Path integral for spin? I'm searching for the path integral formulation for a spin particle and haven't found any precise description yet. Is there a systematic way to construct a non-relativistic path integral formulation for a spin particle?
Edit: Here are some my thoughts on this problem. 
In Pauli theory of spin (non-raletivistic theory of spin), if we only consider the dimension of spin, then after we set a particular $z$ direction, all states of a spin 1/2 particle can be described by an element in $SU(2)$, so the time evolution of spin state will form a connected path in $SU(2)$, I think maybe we can construct path integral over all possible path between two ends that we set. Just a guess. 
 A: Within the Altland "Condensed Matter field theory" there is an entire section on the path integral for spin particles.
As you suggested the basic idea is to parametrize the spin with a representation of SU(2). This is sometimes called a spin-coherent-state (call it $|g\rangle$). On SU(2) you can define a (left) invariant integral and with that you can represent the identity as
\begin{equation}
  \mathrm{id} = C \int_{SU(2)}dg |g\rangle\langle g|,
\end{equation}
where C is for normalization. Observe the structural similarity of this representation and the representation of the bosonic identity by bosonic coherent states.
Now the derivation of a path integral is quite similar to the bosonic case: you do a Trotter decomposition of the time evolution operator and insert identities in the above form. How the diverse spin operators act on $|g\rangle$ are described in the Altland; look there for reference.
As a final side note: this procedure gives yet another way to "show" that spins can only take half-integer values (find that also in the book).
