I think iyami said it well. Just take a formulation for dispersive eps, then exchange eps for mu, and E for H, and you will have the formulation for dispersive mu. When you write your code implement both formulations at the same time.
Are you going to simulate the actual structures (split ring resonator or whatever) that gives you the negative mu and eps? If so, I am skeptical you need the dispersive mu feature. Most materials have negligible magnetic response and the structure itself will automatically produce the "effective" magnetic response. If you are not doing this and modeling the negative index material as an effective medium of if you are actually incorporating materials with a magnetic response, I can understand why you need this.
Anyway, attached are some references that may help you, but they are taking the effective medium approach.
-Tip