headers
Gerard Vermeulen | 2 Oct 2005 21:38
Picon
Favicon

Re: Release of SciPy Core 0.4 (Beta)

On Fri, 30 Sep 2005 00:59:22 -0600
Travis Oliphant <oliphant <at> ee.byu.edu> wrote:

> 
> This is to announce the release of SciPy Core 0.4.X (beta)
> 
> It is available at sourceforge which you can access by going to
> 
> http://numeric.scipy.org
> 
> Thanks to all who helped me with this release, especially
> 
> Robert Kern
> Pearu Peterson
> 
> Now, let's start getting this stable....
> 
> -Travis Oliphant
> 
I have found two problems:

(1) scipy_core-0.4.1 does not compile on this system without ATLAS,
    because there are missing files:
...
 creating build/temp.linux-i686-2.4/lapack_lite
compile options: '-Iscipy/base/include -Ibuild/src/scipy/base -Iscipy/base/src
-I/usr/include/python2.4 -c'
gcc: scipy/corelib/lapack_lite/lapack_litemodule.c
gcc: lapack_lite/dlapack_lite.c
gcc: lapack_lite/dlapack_lite.c: No such file or directory
(Continue reading)

Permalink | Reply | 4 comments |
headers
Eric Firing | 2 Oct 2005 02:19
Favicon
Gravatar

Re: Release of SciPy Core 0.4 (Beta)

Travis,

> This is to announce the release of SciPy Core 0.4.X (beta)

Wonderful progress--I can't thank you enough for the work you are doing 
on this.

I initially tried to build from the tarball, but ran into missing pieces:

compile options: '-Iscipy/base/include -Ibuild/src/scipy/base 
-Iscipy/base/src -I/usr/include/python2.4 -c'
gcc: lapack_lite/blas_lite.c
gcc: lapack_lite/blas_lite.c: No such file or directory
gcc: no input files

I then installed from the rpm (on a Mandriva 2005 LE system) with no 
problems, and proceeded to poke around.  It looks great!

> Now, let's start getting this stable....

Bug: in function_base.py, there are four places where "nx" should be 
"_nx"; see attached diff.

Also attached is a diff for arraymethods, to fix what look to me like 
erroneous docstrings for the reshape and resize methods.  The present 
docstrings indicate that reshape makes the change in place, and resize 
returns a new array, but in both cases the opposite is true--reshape 
returns a reshaped copy, and resize returns None.

Eric
(Continue reading)

Permalink | Reply | 0 comments |
headers
Tim Churches | 3 Oct 2005 06:49
Picon

Re: Release of SciPy Core 0.4 (Beta)

Travis Oliphant wrote:
> 
> This is to announce the release of SciPy Core 0.4.X (beta)
> 
> It is available at sourceforge which you can access by going to
> 
> http://numeric.scipy.org
> 
> Thanks to all who helped me with this release, especially
> 
> Robert Kern
> Pearu Peterson
> 
> Now, let's start getting this stable....

Great work, and we are much relieved that NumPy has a firm future
(albeit under a different name).

However, can I ask if there are plans to add Masked Arrays to SciPy
Core? If not, we'll have to stick with NumPy - missing values are
ubiquitous in the biological sciences and any package which can't handle
them isn't, to put it bluntly, of much use. We are happy to help test MA
for SciPy Core, and our project (www.netepi.org) has over a thousand
unit tests which give the basics of NumPy and MA a good work-out - so
if/when we convert to SciPy Core the unit tests will be helpful. However
we are not familiar enough with how SciPy Core differs from NumPy to
help with the porting, I'm afraid.

Tim C
(Continue reading)

Permalink | Reply | 14 comments |
headers
Tim Churches | 3 Oct 2005 07:51
Picon

Re: Release of SciPy Core 0.4 (Beta)

Paul F. Dubois wrote:
> I was going to see if I could port MA. It is pure Python so shouldn't be
> too hard.

That would eb a most welcome development. If you are in the mood for
some minor extensions to MA (you probably aren't, but just on the
off-chance...), then support for multiple masking values would be great.
In other words, instead of the mask just taking the values zero or one,
it can also take 2, 3, etc (within the constraints of the Int8 data
type), where any non-zero value means masked. That would allow the
reason why a value in an array is masked to be neatly encoded, not just
the fact that it is masked (missing). For example, in (human) survey
data, a scalar value (eg age at menopasue) may be missing because the
respondent did not meet the precondition(s) for the question (eg sex not
female), or because they have not yet undergone menopause, or because
they refuse to answer, or because they don't know or don't recall.
That's at least 4 different reasons why the value may be missing
(masked). But don't ask me the the mask value should be when you
multiply a masked-because-not-asked value by a
masked-because-refused-to-answer value. Perhaps a reasonable rule would
be max(mask0,mask1)?

Tim C

> 
> Tim Churches wrote:
> 
>> Travis Oliphant wrote:
>>
>>> This is to announce the release of SciPy Core 0.4.X (beta)
(Continue reading)

Permalink | Reply | 4 comments |
headers
Paul F. Dubois | 3 Oct 2005 07:16

Re: Release of SciPy Core 0.4 (Beta)

I was going to see if I could port MA. It is pure Python so shouldn't be 
too hard.

Tim Churches wrote:
> Travis Oliphant wrote:
> 
>>This is to announce the release of SciPy Core 0.4.X (beta)
>>
>>It is available at sourceforge which you can access by going to
>>
>>http://numeric.scipy.org
>>
>>Thanks to all who helped me with this release, especially
>>
>>Robert Kern
>>Pearu Peterson
>>
>>Now, let's start getting this stable....
> 
> 
> Great work, and we are much relieved that NumPy has a firm future
> (albeit under a different name).
> 
> However, can I ask if there are plans to add Masked Arrays to SciPy
> Core? If not, we'll have to stick with NumPy - missing values are
> ubiquitous in the biological sciences and any package which can't handle
> them isn't, to put it bluntly, of much use. We are happy to help test MA
> for SciPy Core, and our project (www.netepi.org) has over a thousand
> unit tests which give the basics of NumPy and MA a good work-out - so
> if/when we convert to SciPy Core the unit tests will be helpful. However
(Continue reading)

Permalink | Reply | 7 comments |
headers
Stephen Walton | 3 Oct 2005 17:23
Favicon

Re: [Numpy-discussion] Release of SciPy Core 0.4 (Beta)

Tim Churches wrote:

>That would eb a most welcome development. If you are in the mood for
>some minor extensions to MA (you probably aren't, but just on the
>off-chance...), then support for multiple masking values would be great.
>  
>
This sounds like an overly complicated addition to MA.  If I may be so 
bold, it seems to me that what you might want is your own object which 
maintains parallel arrays of values and "mask reasons," if you will, and 
generates a mask array accordingly
Permalink | Reply | 1 comment |
headers
Chris Barker | 3 Oct 2005 18:53
Picon
Favicon

Re: Release of SciPy Core 0.4 (Beta)

> Tim Churches wrote:
>>  missing values are
>> ubiquitous in the biological sciences and any package which can't handle
>> them isn't, to put it bluntly, of much use.

MA is great, but I wonder if many of the simple "missing value" use 
cases could be covered by robust handling of NaNs.

Which brings up the question: How does scipy_core handle NaN and the 
other IEEE special values? This was a major weakness in Numeric for me.

-Chris

--

-- 
Christopher Barker, Ph.D.
Oceanographer
                                     		
NOAA/OR&R/HAZMAT         (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chris.Barker <at> noaa.gov

-------------------------------------------------------
This SF.Net email is sponsored by:
Power Architecture Resource Center: Free content, downloads, discussions,
and more. http://solutions.newsforge.com/ibmarch.tmpl
Permalink | Reply | 1 comment |
headers
Eric Firing | 3 Oct 2005 21:46
Favicon
Gravatar

ma is in scipy_core

Paul, Tim,

Masked arrays *are* in scipy_core.

import scipy.base.ma as ma

In addition, a quick look indicates that NaN-handling is in good shape.

Eric

-------------------------------------------------------
This SF.Net email is sponsored by:
Power Architecture Resource Center: Free content, downloads, discussions,
and more. http://solutions.newsforge.com/ibmarch.tmpl
Permalink | Reply | 46 comments |
headers
Stephen Walton | 3 Oct 2005 22:02
Favicon

Re: [Numpy-discussion] Release of SciPy Core 0.4 (Beta)

Tim Churches wrote:

>However, can I ask if there are plans to add Masked Arrays to SciPy
>Core?
>
Eric Firing pointed out on the matplotlib mailing list that they're 
already there:

import scipy.base.ma as ma
Permalink | Reply | 3 comments |
headers
Chris Barker | 3 Oct 2005 22:48
Picon
Favicon

Re: Release of SciPy Core 0.4 (Beta)

Travis Oliphant wrote:
> Chris Barker wrote:
>> Which brings up the question: How does scipy_core handle NaN and the 
>> other IEEE special values? This was a major weakness in Numeric for me.
> 
> They can exist in your arrays and  isnan, isinf, isposinf, isneginf  
> detects them.

Wonderful! I'm really looking forward to this being stable.

thanks, Travis!

-CHB

--

-- 
Christopher Barker, Ph.D.
Oceanographer
                                     		
NOAA/OR&R/HAZMAT         (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chris.Barker <at> noaa.gov

-------------------------------------------------------
This SF.Net email is sponsored by:
Power Architecture Resource Center: Free content, downloads, discussions,
and more. http://solutions.newsforge.com/ibmarch.tmpl
Permalink | Reply | 0 comments |

Gmane