Kernel module Defines basic kernel functions of the form:
funcname(x_p, x_q, par) that return a covariance matrix. x_p is n1*d, x_q is n2*d, the result should be n1*n2 par can be a scalar, array, or list of these.
All kernels must allow x_q=None, and efficiently compute the diagonal of K(x_p, x_p) as a (n1,) shaped vector.
Multi-task kernels must begin with mt_, and such kernels must use the last dimension of x_p and x_q as an indicator of the task.
This file also contains code for composing these kernels into multi-use objects.
Functions
| auto_range(user_kernel) | |
| cdist(XA, XB[, metric, p, V, VI, w]) | Computes distance between each pair of the two collections of inputs. |
| chisquare(x_p, x_q[, eps]) | |
| compose(user_kernel) | |
| describer(user_kernel) | |
| gaussian(x_p, x_q, LS) | |
| laplace(x_p, x_q, LS) | |
| matern3on2(x_p, x_q, LS) | |
| mt_weights(x_p, x_q, params) | |
| named_target(covfn, fn_cache) | |
| non_stationary(x_p, x_q, params) | Implementation of Paciorek's kernel where length scale is defined as |
| nonstat_rr(x_p, x_q, params) | |
| sin(x_p, x_q, params) | |
| tree(x_p, x_q, params) | Implementation of Paciorek's kernel where length scale is defined as |
| tree1D(x_p, x_q, params) | Implementation of Paciorek's kernel where length scale is defined as |
Classes
| Printer([val]) | |
| Range(lowerBound, upperBound, initialVal) |