From singer@itl.atr.co.jp Tue Dec 2 04:28:42 1997
From: singer@itl.atr.co.jp (Harald Singer)
Date: Tue, 02 Dec 1997 13:28:42 +0900
Subject: [MATRIX-SIG] [Q] casting double to char in NumPy1.0b3
Message-ID: <9712020428.AA05139@atrh16.itl.atr.co.jp>
Hi,
I am trying to map an array of doubles to an array of char (values
between 0 and 63) as entries into a colormap. In an older version
of NumPy this worked without any problems but in the latest version
I get:
>>> from Numeric import *
>>> x = array([1.,2.,3.])
>>> y = array(x,Int8)
Traceback (innermost last):
File "<stdin>", line 1, in ?
TypeError: Array can not be safely cast to required type
>>> x = array([1,2,3])
>>> y = array(x,Float64)
>>> y = array(x,Float32)
Traceback (innermost last):
File "<stdin>", line 1, in ?
TypeError: Array can not be safely cast to required type
So what is an efficient way of doing this type of operation?
Up to now, I have been doing something like
ScaleOffset = min(yy.flat)
ScaleFactor = 63.0/(max(yy.flat)-ScaleOffset)
yyy = array(((yy-ScaleOffset)*ScaleFactor)[:,:fftorder/2],Int8)
where yy is a 2-dimensional array of doubles (time, frequency bins)
and yyy is the same thing but mapped to 8bit values between 0 and 63,
which I use as entries into a colormap.
BTW, the moment I get a solution, I will post my spectrogram display
python script to the image-sig and matrix-sig mailing lists.
/harry
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From jhauser@ifm.uni-kiel.de Tue Dec 2 14:43:47 1997
From: jhauser@ifm.uni-kiel.de (Janko Hauser)
Date: Tue, 2 Dec 1997 15:43:47 +0100 (CET)
Subject: [MATRIX-SIG] [Q] casting double to char in NumPy1.0b3
In-Reply-To: <9712020428.AA05139@atrh16.itl.atr.co.jp>
References: <9712020428.AA05139@atrh16.itl.atr.co.jp>
Message-ID: <m0xctYB-000suOC@lisboa.ifm.uni-kiel.de>
Hi, try this, it should work :0)
>>> a=arange(10,typecode=Float32)
>>> a
array([ 0., 1., 2., 3., 4., 5., 6., 7., 8., 9.],'f')
>>> a.astype(Int8)
array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9],'1')
>>>
Now, we are eager to see the new great NumPy application :-)
__Janko, only speaking for himself.
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From busby@icf.llnl.gov Wed Dec 3 19:58:28 1997
From: busby@icf.llnl.gov (L. Busby)
Date: Wed, 3 Dec 97 11:58:28 PST
Subject: [MATRIX-SIG] Python-1.5b1 + LLNL extensions available
Message-ID: <9712031958.AA18761@icf.llnl.gov.llnl.gov>
I've done a quick, hopefully not too dirty, transfer of most of the
LLNL Python extensions to Python-1.5b1. See
ftp://ftp-icf.llnl.gov/pub/python/Python-1.5b1+LLNL.tgz.
Apologies in advance for the size of this file. I stress that this is
an initial release for those early adopters who need our extensions.
Our future plans per NumPy stewardship, unbundling of extensions, etc.
are unchanged. (They're still in the future. :-)
All our extensions save for the C++ changes are in the above file.
We're investigating if those can be added to Python-1.5 without changes
to the core code. If you need that extension *now* in Python-1.5b1, see
the file
ftp://ftp-icf.llnl.gov/pub/python/diffs-1.4-to-llnl.
This file is diff output for those files in Python-1.4 modified by
LLNL. A motivated patcher could add our C++ extensions to 1.5b1 in
fairly short order by inspecting those differences.
Lee Busby
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From m.hadfield@niwa.cri.nz Sun Dec 7 22:55:47 1997
From: m.hadfield@niwa.cri.nz (Mark Hadfield)
Date: Mon, 8 Dec 1997 11:55:47 +1300
Subject: [MATRIX-SIG] Subscription request
Message-ID: <01bd0363$41a1a480$b638cb83@gust.niwa.cri.nz>
This is a multi-part message in MIME format.
------=_NextPart_000_0011_01BD03D0.389AEC80
Content-Type: text/plain;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
subscribe
----
Mark Hadfield, m.hadfield@niwa.cri.nz, http://gust.niwa.cri.nz/
National Institute for Water and Atmospheric Research
PO Box 14-901, Wellington, New Zealand
------=_NextPart_000_0011_01BD03D0.389AEC80
Content-Type: text/html;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
<!DOCTYPE HTML PUBLIC "-//W3C//DTD W3 HTML//EN">
<HTML>
<HEAD>
<META content=3Dtext/html;charset=3Diso-8859-1 =
http-equiv=3DContent-Type>
<META content=3D'"MSHTML 4.71.2016.0"' name=3DGENERATOR>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT color=3D#000000 size=3D2>subscribe</FONT></DIV>
<DIV><FONT color=3D#000000 size=3D2>----<BR>Mark Hadfield, <A=20
href=3D"mailto:m.hadfield@niwa.cri.nz">m.hadfield@niwa.cri.nz</A>, =
<A=20
href=3D"http://gust.niwa.cri.nz/">http://gust.niwa.cri.nz/</A><BR>Nationa=
l=20
Institute for Water and Atmospheric Research<BR>PO Box 14-901, =
Wellington, New=20
Zealand</FONT> </DIV>
<DIV><FONT color=3D#000000 size=3D2></FONT> </DIV></BODY></HTML>
------=_NextPart_000_0011_01BD03D0.389AEC80--
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From V4N14Zu8X@le1ttrbox.com Sun Dec 7 04:11:50 1997
From: V4N14Zu8X@le1ttrbox.com (V4N14Zu8X@le1ttrbox.com)
Date: 07 Dec 97 4:11:50 AM
Subject: [MATRIX-SIG] We bulk Email for You
Message-ID: <37nIU707a4xWEsTM>
LET US DO YOUR BULK MAILINGS!!!
...$350 PER MILLION ADDRESSES SENT
...$250 PER 1/2 MILLION ADDRESSES SENT
THE WAY OF THE FUTURE FOR SUCCESS IN YOUR BUSINESS!
Our company will do bulk emailing for your product/service.
DON'T BE LEFT IN THE DUST OF YOUR COMPETORS, ADVERTISE TODAY!!!
Addresses are extracted daily by 6 of our computers, which run 24
hours a day 7 days a week, scanning the net for new addresses.
Estimated 60-80,000 addresses extracted daily. They are fresh!
Over 40 million addresses on file. More computers with cable
modems are being added due to increasing responses.
All ads can not be more than 2 pages (50 lines) and 68 characters
wide. More than 50 lines, there will be an additional charge of $50
per page or 25 lines or less.
No porn, no foul language or sex relative material will be
sent.
->->->->->->->-> TARGETED MAILINGS <-<-<-<-<-<-<-<-
->-> $150 per 50,000 addresses extracted or less
->-> This price includes mailing of material
->-> Allow 3-7 days for extracting of addresses
->-> Mailings to US only (general addresses) $500 per 500,000
->-> Many targeted list on file
->-> Purchased target list $150 per 50,000 or less
If you have your own list, we will email your list for $200
per 500,000 or less. All list must be in text format and
each address must be on individual lines.
***Mailings are done all at once, not broken down into
multiple mailings.***
Addrsses are processed against our remove list of those
individuals who have asked to be elimated from any of our
mailings. All mailings are de-duped for duplicates. No
.mil, .gov or .edu included.
There are no lower prices on the net. Your mailing
can be done in a matter of hours. Turn around time for
mailings range from 3 to 5 days. This time varies according
to the demand of mailings.
For the fastest service, cheapest prices and cleanest
mailings call our processing and new accounts office
at 904-282-0945, Monday - Friday 9 - 5 EST (English only).
If the line is busy or no repsonse, please keep trying,
as bulk mailing is growing fast and we are busy responding
to those who are interested in advertsing on the internet.
We DO want to work with you to advertise your product/service.
No collect calls.
Our computers are working 24 hours a day to better serve YOU!
To have your name removed, call our processing office.
Any negative responses will be dealt with accordingly.
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From pa@tekla.fi Mon Dec 8 12:23:46 1997
From: pa@tekla.fi (Harri Pasanen)
Date: Mon, 8 Dec 1997 14:23:46 +0200
Subject: [MATRIX-SIG] NumPy with Python 1.5 for Win32
Message-ID: <9712081223.AA26012@tahma.tekla.fi>
How's the progress on getting NumPy to work on Win32?
I took a quick stab at compiling it myself, but it failed.
Is somebody working on it actively, with an expected date when it will
be done? If not, I will do it, I just don't want to do duplicated work.
Harri
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From caceres@myna.com Sun Dec 14 22:00:46 1997
From: caceres@myna.com (CACERES)
Date: Sun, 14 Dec 1997 17:00:46 -0500
Subject: [MATRIX-SIG] openGL
Message-ID: <3494570E.387D265C@myna.com>
can i open gl in win95???
i have a mystique220 .4meg!!
thanx!
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From da@skivs.ski.org Mon Dec 15 04:22:42 1997
From: da@skivs.ski.org (David Ascher)
Date: Sun, 14 Dec 1997 20:22:42 -0800 (PST)
Subject: [MATRIX-SIG] openGL
In-Reply-To: <3494570E.387D265C@myna.com>
Message-ID: <Pine.SUN.3.96.971214202157.17439A-100000@skivs.ski.org>
On Sun, 14 Dec 1997, CACERES wrote:
> can i open gl in win95???
>
> i have a mystique220 .4meg!!
Yes. See http://starship.skyport.net/~da/PyOpenGL.html.
Let me know if you have problems.
I'll make a version for Python 1.5 once 1.5final is released.
--da
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From Jack.Jansen@cwi.nl Mon Dec 15 17:38:10 1997
From: Jack.Jansen@cwi.nl (Jack Jansen)
Date: Mon, 15 Dec 1997 18:38:10 +0100
Subject: [MATRIX-SIG] NumPy and 1.5, and NumPy and MacPython
Message-ID: <UTC199712151738.SAA05055.jack@snelboot.cwi.nl>
Folks, two remarks,
- as you are probably already aware the NumPy test suite fails inder
1.5 because of the new exception handling.
- The download page page for NumPy mentions the windows- and
unix-distributions, but not the mac distribution. As of 1.5 NumPy is
included in the standard MacPython distribution as an optional
install option, and NumPy for MacPython 1.4 is available through
http://www.cwi.nl/~jack/macpython.
Is a new NumPy release planned for the near future (I'm using 1.0b3 at
the moment, which still seems to be the latest release)?
Please cc me on any replies since I'm not a member of the matrix-sig,
--
Jack Jansen | ++++ stop the execution of Mumia Abu-Jamal ++++
Jack.Jansen@cwi.nl | ++++ if you agree copy these lines to your sig ++++
http://www.cwi.nl/~jack | see http://www.xs4all.nl/~tank/spg-l/sigaction.htm
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From acannon@geog.ubc.ca Tue Dec 16 06:47:44 1997
From: acannon@geog.ubc.ca (Alex Cannon)
Date: Mon, 15 Dec 1997 22:47:44 -0800 (PST)
Subject: [MATRIX-SIG] NumPy and 1.5, and NumPy and MacPython
In-Reply-To: <UTC199712151738.SAA05055.jack@snelboot.cwi.nl>
Message-ID: <Pine.GSO.3.95.971215223259.1458A-100000@shrike>
On Mon, 15 Dec 1997, Jack Jansen wrote:
> Folks, two remarks,
> - as you are probably already aware the NumPy test suite fails inder
> 1.5 because of the new exception handling.
> - The download page page for NumPy mentions the windows- and
> unix-distributions, but not the mac distribution. As of 1.5 NumPy is
> included in the standard MacPython distribution as an optional
> install option, and NumPy for MacPython 1.4 is available through
> http://www.cwi.nl/~jack/macpython.
Has anyone successfully gotten 1.4 (CFM68K) with NumPy up and running?
The core Numeric functions seem to work fine but certain functions in
other modules cause the interpreter to die (RandomArray.random, FFT.fft
etc); others seem to work fine though (RandomArray.randint for instance).
Any ideas?
A quick search for the 1.5b2 Mac version came up empty; I'm assuming the
betas aren't available for the Mac?
Alex
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From Jack.Jansen@cwi.nl Tue Dec 16 10:31:13 1997
From: Jack.Jansen@cwi.nl (Jack Jansen)
Date: Tue, 16 Dec 1997 11:31:13 +0100
Subject: [MATRIX-SIG] NumPy and 1.5, and NumPy and MacPython
In-Reply-To: Message by Alex Cannon <acannon@geog.ubc.ca> ,
Mon, 15 Dec 1997 22:47:44 -0800 (PST) , <Pine.GSO.3.95.971215223259.1458A-100000@shrike>
Message-ID: <UTC199712161031.LAA13689.jack@snelboot.cwi.nl>
> A quick search for the 1.5b2 Mac version came up empty; I'm assuming the
> betas aren't available for the Mac?
Expect it shortly, I'm working on it. It'll contain NumPy, but the cfm68k
version will again be only lightly tested, if at all, so feedback is very
welcome. I'll announce it to macpython-sig and comp.lang.python/python-list.
--
Jack Jansen | ++++ stop the execution of Mumia Abu-Jamal ++++
Jack.Jansen@cwi.nl | ++++ if you agree copy these lines to your sig ++++
http://www.cwi.nl/~jack | see http://www.xs4all.nl/~tank/spg-l/sigaction.htm
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From busby@icf.llnl.gov Tue Dec 16 19:31:55 1997
From: busby@icf.llnl.gov (L. Busby)
Date: Tue, 16 Dec 97 11:31:55 PST
Subject: [MATRIX-SIG] NumPy and 1.5, and NumPy and MacPython
Message-ID: <9712161931.AA11313@icf.llnl.gov.llnl.gov>
[Jack Jansen <Jack.Jansen@cwi.nl> wrote ]
>Folks, two remarks,
>- as you are probably already aware the NumPy test suite fails inder
> 1.5 because of the new exception handling.
>- The download page page for NumPy mentions the windows- and
> unix-distributions, but not the mac distribution. As of 1.5 NumPy is
> included in the standard MacPython distribution as an optional
> install option, and NumPy for MacPython 1.4 is available through
> http://www.cwi.nl/~jack/macpython.
>
>Is a new NumPy release planned for the near future (I'm using 1.0b3 at
>the moment, which still seems to be the latest release)?
[ "Paul F. Dubois" <dubois1@llnl.gov> replied ]
>We (LLNL) will release a 1.5-compatible NumPy when we have one and can
>compile it on Windows. We'll send it to you first so you can see if it works
>on the Mac.
Let's (matrix-sig) talk a little more about the release process for
NumPy. The most direct answer I can give now to Jack's question is
that I want to put out a release of NumPy as soon as we practically can
after Python-1.5 is officially released. Here's some questions that
occur to me:
1) How do we ``release'' NumPy?
a) Which site has the ``master'' copy?
b) Which are the mirror sites?
c) Who takes care of mirroring?
d) How do we accomodate our 3 primary architectures in the release?
+ UNIX
+ WINDOWS
+ MACINTOSH
2) What goes into NumPy-1.0?
a) Compatible with Python-1.5 (and 1.4, ...).
b) What else?
I would appreciate all comments to help me structure the release to be
simple and accessible. I'll use your comments and requests to develop
a release procedure, then post it for your review.
Thanks - Lee Busby
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From pa@tekla.fi Wed Dec 17 07:49:01 1997
From: pa@tekla.fi (Harri Pasanen)
Date: Wed, 17 Dec 1997 09:49:01 +0200
Subject: [MATRIX-SIG] NumPy and 1.5, and NumPy and MacPython
In-Reply-To: <9712161931.AA11313@icf.llnl.gov.llnl.gov>
References: <9712161931.AA11313@icf.llnl.gov.llnl.gov>
Message-ID: <9712170749.AA23734@tahma.tekla.fi>
L. Busby writes:
> [Jack Jansen <Jack.Jansen@cwi.nl> wrote ]
> >Folks, two remarks,
> >- as you are probably already aware the NumPy test suite fails inder
> > 1.5 because of the new exception handling.
Should be noted, that it is the test_items.py that needs fixing, not
NumPy itself.
> >- The download page page for NumPy mentions the windows- and
> > unix-distributions, but not the mac distribution. As of 1.5 NumPy is
> > included in the standard MacPython distribution as an optional
> > install option, and NumPy for MacPython 1.4 is available through
> > http://www.cwi.nl/~jack/macpython.
> >
> >Is a new NumPy release planned for the near future (I'm using 1.0b3 at
> >the moment, which still seems to be the latest release)?
>
> [ "Paul F. Dubois" <dubois1@llnl.gov> replied ]
> >We (LLNL) will release a 1.5-compatible NumPy when we have one and can
> >compile it on Windows. We'll send it to you first so you can see if it works
> >on the Mac.
>
> Let's (matrix-sig) talk a little more about the release process for
> NumPy. The most direct answer I can give now to Jack's question is
> that I want to put out a release of NumPy as soon as we practically can
> after Python-1.5 is officially released. Here's some questions that
> occur to me:
>
> 1) How do we ``release'' NumPy?
> a) Which site has the ``master'' copy?
I think www.python.org is the logical place for the master copy, it gets
automatically mirrored.
But about any place is fine, provided it is prominently displayed and
has links from the web pages at www.python.org.
> b) Which are the mirror sites?
> c) Who takes care of mirroring?
> d) How do we accomodate our 3 primary architectures in the release?
> + UNIX
> + WINDOWS
> + MACINTOSH
>
> 2) What goes into NumPy-1.0?
> a) Compatible with Python-1.5 (and 1.4, ...).
> b) What else?
>
> I would appreciate all comments to help me structure the release to be
> simple and accessible. I'll use your comments and requests to develop
> a release procedure, then post it for your review.
>
> Thanks - Lee Busby
>
I have a compiled NumPy on NT for Python 1.5. It is based on Konrad
Hinsen's NumPy release, which is a bit modified presumably for easier
installation on Unix systems. As it is, Konrad's release modifies the
C API slightly, so that old NumPy.dll is no longer compatible, and is
in fact replaced by _numpy.pyd, which is an internally imported python
extension module. I haven't done much with it so far, but I've
verified that the test_all.py script passes without errors (meaning I
have fixed test_items.py to comply with Python 1.5). It should be
possible to do test_*.py scripts that run properly under both Python
1.4 and 1.5, but I did not bother.
Note that the sources could use some cleanup, they compile with tons
of warnings, but without errors. Also I'm not sure how widely people
use the C API directly, and if Konrad's changes are acceptable to
everyone. I've made some changes to sources (mainly .def files), but
the distribution should still compile without added problems on Unix.
I've packaged the binaries to an installer so that it drops in place
on top of Python 1.5 binary distribution, much as the old numpy did.
If anyone would like to try this, drop me a line. Also, I can provide
the source + project workspace if there is interest. I don't know
what version to stamp on those, and I'm not looking to be a
maintainer, so I'll gladly donate these to the 'central authority',
which shall be LLNL if I've understood correctly?
Thanks,
Harri
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From hinsen@ibs.ibs.fr Wed Dec 17 10:47:25 1997
From: hinsen@ibs.ibs.fr (Konrad Hinsen)
Date: Wed, 17 Dec 1997 11:47:25 +0100
Subject: [MATRIX-SIG] NumPy and 1.5, and NumPy and MacPython
In-Reply-To: <9712170749.AA23734@tahma.tekla.fi> (message from Harri Pasanen
on Wed, 17 Dec 1997 09:49:01 +0200)
Message-ID: <199712171047.LAA29556@lmspc1.ibs.fr>
> > 1) How do we ``release'' NumPy?
> > a) Which site has the ``master'' copy?
>
> I think www.python.org is the logical place for the master copy, it gets
> automatically mirrored.
I agree. Of course LLNL's site would be another obvious choice, but
at least for me it's practically inaccessible (2 bytes/second!).
> > d) How do we accomodate our 3 primary architectures in the release?
> > + UNIX
> > + WINDOWS
> > + MACINTOSH
I'd prefer a general source distribution for all platforms, plus
binaries for Windows/Intel and the Mac. That should cover most
needs.
> > 2) What goes into NumPy-1.0?
> > a) Compatible with Python-1.5 (and 1.4, ...).
> > b) What else?
In the long run, I'd like to see a package structure for NumPy.
Maybe that's not doable sufficiently fast, but at least a new
release should be structured to take into account an eventual
package version.
> I have a compiled NumPy on NT for Python 1.5. It is based on Konrad
> Hinsen's NumPy release, which is a bit modified presumably for easier
> installation on Unix systems. As it is, Konrad's release modifies the
> C API slightly, so that old NumPy.dll is no longer compatible, and is
The C API is not modified at the source code level, but to the
linker it looks very different. Nevertheless, modules compiled
with the "traditional" arrayobject.h should continue to work
with a DLL made from my modified version, but inversely it wouldn't
work.
> Note that the sources could use some cleanup, they compile with tons
> of warnings, but without errors. Also I'm not sure how widely people
> use the C API directly, and if Konrad's changes are acceptable to
> everyone. I've made some changes to sources (mainly .def files), but
> the distribution should still compile without added problems on Unix.
I don't know whether my API modification is necessary or advantageous
under Windows. If the old one is better, a simple system-dependent
modification in arrayobject.h would be sufficient. For Unix, the
API modifications are essential to permit uniform installation of
other C extension modules on all machines.
--
-------------------------------------------------------------------------------
Konrad Hinsen | E-Mail: hinsen@ibs.ibs.fr
Laboratoire de Dynamique Moleculaire | Tel.: +33-4.76.88.99.28
Institut de Biologie Structurale | Fax: +33-4.76.88.54.94
41, av. des Martyrs | Deutsch/Esperanto/English/
38027 Grenoble Cedex 1, France | Nederlands/Francais
-------------------------------------------------------------------------------
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From Jack.Jansen@cwi.nl Wed Dec 17 13:02:22 1997
From: Jack.Jansen@cwi.nl (Jack Jansen)
Date: Wed, 17 Dec 1997 14:02:22 +0100
Subject: [MATRIX-SIG] NumPy and 1.5, and NumPy and MacPython
In-Reply-To: Message by busby@icf.llnl.gov (L. Busby) ,
Tue, 16 Dec 97 11:31:55 PST , <9712161931.AA11313@icf.llnl.gov.llnl.gov>
Message-ID: <UTC199712171302.OAA26639.jack@snelboot.cwi.nl>
Recently, busby@icf.llnl.gov (L. Busby) said:
>
> Let's (matrix-sig) talk a little more about the release process for
> NumPy. The most direct answer I can give now to Jack's question is
> that I want to put out a release of NumPy as soon as we practically can
> after Python-1.5 is officially released. Here's some questions that
> occur to me:
>
> 1) How do we ``release'' NumPy?
> a) Which site has the ``master'' copy?
> b) Which are the mirror sites?
> c) Who takes care of mirroring?
> d) How do we accomodate our 3 primary architectures in the release?
> + UNIX
> + WINDOWS
> + MACINTOSH
I've included NumPy as an optional part in the core Python
distribution after a lot of demand for it in the python-mac SIG: NumPy
and PIL are used so often that people felt it would be handy if they
were included. Also, it would point newcomers immedeately at two large
and hady extension packages.
There is however the question of non-synchronicity between releases of
MacPython and NumPy. First, I would like to release MacPython 1.5 as
soon as possible after Guido releases Unix-Python 1.5, so I don't want
to delay too long waiting for NumPy. Second, if new releases of NumPy
come available between major releases of MacPython there's a 99%
chance that mac-users will never notice this (and possibly continue
struggling against bugs that have long been solved in the new
release).
I am unsure what the right answer is. I am tempted to say that the
advantage of immedeately giving people installing Python a number of
fun/useful packages is the more important one, but I'd like to hear
what other people think.
--
Jack Jansen | ++++ stop the execution of Mumia Abu-Jamal ++++
Jack.Jansen@cwi.nl | ++++ if you agree copy these lines to your sig ++++
http://www.cwi.nl/~jack | see http://www.xs4all.nl/~tank/spg-l/sigaction.htm
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From npat@ariadne.di.uoa.gr Wed Dec 17 21:38:24 1997
From: npat@ariadne.di.uoa.gr (Nick Patavalis)
Date: Wed, 17 Dec 1997 23:38:24 +0200 (EET)
Subject: [MATRIX-SIG] Float32 and type conversions.
Message-ID: <199712172138.XAA29306@ariadne.di.uoa.gr>
Hi,
I'm trying to do something like this and it fails:
a = ones([5,5], Float32)
a[2:4] = a[2:4] * 0.5
Python prints the error message:
TypeError: Array can not be safely cast to required type
Of course I understaind what goes wrong! Is there any way
around this problem? Is there any way to make python or NumPy
tread 0.5 (or any other number for that matter) as a float and
not as a double? I'm asking this because it is really important
for my application to use 'float' and not 'double' arrays.
(It interfaces to a lot of C-code, the arrays are _huge_, and
so on...)
Thanks in advance.
Nick Patavalis
(npat@ariadne.di.uoa.gr)
University of Athens
Athens, Dec 17 1997
P.S. Please also send your answers to my personal address, since I not
a member of the list (yet!)
I'm using NumPy v1.0b3 and Python v1.4
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From hinsen@ibs.ibs.fr Thu Dec 18 10:44:28 1997
From: hinsen@ibs.ibs.fr (Konrad Hinsen)
Date: Thu, 18 Dec 1997 11:44:28 +0100
Subject: [MATRIX-SIG] Float32 and type conversions.
In-Reply-To: <199712172138.XAA29306@ariadne.di.uoa.gr> (message from Nick
Patavalis on Wed, 17 Dec 1997 23:38:24 +0200 (EET))
Message-ID: <199712181044.LAA02530@lmspc1.ibs.fr>
> I'm trying to do something like this and it fails:
>
> a = ones([5,5], Float32)
> a[2:4] = a[2:4] * 0.5
>
> Python prints the error message:
>
> TypeError: Array can not be safely cast to required type
>
> Of course I understaind what goes wrong! Is there any way
> around this problem? Is there any way to make python or NumPy
> tread 0.5 (or any other number for that matter) as a float and
> not as a double? I'm asking this because it is really important
Plain numbers don't exist as float in Python. You have two choices:
1) Put the factor 0.5 into an array of shape (1,) of type float:
a = ones([5,5], Float32)
half = array([0.5], Float32)
a[2:4] = a[2:4] * half
2) Cast the result of the multiplication back to float:
a = ones([5,5], Float32)
a[2:4] = (a[2:4] * 0.5).astype(Float32)
The second method produces a double intermediary, which may be a problem
if your arrays are very big.
--
-------------------------------------------------------------------------------
Konrad Hinsen | E-Mail: hinsen@ibs.ibs.fr
Laboratoire de Dynamique Moleculaire | Tel.: +33-4.76.88.99.28
Institut de Biologie Structurale | Fax: +33-4.76.88.54.94
41, av. des Martyrs | Deutsch/Esperanto/English/
38027 Grenoble Cedex 1, France | Nederlands/Francais
-------------------------------------------------------------------------------
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From jhauser@ifm.uni-kiel.de Thu Dec 18 13:31:03 1997
From: jhauser@ifm.uni-kiel.de (Janko Hauser)
Date: Thu, 18 Dec 1997 14:31:03 +0100 (CET)
Subject: [MATRIX-SIG] Final NumPy 1.0
Message-ID: <m0xig2Z-000szJC@lisboa.ifm.uni-kiel.de>
Hello, I welcome the decision to build to a final NumPy ver. 1.0. I
would like to raise some questions regarding this.
I think this package will be something for a pythoneer, who want's to
build some numbercrunching programs. We are currently not on the way
to a real integrated development enviroment for scientific
computation. But I also think, something like this is very doable. But
there is the question, if this is wanted. This is especially important
for the documentation of the hole package (not the different
functions). So, should somebody without knowlege of python be able to
grab a tar ball and install a working NumPy, or do we focus currently
only a pythoneer who wants to install an extension?
Do we help people, if we include more code, which are wrappers over
the more general functions (sum, cumsum vs. add.reduce)?
For the further extensibility we need e repository with more
code. It's important to have the code at one place and don't have
links to various locations, so a homogeneous representation to the
public is possible.
I think there should be included some code to read and write some
ascii-data. Everybody has some data lying around, with which he/she
wants to work.
And Timothy Hochberg has perhaps the currently better implementation
of a UserArray class at his starship side.
And at the end, I thank you for your efforts to do this work.
OK, that's it :-)
__Janko
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From owner-matrix-sig@python.org Thu Dec 18 17:13:56 1997
From: owner-matrix-sig@python.org (owner-matrix-sig@python.org)
Date: Thu, 18 Dec 1997 12:13:56 -0500 (EST)
Subject: No subject
Message-ID: <199712181713.MAA25040@python.org>
From jhauser@ifm.uni-kiel.de Thu Dec 18 18:22:20 1997
From: jhauser@ifm.uni-kiel.de (Janko Hauser)
Date: Thu, 18 Dec 1997 19:22:20 +0100 (CET)
Subject: [MATRIX-SIG] Final NumPy 1.0
In-Reply-To: <t9d8iuwpe6.fsf@uinpluxb.npl.uiuc.edu>
References: <m0xig2Z-000szJC@lisboa.ifm.uni-kiel.de> <t9d8iuwpe6.fsf@uinpluxb.npl.uiuc.edu>
Message-ID: <m0xikaS-000ssLC@lisboa.ifm.uni-kiel.de>
I understand that sys.input is something to facilitate the reading
from standard in to build command line tools and scripts which get
used in pipes.
I use currently this code, which unfortunatly needs sscanfmodule. But
it's the fastes way to read ``trusted'' data (No testing for to big,
equally formated)
def read_table(fname,header=0):
"""read_table(filename,header=number_of_header_lines)
Reads from filename number_of_fields numbers per line
to end of file. If number of num_fields per line doesn't
match, skip line"""
from sscanf import sscanf
import string
try:
fp=open(fname,'r')
except:
print "Can't open file or wrong filename in function read_table"
return
scanline='%f'
lines = fp.readlines()
n=len(string.split(lines[header]))
new_table=ones((len(lines)-header,n),typecode = Float)
while n-1:
n=n-1
scanline=scanline+' %f'
for item in xrange(len(lines)-header):
a=array(sscanf(lines[item],scanline))
new_table[item,:] = a
return(new_table)
Has someone something faster, or really sophisticated?
(sscanf is much faster with the conversion of string to float, I
tested this only with 1.4 not 1.5, where the string handling should be
faster)
__Janko
Mike Miller writes:
> >>>>> "janko" == Janko Hauser <jhauser@ifm.uni-kiel.de> writes:
> > I think there should be included some code to read and
> > write some ascii-data. Everybody has some data lying
> > around, with which he/she wants to work.
>
> There was a discussion about this on comp.lang.python a few weeks
> ago. The subject was "Proposal: sys.input()". Some code was
> posted there...
>
> Mike
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From pa@tekla.fi Thu Dec 18 22:58:59 1997
From: pa@tekla.fi (Harri Pasanen)
Date: 19 Dec 1997 00:58:59 +0200
Subject: [MATRIX-SIG] ANNOUNCE: unofficial NumPy binaries for Python 1.5 on NT
Message-ID: <cs7m92z2a4.fsf@tahma.tekla.fi>
The following message is a courtesy copy of an article
that has been posted to comp.lang.python as well.
For the impatient
http://starship.skyport.net/crew/harri_pasanen/NumPy10-15.exe
Drops on place on top of Python 1.5b2, (probably works on 1.5b1 as 1.5
final as well).
Regards,
Harri.Pasanen@iki.fi
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From pa@tekla.fi Thu Dec 18 22:58:59 1997
From: pa@tekla.fi (Harri Pasanen)
Date: 19 Dec 1997 00:58:59 +0200
Subject: [MATRIX-SIG] ANNOUNCE: unofficial NumPy binaries for Python 1.5 on NT
Message-ID: <cs7m92z2a4.fsf@tahma.tekla.fi>
The following message is a courtesy copy of an article
that has been posted to comp.lang.python as well.
For the impatient
http://starship.skyport.net/crew/harri_pasanen/NumPy10-15.exe
Drops on place on top of Python 1.5b2, (probably works on 1.5b1 as 1.5
final as well).
Regards,
Harri.Pasanen@iki.fi
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From pa@tekla.fi Fri Dec 19 08:44:57 1997
From: pa@tekla.fi (Harri Pasanen)
Date: 19 Dec 1997 10:44:57 +0200
Subject: [MATRIX-SIG] Re: ANNOUNCE: unofficial NumPy binaries for Python 1.5 on NT
In-Reply-To: Harri Pasanen's message of 19 Dec 1997 00:58:59 +0200
References: <cs7m92z2a4.fsf@tahma.tekla.fi>
Message-ID: <csk9d1hgc6.fsf@tahma.tekla.fi>
The following message is a courtesy copy of an article
that has been posted to comp.lang.python as well.
Harri Pasanen <pa@tekla.fi> writes:
> For the impatient
>
> http://starship.skyport.net/crew/harri_pasanen/NumPy10-15.exe
>
> Drops on place on top of Python 1.5b2, (probably works on 1.5b1 as 1.5
> final as well).
>
Oops,...
Installer puts the wrong path at
HKEY_LOCAL_MACHINE\SOFTWARE\Python\PythonCore\1.5.0\PythonPath\NumPy
You have three choices:
1) Use regedit to fix it to point to NumPy under your Python tree.
2) Uninstall it from control-panel -> Add/Remove programs.
3) Uninstall and wait for a fixed version, ETA 24 hours, probably
sooner.
Sorry for the confusion, thanks for the prompt feedback.
Harri
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From jhauser@ifm.uni-kiel.de Sun Dec 21 16:28:08 1997
From: jhauser@ifm.uni-kiel.de (Janko Hauser)
Date: Sun, 21 Dec 1997 17:28:08 +0100 (CET)
Subject: [MATRIX-SIG] Get array elements in c?
Message-ID: <m0xjoEa-000ssvC@lisboa.ifm.uni-kiel.de>
I need to copy an array to this struct:
typedef float Matrix[4][4];
For this I try to write a function:
#define A_DATA(a) (((PyArrayObject *)a)->data)
#define GET_ARR(ap,op,type,dim) \
Py_Try(ap=(PyArrayObject
*)PyArray_ContiguousFromObject(op,type,dim,dim))
static void cp_to_Matrix(PyObject *arg, Matrix mat) {
PyArrayObject *arr;
int i, j;
double *data;
GET_ARR(arr, arg, PyArray_FLOAT, 2);
data = (double *)A_DATA(arr);
for (i=0; i<4; i++){
for (j=0; j<4; j++){
mat[i][j] = data[i][j]; /* this is wrong. right? */
}
}
return;
}
Must I address data here linearly? Or is there another way to get the
NumPyArray into the Matrix?
Something similar for
typedef float Vector[4];
does function, with some warnings.
Sorry for my C ignorance (that's why I use Python),
__Janko
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From bsd@scripps.edu Tue Dec 23 20:01:44 1997
From: bsd@scripps.edu (Bruce Duncan)
Date: Tue, 23 Dec 1997 12:01:44 -0800 (PST)
Subject: [MATRIX-SIG] ** problems with Float32
Message-ID: <199712232001.MAA03491@sheba.scripps.edu>
Greetings,
I'm not sure if I'm using the most up to date version of Numeric but...
num_array ** 2 doesn't work if num_array is Float32 (Float64 works fine).
Has this been fixed in a more recent release?
-bsd-
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From robert@directed.edu Wed Dec 24 17:47:17 1997
From: robert@directed.edu (robert@directed.edu)
Date: Wed, 24 Dec 1997 12:47:17 -0500
Subject: [MATRIX-SIG] P R E S S R E L E A S E
Message-ID: <eaep.3.0.reg.RobD9U.35788.5507513889@cbbc.cbbc.ns.ca>
DirectED - The New Way to Learn!
DirectED is a distance education school that will revolutionize the way people will think about
education in the future. Students can now study in the comfort of their own home to earn their
diploma. What is unique about DirectED is that it is completely Internet based. Students attend
classes in a virtual campus that has an employment center, library, cafeteria, computer lab, book
store, faculty lounge, business office, student residence, and virtual classrooms. You can visit the
campus at http://www.directed.edu.
All DirectED Students will receive:
Acer Multimedia Computer
Lexmark Color Printer
Windows 95
Office 97 Professional
Books
Prepaid Internet Connection for 1 Year
Access to DirectED Web Site
Toll Free Support for 1 Year
DirectED programs are certified and registered with Department of Training and Advanced
Education.
DirectED is a member of the Better Business Bureau.
DirectED partners include the Toronto Dominion Bank, MTS Sympatico, Acer, Course Technology,
Microsoft, and Computer Associates.
To find out more about THE NEW WAY TO LEARN please visit DirectED's business office at
http://www.directed.edu/bus.html
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From hinsen@ibs.ibs.fr Fri Dec 26 08:49:12 1997
From: hinsen@ibs.ibs.fr (Konrad Hinsen)
Date: Fri, 26 Dec 1997 09:49:12 +0100
Subject: [MATRIX-SIG] Get array elements in c?
In-Reply-To: <m0xjoEa-000ssvC@lisboa.ifm.uni-kiel.de> (message from Janko
Hauser on Sun, 21 Dec 1997 17:28:08 +0100 (CET))
Message-ID: <199712260849.JAA09449@lmspc1.ibs.fr>
> I need to copy an array to this struct:
>
> typedef float Matrix[4][4];
The data layout of this type is just 16 floats, as for a NumPy array
of Float32. As always, there are several ways to do what you want.
Perhaps the clearest is to cast array->data to Matrix and copy as
if you had two Matrix variables. Or just copy 16 floats linearly.
--
-------------------------------------------------------------------------------
Konrad Hinsen | E-Mail: hinsen@ibs.ibs.fr
Laboratoire de Dynamique Moleculaire | Tel.: +33-4.76.88.99.28
Institut de Biologie Structurale | Fax: +33-4.76.88.54.94
41, av. des Martyrs | Deutsch/Esperanto/English/
38027 Grenoble Cedex 1, France | Nederlands/Francais
-------------------------------------------------------------------------------
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________
From hinsen@ibs.ibs.fr Fri Dec 26 08:51:34 1997
From: hinsen@ibs.ibs.fr (Konrad Hinsen)
Date: Fri, 26 Dec 1997 09:51:34 +0100
Subject: [MATRIX-SIG] ** problems with Float32
In-Reply-To: <199712232001.MAA03491@sheba.scripps.edu> (message from Bruce
Duncan on Tue, 23 Dec 1997 12:01:44 -0800 (PST))
Message-ID: <199712260851.JAA09460@lmspc1.ibs.fr>
> I'm not sure if I'm using the most up to date version of Numeric but...
> num_array ** 2 doesn't work if num_array is Float32 (Float64 works fine).
> Has this been fixed in a more recent release?
It works for me, and I can't remember that it didn't work. However,
it is always worth checking that you have the most recent version,
since there were important changes.
--
-------------------------------------------------------------------------------
Konrad Hinsen | E-Mail: hinsen@ibs.ibs.fr
Laboratoire de Dynamique Moleculaire | Tel.: +33-4.76.88.99.28
Institut de Biologie Structurale | Fax: +33-4.76.88.54.94
41, av. des Martyrs | Deutsch/Esperanto/English/
38027 Grenoble Cedex 1, France | Nederlands/Francais
-------------------------------------------------------------------------------
_______________
MATRIX-SIG - SIG on Matrix Math for Python
send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org
_______________