Mutual information between two sets of variates within a multivariate normal distribution

Function computing the mutual information between two exhaustive and mutually exclusive splits of a set of multivariate normal random variables.

GGMmutualInfo(S, split1)

Arguments

S

A positive-definite covariance matrix.

split1

A numeric, indicating the variates (by column number) forming the first split. The second split is automatically formed from its complement.

Value

A numeric, the mutual information between the variates forming split1 and those forming its complement.

References

Cover, T.M., Thomas, J.A. (2012), Elements of information theory.

See also

covML, ridgeP.

Author

Wessel N. van Wieringen, Carel F.W. Peeters <carel.peeters@wur.nl>

Examples

# create a covariance matrix
Sigma <- covML(matrix(rnorm(100), ncol=5))

# impulse response analysis
GGMmutualInfo(Sigma, c(1,2))
#> [1] 0.7200012