[Python] How to index an ndarray of unknown dimensions

I have a python code that uses a numpy array to store multi-dimensional data. The dimensionality of the array is determined at runtime so I cannot know in advance the exact number of dimensions, which can be from 3 to 6. One thing I know, though, is that when a dimension exists in the array, it has 100 elements. So right now my analysis code looks something like this:

ndim = myarray.ndim
if ndim == 2:
for p0 in range(100):
do_something(myarray[:, p0])
if ndim == 3:
for p0 in range(100):
for p1 in range(100):
do_something(myarray[:, p0, p1])
elif ndim == 4:
for p0 in range(100):
for p1 in range(100):
for p2 in range(100):
do_something(myarray[:, p0, p1, p2])
elif ndim == 5:
for p0 in range(100):
for p1 in range(100):
for p2 in range(100):
for p3 in range(100):
do_something(myarray[:, p0, p1, p2, p3])
elif ndim == 6:
for p0 in range(100):
for p1 in range(100):
for p2 in range(100):
for p3 in range(100):
for p4 in range(100):
do_something(myarray[:, p0, p1, p2, p3, p4])


Of course this works, but I find the code not very elegant and very much cluttered. Is there a better way to do this? I am pretty sure there must be a way to index the array without knowing a priori the number of dimensions, but I cannot find the proper function in the numpy documentation.

Continue reading...