#8547: implement hidden markov models in Cython from scratch (so can remove the
GHMM standard package from Sage)
---------------------------+------------------------------------------------
Reporter: was | Owner: amhou
Type: enhancement | Status: needs_review
Priority: major | Milestone: sage-4.4
Component: statistics | Keywords:
Author: | Upstream: N/A
Reviewer: | Merged:
Work_issues: |
---------------------------+------------------------------------------------
Comment(by was):
* I've attached a part 2 patch. I made sure all cdef's methods have
docstrings and also that doctests are properly sphinx formated (some
weren't since they were copied from the old code).
* Jason said above that IntList.sum doesn't have a doctest for the
overflow case... but it does, so I don't know what he meant.
* I read the source code for {{{sage.misc.misc_c.normalize_index}}} and
cannot bring myself to use that in this situation. That function
actually returns a Python *list* of Python ints for every single index
into the list that is being sliced! That would easily lead to factor of
50-100 slowdowns on realistic operations:
{{{
sage: timeit('z = sage.misc.misc_c.normalize_index(slice(1,10^5),10^5)')
# slow because constructions a big python list
125 loops, best of 3: 2.17 ms per loop
sage: a = stats.IntList([1..10^5])
sage: timeit('a[1:10^5]') # slice is just a memcpy
625 loops, best of 3: 48.4 µs per loop
sage: 2.17/0.0484
44.8347107438017
}}}
Here's an example with a step:
{{{
sage: a = stats.IntList([1..10^5])
sage: timeit('a[1:10^5:2]')
625 loops, best of 3: 92.2 µs per loop
sage: timeit('z = sage.misc.misc_c.normalize_index(slice(1,10^5,2),10^5)')
625 loops, best of 3: 772 µs per loop
}}}
and that 772 microseconds is *before* we do the actual iteration through
the returned list of python ints, convert them to c ints, copy stuff
around in memory, etc.
This stats code I'm writing is really meant to be industrial strength --
the sort of code maybe somebody would use in "realtime" processing of
large datastreams. I don't want slow functions anywhere in there.
-- William
--
Ticket URL: <http://trac.sagemath.org/sage_trac/ticket/8547#comment:13>
Sage <http://www.sagemath.org>
Sage: Creating a Viable Open Source Alternative to Magma, Maple, Mathematica,
and MATLAB
--
You received this message because you are subscribed to the Google Groups
"sage-trac" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/sage-trac?hl=en.
