Modified python code

#!/usr/bin/env python

from numpy.random import randn
from numpy import sqrt,row_stack,array,dot,transpose,sum,zeros,inner,zeros,mean,float32
def GMMGaussianMC(N=100):
    T=100;mu=0.1;sigma=0.2
    invT = 1./T
    coverage=zeros((2,N),dtype=int)
    d = array([[-1,0],[0,0]],dtype=float32)
    xmat = mu + sigma*randn(T,N)
    muhats =  invT * sum(xmat, axis=0)
    f1s = xmat - muhats
    sigma_hats = sqrt(invT * sum(f1s**2,axis=0))
    f2s = f1s**2 - sigma_hats**2
    f = row_stack((f1s,f2s))
    for n in xrange(0,N-1):
    #d is now inv(d) as it is easy to compute
        d[1,1] = (-2*sigma_hats[n])
        temp = f[:,n].reshape(2,T)

        S=invT*inner(temp,temp)
        V=dot(dot(d,S),d)

        SE_mu = sqrt(invT*V[0,0])
        SE_sigma = sqrt(invT*V[1,1])
        coverage[0,n]=(abs(muhats[n]-mu)<1.96*SE_mu)
        coverage[1,n]=(abs(sigma_hats[n]-sigma)<1.96*SE_sigma)
    y=mean(coverage,axis=1)
    return y

if __name__=='__main__':
    #import cProfile
    #cProfile.run('GMMGaussianMC(100000)','prof_jason.dat')
    import time
    t1=time.time()
    y=GMMGaussianMC(100000)
    print time.time()-t1