def test_boundscheck(x):
"""
>>> test_boundscheck([1, 2, 3])
3
>>> try: test_boundscheck([1, 2])
... except IndexError: pass
"""
with cython.boundscheck(True):
return x[2]
def test_boundscheck(x):
"""
>>> test_boundscheck([1, 2, 3])
3
>>> try: test_boundscheck([1, 2])
... except IndexError: pass
"""
with cython.boundscheck(True):
return x[2]
if mu.shape[1] > 1:
mu = mu.T
cdef int D = int_max(xs.shape[0],xs.shape[1])
return sum( -O*(0.5*np.log(2*np.pi*sigma)) - ((xs-mu)**2)/(2*(O)*sigma))
@cython.boundscheck(False)
@cython.wraparound(False)
@cython.initializedcheck(False)
@cython.cdivision(True)
cdef lfrogupdate(np.ndarray[np.float64_t, ndim=2] E, np.ndarray[np.float64_t, ndim=2] X, np.ndarray[np.float64_t, ndim=1] Y, np.ndarray[np.float64_t, ndim=2] f, int D, double alpha, np.ndarray[np.float64_t, ndim=2] wnew):
"""Update momentum given current data in leapfrog iterations
"""
return ( np.dot(X.T,( Y[:,None] - (1./(1+np.exp(-f))) )) - E*(1/alpha)@wnew)
cython.boundscheck(False)
@cython.wraparound(False)
@cython.initializedcheck(False)
def BLR_sghmc_cython(int C, int Bhat, int D, np.ndarray[np.float64_t, ndim=2] Mass, np.ndarray[np.float64_t, ndim=2] w, double m, np.ndarray[np.float64_t, ndim=2] X, np.ndarray[np.float64_t, ndim=1] Y, np.ndarray[np.float64_t, ndim=2] f, double alpha, double StepSize, int BurnIn, int niters, double CurrentLJL):
"""Bayesian Linear Regression using HMC algorithm implemented using Cython
C is a user specified constant
Bhat is an approximate set to 0 here (it should converge to 0)
D is shape of data
Mass is the mass matrix of kinetic energy
w is a vector of coefficients to estimate
m is number of iterations for Monte-Carlo
X is the explanatory data matrix
Y is the explained vector
f fit given initial coefficients (0s)
alpha is variance of prior
StepSize dt for dynamics
