wrong in TransitionDensity.py about pymle HOT 1 CLOSED

Thank you for this great catch! Indeed, the previous code was not suitable for the time-inhomogenous coefficient case. I have refactored the code so that it should support this case, and added some unit tests for the existing (time homogenous) models. I have not yet thoroughly tested time-inhomogenous coefficient case, so please let me know if you see any other issues. See tests\fit\Test_AnalyticalMLE.py for an example usage, specifically:

    Euler_est = AnalyticalMLE(sample=sample, param_bounds=param_bounds, 
                              dt=dt*np.ones(len(sample) - 1),
                              density=EulerDensity(model),
                              t0=np.linspace(0, T-dt, len(sample) - 1),
                              ).estimate_params(guess)

Note how now you have the option to pass in dt as a vector (in this case it's uniform, same dt for each observation) as well as a vector of t0, which is where the coefficients will be evaluated. Please reach out with any questions,

• Justin

from pymle.