MVTDST

Python wrappers to MVTDST subroutine are generated by f2py.

MVTDST computes an integral of a multivariate Student-T distribution, thus producing a numerical estimate for a multivariate Student-T CDF.

All credit for the integrator goes to Alan Genz. See the license file for details. BSD-3clause compatible.

Install

To manually build the python extension module:

f2py -c mvtdstpack_custom.pyf mvtdstpack.f

Depends:

- numpy (includes f2py)
- a functional fortran compiler

Example Use

In python, import the mvtdstpack module:

import mvtdstpack as mvt

Use the mvtdst routine to integrate a 2D Student-T distribution:

# 2d student t settings
dim, dof = 2, 10
# corr coeff parameter
rho = np.array([0.7])
#
# integration settings (int from -inf, to upperb)
lowerb = np.array([0.0, 0.0], dtype='double')
upperb = np.array([0.2, 0.2], dtype='double')
inFin = np.zeros(upperb.size, dtype='int')     # integration limit setting
delta = np.zeros(upperb.size, dtype='double')  # non centrality params
maxpts = int(1000 * dim)
abseps = 0.001
releps = 0.0
#
# Perform integration
error, value, status = mvt.mvtdst(dof, lowerb, upperb, inFin, rho, delta)
#
# It is optional to supply dim, if not supplied dim=len(lowerb)
error, value, status = mvt.mvtdst(dof, lowerb, upperb, inFin, rho, delta, n=dim)
#
# It is optional to suppy rel and abs error
error, value, status = mvt.mvtdst(dof, lowerb, upperb, inFin, rho, delta, maxpts, abseps, releps)

Note: mvtdst only accepts integers for degrees of freedom parameter.