Accessing Matrix row/col ranges produces incorrect results

[crgfn]

Initialization

from sympy import Matrix, shape
M = Matrix([[1, 2, 3], [4, 5, 6], [7, 8, 9]])

print(M)
Matrix([[1, 2, 3], [4, 5, 6], [7, 8, 9]])

pprint(shape(M))
(3, 3)

Accessing ranges produces incorrect results

print(M[0:1,0:1])
Matrix([[1]])

print(M[0:2,0:2])
Matrix([[1, 2], [4, 5]])

print(M[0:3,0:3])
Matrix([[1, 2, 3], [4, 5, 6], [7, 8, 9]])

print(M[0:9,0:9])
Matrix([[1, 2, 3], [4, 5, 6], [7, 8, 9]])

[wermos]

Can you elaborate on which one of these you think is the incorrect result? Also, it would be nice if you use the code font for the code snippets. You can use single backticks (`) for single lines, and triple backticks (```) for multiline snippets.

As far as print(M[0:1,0:1]) and print(M[0:2,0:2]) and print(M[0:3,0:3]) is concerned, this is expected and documented behavior. See this section for a reference.

I agree that the output of print(M[0:9,0:9]) is surprising.

[wermos]

I did a little investigating, and the behavior is arising from this part of the code:

sympy/sympy/matrices/repmatrix.py

Lines 966 to 978 in b412fa4

	if isinstance(i, slice):
	i = range(self.rows)[i]
	elif is_sequence(i):
	pass
	else:
	i = [i]
	if isinstance(j, slice):
	j = range(self.cols)[j]
	elif is_sequence(j):
	pass
	else:
	j = [j]
	return self.extract(i, j)

Here, if i and j are slice instances, their max values are being clamped to the size of the matrix.

I'm not sure if this is the intended/desired behavior or not. If it is, then it should definitely be documented somewhere.

[oscarbenjamin]

This is just how slicing in Python works:

In [58]: l = [1,2,3,4]

In [59]: l[:10]
Out[59]: [1, 2, 3, 4]

[sylee957]

I don't think that we should change this. It is reasonable and robust that matrix should follow how Python slicing just works, and changing that gives further confusion that benefit.

Also, in general programming advice, you should avoid writing code like x[1:100] in favor of x[1:] or x[2:], if you don't know the length of the list.
It is more robust when the shape of the list changes.