You are not logged in.

#1 2006-10-24 19:33:53

nv
Member
Registered: 2006-10-24
Posts: 39

python numeric goes numpy

Hello everyone,

I am using a lot of science packages for python. The current 'standard' package for numerical computation in python is 'python-numeric'. Tomorrow (october 25th) is the official release day for the replacement of numeric, called numpy. It is from the same author, Travis Oliphant, and has a very active user community. A few python packages that were using numeric are already enabled to use numpy, e.g. matplotlib and pytables. Some packages need a recompile (e.g. pytables), while the pure python packages, e.g. matplotlib, can use it at runtime.

There is currently a 'numpy' package (rc2) in AUR, along with two dependencies in AUR, and even though I already voted for it, I would like to know if there are any immediate plans to include python-numpy into the official arch tree? This would make it a lot easier to build numeric python extensions for arch. I think it is already as important as python-numeric, judging from the traffic on their official mailing list.

Best regards,

Niklas.

Offline

#2 2006-10-24 20:42:08

cactus
Taco Eater
From: t͈̫̹ͨa͖͕͎̱͈ͨ͆ć̥̖̝o̫̫̼s͈̭̱̞͍̃!̰
Registered: 2004-05-25
Posts: 4,622
Website

Re: python numeric goes numpy

interesting.. does numpy require python 2.5, or is it meant as a full retroactive replacement of numeric?


"Be conservative in what you send; be liberal in what you accept." -- Postel's Law
"tacos" -- Cactus' Law
"t̥͍͎̪̪͗a̴̻̩͈͚ͨc̠o̩̙͈ͫͅs͙͎̙͊ ͔͇̫̜t͎̳̀a̜̞̗ͩc̗͍͚o̲̯̿s̖̣̤̙͌ ̖̜̈ț̰̫͓ạ̪͖̳c̲͎͕̰̯̃̈o͉ͅs̪ͪ ̜̻̖̜͕" -- -̖͚̫̙̓-̺̠͇ͤ̃ ̜̪̜ͯZ͔̗̭̞ͪA̝͈̙͖̩L͉̠̺͓G̙̞̦͖O̳̗͍

Offline

#3 2006-10-24 21:23:35

nv
Member
Registered: 2006-10-24
Posts: 39

Re: python numeric goes numpy

numpy works with 2.4 and (as far as I remember from their mailing list) with 2.5. I don't know about 2.3 though. The core of the numpy package (the array protocol) was actually meant to be put into python 2.5, but didn't make it on time.

Numpy is the successor and a full replacement for both numeric and numarray. On the C-API side, there is a compatibility layer (or something similar), so that existing C-extensions written for numeric can easily switch to numpy. The biggest disadvantage of numeric, the time for the creation of new arrays, has been solved and is supposedly even better than in numarray.

A lot more information can be found on http://www.scipy.org.

Offline

Board footer

Powered by FluxBB