In filter, raise TypeError or ValueError.

type.is_valid_value relies on filter() raising either TypeError
or ValueError for invalid values, not AssertionError.

Sparse indexing

Optimization to move tensordot to GPU

For instance, m[:,a:b], or m[a:,:b]

Some types (like Generic) allow for a value of None.
We should not consider it an error.

updated Assert docstring as Fred suggested

Fix misc buildbot tests

Some doc

Error was:
ImportError: No module named theano

Doc2

dtype for different Tensor CAReduce operations

Ops:GetItem2D&GetItemScalar to return a subtensor/scalar from a sparse m...

gpu accelerated tensordot by conversion to matrix product

Ops:GetItem2D&GetItemScalar to return a subtensor/scalar from a sparse matrix tested. grad() not implemented because too complicated for now