A vector, where all elements are bounded between 0 and 1, and sum(vector)=1.

[murrellb]

Love the package.

I often have components of models where I need to parameterize a categorical probability distribution. I use the following trick:

σ(z::Real) = one(z) / (one(z) + exp(-z))
function unconstrained2probvec(v)
    l = length(v)
    remaining = 1.0
    o = zeros(l+1)
    for i in 1:l
        #log(l+1-i) is a "balancing" shift to ensure that a vector of zeros givens you [1/N ... 1/N]
        o[i] = remaining*σ(v[i] - log(l+1-i))
        remaining -= o[i]
    end
    o[end] = max(0.0,remaining)
    return o
end

This takes a vector of N-1 unbounded values and returns a vector of N values, bounded between 0 and 1, that sums to 1. When input is all zeros, the output is all 1/N.

I'm too stupid to figure out how to build this into your package, but if you think it would be sensible to do so, please consider this a feature request.

[willtebbutt]

I would also be keen to have this as a feature, thanks for bringing it up!