[theano-users] Using Theano for Symbolic Matrix and Tensor Differentiation

['Herbert Meier' via theano-users]

Dear Theano experts,

I'm having a question regarding theano's capability for a
non-machine-learning application, as I am actually interested in obtaining
derivatives for symbolic tensor and matrix expressions.
A colleague of mine suggested to have a look at theano for this purpose as
other software such as Mathematica, MATLAB and Maple are not able to
achieve what I am looking for.

In fact, I am looking for a software that is able to compute symbolically
Jacobians and Hessians of vector valued functions such as
f(X) = Y*Z
where Y=Y(X) and Z=Z(Y), f(): R^n -> R^m.

Is this possible with theano? If so, where to find some corresponding code
snippets/examples?

Thanks a lot for your help in advance!