implemented derivation of gradient weights via rotation

derive gradient weights of other direction with already calculated weights of one direction via rotation and apply them to a field.

For the 2D case in the as_matrix function the result is converted from a MutableDenseNDimArray to a MutableDenseMatrix to keep the compatibility in the test cases since MutableDenseMatrix can not be used for 3D cases. I am not sure whether it is better to change the datatype to MutableDenseNDimArray in general. Working with both types might lead to inconsistency problems somewhere sometime.

code snippet where the datatype is changed in derivation.py:

result = sp.MutableDenseNDimArray([0] * number_of_elements, shape)

if dim == 2:
    for direction, weight in zip(self.stencil, self.weights):
        result[max_offset - direction[1], max_offset + direction[0]] = weight
    result = result.tomatrix()

We could rename the function you implemented to as_array. This function could consistently return a DenseNDimArray. The old as_matrix function can be deprecated then by using a deprecation warning.

All tests should then also use the new as_array function.

added 1 commit

• 248a5e0d - implemented derivation of gradient weights via rotation

Compare with previous version

Everything done

merged

mentioned in commit dc8453b3

mentioned in issue #8 (closed)