Dear Stats-Experts:

I am at loss in my research project and would appreciate any input, reference to literature, publications, etc. Here it goes:

I have a set of directional data which I know is a combination of two von Mises distributions. I have been able to isolate one contributor through experiments but I cannot measure the other directly. So the only way is measuring the combination.

Unfortunately, I have not been able to find any literature on how I can mathematically solve for a von Mises model of the unknown contributor.

Does anyone know how I can "subtract" one von Mises distribution from another to identify the parameters of the unknown distribution?

Thank you for your help.

-Tuco