Thanks for the hint. I am currently happy with %prun followed by %lprun.
The only thing I dislike about this is that I have to guess who calls the
most offending method or function, but it's not much of a hassle, since
%prun also includes the cumulative timings.
Martin
Am Freitag, 15. Septem
By the way, if you are doing a lot of optimisation you should consider using
gprof2dot https://github.com/jrfonseca/gprof2dot . It can turn your python
profiling data into a really useful grafical visualisation of the call graph,
instantly showing you which parts of the code to optimize. And I i
Thu 2017-09-14 18:06:48 UTC, Martin R:
> almost by accident, I just noticed the following:
>
> [...]
>
> Is there any drawback to the faster method?
No! Good that you tested and found what works faster.
Sorry if anything I said was misleading.
Are you working on speeding up the 'to_matrix' and
Dear Samuel,
almost by accident, I just noticed the following:
def to_matrix_1(self):
entries = { (v-1, i): 1 for i, v in enumerate(self) }
return MatrixSpace(ZZ, len(self), sparse = True)(entries)
def to_matrix_2(self):
entries = { (v-1, i): 1 for i, v in enumerate(self) }
retur
Mon 2017-09-11 10:07:01 UTC-5, Martin R:
> Is there a recommended (fast) way to turn input
> into a (square) matrix (with entries in ZZ)?
> Currently I do "corner = matrix(corner)"
Suppose you defined
m_list = [[0, 1, 2], [3, 4, 5], [6, 7, 8]]
and you want to turn m_list into a 3 by 3 matri
Thank you! Is there a recommended (fast) way to turn input into a (square)
matrix (with entries in ZZ)?
currently I do "corner = matrix(corner)"
Thanks again,
Martin
Am Montag, 11. September 2017 17:00:20 UTC+2 schrieb Samuel Lelievre:
>
> Mon 2017-09-11, 9:38:45 (UTC-5), Martin R:
> > Appa
Mon 2017-09-11, 9:38:45 (UTC-5), Martin R:
> Apparently, corner[i,j] is a lot faster than corner[i][j].
Yes: if m is a matrix, m[i, j] accesses the element (i, j) of m, while
m[i][j] first extracts line i of m, stores it as a vector, then extracts
element j of that vector. This explains the speed
Apparently, corner[i,j] is a lot faster than corner[i][j]. With this
modification, the two versions are roughly equally fast. I don't get it,
though, because the second version should be doing a lot more work.
Martin
Am Montag, 11. September 2017 16:23:07 UTC+2 schrieb Martin R:
>
> I don't u