The numerical scaling may be bad. NR = 7.9600e-14 might cause havoc numerically. Try to change units in order to get all non-zero input data to be within a small number of orders of magnitude of 1.
Also, if you have Mosek available as solver, use it; otherwise follow the advice in CVXQUAD: How to use CVXQUAD's Pade Approximant instead of CVX's unreliable Successive Approximation for GP mode, log, exp, entr, rel_entr, kl_div, log_det, det_rootn, exponential cone. CVXQUAD's Quantum (Matrix) Entropy & Matrix Log related functions .