0;
    for (size_t i = 0; i < n; ++i) {
        const double da = a[i] - mean_a;
        const double db = b[i] - mean_b;
        cov += da * db;
        var_a += da * da;
        var_b += db * db;
    }
 
    const double denom = std::sqrt(var_a * var_b);
    if (denom < 1e-12)
        return 0.0;
 
    return std::clamp(cov / denom, -1.0, 1.0);
}

References a, and b.

Referenced by coherence(), estimate_period(), and relation_at_lag().

Here is the caller graph for this function: