Using the MCS functionality:

I am trying to assign a distribution to a parameter alpha = Parameter(index=[1:5, 1:5]). I want to enforce that **sum(alpha[:,:]) = 1 for all runs**. I can see two ways to do this:

- Assign e.g. a normal distribution to all elements of alpha, and divide by the sum of the values taken: alpha[:,:] = Normal(1,0) / sum(alpha[:,:]). However this is not supported by @defmcs.
- Assign a Dirichlet distribution to alpha, which gives an array of values of sum=1: alpha[:,:] = Dirichlet(25,1). Yet this gives me a 25-element Array{Float64,1}, while I would like a 5*5 Array{Float64,2}. A way to get to the latter shape would be to use reshape(Dirichlet(25,1), 5,5). However this is not supported by @defmcs either.

Do you have suggestions as to how to solve this?