cuda.Stats

Module Contents

class cuda.Stats.Stats

Statistics Routines for cuda vectors ana matrices

amin(a)

Return the minimum of a vector or a matrix or minimum along an axis. :param a: a cuda vector or matrix :return: the minimum value

amax(a)

Return the maximum of a vector or a matrix or maximum along an axis. :param a: a cuda vector or matrix :return: the maximum value

l1norm(x, size=None, stride=1)

L1 norm of a vector sum_i |x_i| :return:

l2norm(x, size=None, stride=1)

L2 norm of a vector sqrt{sum_i x_i^2} :return:

linfnorm(x, size=None, stride=1)

L-Infinity norm of a vector max|x_i| :return:

covariance(x, y, out=None, axis=0, ddof=1)

Compute covariance between two vectors or matrices (along row or col) cov(X,Y) = E[(X-E[X])(Y-E[Y])] :param x, y: two input vectors or matrices :param axis: (for matrices only) 0/1 = along row/col :param out: (for matrices only) vector of size shape[1]/shape[0] for axis = 0/1 :return: float for vectors, vector for matrices

correlation(x, y, axis=0, out=None)

Compute correlation between two vectors or matrices (along row or col) cor(X,Y) = cov(X, Y)/ (std(X)std(Y)) :param y: :param axis: :param out: :return:

max_diff(x, y)

compute maximum difference between elements of two vectors or matrices max{|x_i - y_i|} :param x, y: two vectors or matrices :return: the maximum difference max|x_i - y_i|