I have written the following function in cython to estimate the log-likelihood
@cython.boundscheck(False)
@cython.wraparound(False)
def likelihood(double m,
double c,
np.ndarray[np.double_t, ndim=1, mode='c'] r_mpc not None,
np.ndarray[np.double_t, ndim=1, mode='c'] gtan not None,
np.ndarray[np.double_t, ndim=1, mode='c'] gcrs not None,
np.ndarray[np.double_t, ndim=1, mode='c'] shear_err not None,
np.ndarray[np.double_t, ndim=1, mode='c'] beta not None,
double rho_c,
np.ndarray[np.double_t, ndim=1, mode='c'] rho_c_sigma not None):
cdef double rscale = rscaleConstM(m, c,rho_c, 200)
cdef Py_ssize_t ngals = r_mpc.shape[0]
cdef np.ndarray[DTYPE_T, ndim=1, mode='c'] gamma_inf = Sh(r_mpc, c, rscale, rho_c_sigma)
cdef np.ndarray[DTYPE_T, ndim=1, mode='c'] kappa_inf = Kap(r_mpc, c, rscale, rho_c_sigma)
cdef double delta = 0.
cdef double modelg = 0.
cdef double modsig = 0.
cdef Py_ssize_t i
cdef DTYPE_T logProb = 0.
#calculate logprob
for i from ngals > i >= 0:
modelg = (beta[i]*gamma_inf[i] / (1 - beta[i]*kappa_inf[i]))
delta = gtan[i] - modelg
modsig = shear_err[i]
logProb = logProb -.5*(delta/modsig)**2 - logsqrt2pi - log(modsig)
return logProb
but when I run the compiled version of this function, I get the following error message:
File "Tools.pyx", line 3, in Tools.likelihood
def likelihood(double m,
ValueError: ndarray is not C-contiguous
I could not quite understand why this problem occurs??!!! I will appreciate to get any useful tips.
1 Answer
Just before you get the error, try printing the flags attribute of the numpy array(s) you're passing to likelihood. You'll probably see something like:
In [2]: foo.flags
Out[2]:
C_CONTIGUOUS : False
F_CONTIGUOUS : True
OWNDATA : True
WRITEABLE : True
ALIGNED : True
UPDATEIFCOPY : False
Note where it says C_CONTIGUOUS : False, because that's the issue. To fix it, simply convert it to C-order:
In [6]: foo = foo.copy(order='C')
In [7]: foo.flags
Out[7]:
C_CONTIGUOUS : True
F_CONTIGUOUS : False
OWNDATA : True
WRITEABLE : True
ALIGNED : True
UPDATEIFCOPY : False
