Re:Fast Non-Standard Data Structures for Python
by sbrother on Monday January 28, 2013 @01:04PM
(#5129673)
I'd like to add scipy.spatial.cKDTree for really fast kd-trees. Just used it for a geospatial data analysis application (find the closest N facilities to a given point) and it blew the numpy and standard python alternatives out of the water in terms of speed.
Re:Fast Non-Standard Data Structures for Python
by willvarfar on Monday January 28, 2013 @07:32AM
(#5128221)
Just a random anecdote: I've been badly burned by the heapq for timeouts; although I understand the advantages of the heap in general I think the Python library version is shot-down in practice due to being a purely-Python implementation. I moved to rbtree and suddenly my throughput shot up.
Re:Fast Non-Standard Data Structures for Python
by z-factor on Monday January 28, 2013 @10:15AM
(#5128774)
I want to add that I've been using heapq for scheduling in every piece of software as well and it's never been a problem even in heavily loaded projects.
I've used heapq for several of my projects, including one where I did extensive performance testing, and I haven't found it to be a problem.