Calculate the generalised distance between vectors t and t' using an anisotropy matrix A.
distMat and distMatSq calculate:
$$ [(t - t')^{p/2}]^T A (t - t')^{p/2} $$
distMatLinear and distMatLinearSq calculate:
$$ t^T A t' $$
distMat(input, inputNew, A, power)distMatSq(input, A, power)
distMatLinear(input, inputNew, A)
distMatLinearSq(input, A)
A matrix
Vector of the input coordinate t
Vector of the input coordinate t'
Anisotropy matrix A
Power value p
The distMatSq and distMatLinearSq functions are
used when input vectors t and t' are identical, returning a symmetric matrix.
When distMat and distMatSq functions are used in
powered exponential kernels, power=1 gives the exponential kernel and
power=2 gives the squared exponential one.
distMatLinear and distMatLinearSq functions are used in the
linear covariance kernel.