CUDA derangement implementation

Supposing that there is an array `a[n]` without any duplicate element (i.e. `a[i] != a[j]` for each `i != j`), I want to shuffle this array randomly and get an array `b[n]` that satisfies `b[k] != a[k]` for each `k`. This is known as [derangement](https://en.wikipedia.org/wiki/Derangement).

As far as I know, thrust and C++ STL library have no function to perform derangement. Is it possible that thrust can provide a GPU implementation of derangement? Thank you.