numpy-discussion — Discussion list for all users of Numerical Python

Re: [Numpy-discussion] some work on arpack

[Keith G. <kwg...@gm...>]

On 10/22/06, Aric Hagberg <ha...@la...> wrote:
> On Sat, Oct 21, 2006 at 02:05:42PM -0700, Keith Goodman wrote:
> > Did you, or anybody else on the list, have any luck making a numpy
> > version of eigs?
>
> I made a start at an ARPACK wrapper, see
> http://projects.scipy.org/scipy/scipy/ticket/231
> and the short thread at scipy-dev
> http://thread.gmane.org/gmane.comp.python.scientific.devel/5166/focus=5175

That looks very promising!

Re: [Numpy-discussion] some work on arpack

[Albert S. <fu...@gm...>]

Argh.

> <snip>

> In addition to the wrapper there is a Python interface (and some tests).
> I don't know if the interface is like "eigs" - I don't use Matlab.

http://www.mathworks.de/access/helpdesk/help/techdoc/ref/index.html?/access/
helpdesk/help/techdoc/ref/eigs.html

should give you some idea of how MATLAB's eigs works.

Cheers,

Albert

Re: [Numpy-discussion] some work on arpack

[Albert S. <fu...@gm...>]

Hello all

> -----Original Message-----
> From: num...@li... [mailto:numpy-
> dis...@li...] On Behalf Of Aric Hagberg
> Sent: Sunday, October 22, 2006 4:09 PM
> To: Discussion of Numerical Python
> Subject: Re: [Numpy-discussion] some work on arpack
> 
> On Sat, Oct 21, 2006 at 02:05:42PM -0700, Keith Goodman wrote:
> > Did you, or anybody else on the list, have any luck making a numpy
> > version of eigs?
> 
> I made a start at an ARPACK wrapper, see
> http://projects.scipy.org/scipy/scipy/ticket/231
> and the short thread at scipy-dev
> http://thread.gmane.org/gmane.comp.python.scientific.devel/5166/focus=5175
> 
> In addition to the wrapper there is a Python interface (and some tests).
> I don't know if the interface is like "eigs" - I don't use Matlab.

If you're interested in making an interface that looks like MATLAB's, you
can take a peek at:


> 
> It will give you a few eigenvalues and eigenvectors for the standard
> eigenproblem (Ax=lx) for any type of A (symmetric/nonsymmetric,
> real/complex,
> single/double, sparse/nonsparse).
> 
> The generalized and shifted modes are not implemented.
> I need to find some time (or some help) to get it finished.
> 
> Regards,
> Aric
> 
> -------------------------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job
> easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> Numpy-discussion mailing list
> Num...@li...
> https://lists.sourceforge.net/lists/listinfo/numpy-discussion

Re: [Numpy-discussion] some work on arpack

[Aric H. <ha...@la...>]

On Sat, Oct 21, 2006 at 02:05:42PM -0700, Keith Goodman wrote:
> Did you, or anybody else on the list, have any luck making a numpy
> version of eigs?

I made a start at an ARPACK wrapper, see
http://projects.scipy.org/scipy/scipy/ticket/231
and the short thread at scipy-dev
http://thread.gmane.org/gmane.comp.python.scientific.devel/5166/focus=5175

In addition to the wrapper there is a Python interface (and some tests).  
I don't know if the interface is like "eigs" - I don't use Matlab.

It will give you a few eigenvalues and eigenvectors for the standard 
eigenproblem (Ax=lx) for any type of A (symmetric/nonsymmetric, real/complex,
single/double, sparse/nonsparse).  

The generalized and shifted modes are not implemented.
I need to find some time (or some help) to get it finished.

Regards,
Aric 

Re: [Numpy-discussion] Model and experiment fitting.

[<se...@pi...>]

Nadav Horesh napisał(a):
> 1. If at least one of your data sets to be interpulated is on a grid, 
> you can use numpy.ndimage.map function for fast interpolation for 2d (in fact for any dimensional) dataset.

I've already used a splines to interpolate a missing simulated points. 
That procedure works  great and is very fast. But I'll check the 
numpy.ndimage - I haven't used it, yet.


> 2. Isn't there an analytic expression to average the expectration values of SH over all possible orientations 
>  between B and the crystal axis? My experience shows that some analytic work can save 99% of simulation time.

Well, the simulations are already very fast. The time consumption is 
approximately ~0.3s for a single powder spectrum (2.8GHz Pentium D). The 
calculations are held by an external, very fine EPR spectra simulation 
tool. The author must have incorporated into it a lot of 
rationalizations, but this is a binary tool (unfortunately) and I do not 
know, what exactly sits inside of it... All I know, is that the 
orientations are represented by a grid (with an increment step tunable 
by a user). From library documentation: "After having computed the 
spectrum for a number of orientations specified, the simulation function 
interpolates these spectra for additional orientations before summing up 
all spectra." The interpolation is accomplish with a splines.




Thank you for your comment,

best regards

Sebastian

Re: [Numpy-discussion] some work on arpack

[Stefan v. d. W. <st...@su...>]

On Sun, Oct 22, 2006 at 09:27:39AM +0900, Bill Baxter wrote:
> On 10/22/06, Charles R Harris <cha...@gm...> wrote:
> > On 10/21/06, Bill Baxter <wb...@gm...> wrote:
> > > Here's something I've been wondering: is it ok to port Matlab
> > > functions over to python?  If so, then it's maybe an afternoon's wo=
rk
> > > to get eigs working, given eigs.m and python Arpack wrappers.
> > > [...]
> >
> > I haven't the faintest idea. What did the EULA say? What about a copy=
right?
> > This sounds like a job for a lawyer, but I would guess it would be a =
bad
> > idea unless the code is from a third party with a free license.
>=20
> Gee, there's an idea.  :-)
>=20
> ---- from License.txt ----
[...]
> -------------------------
>=20
> So. If you're a Matlab licensee it would be ok to use it in your own
> stuff, but putting it in Numpy is definitely out of the question.

A lot of the code in the Octave Forge is licensed under the BSD
license, so you can use that.  Unfortunately, the main Octave code is
released under the GPL.

Cheers
St=E9fan

[Numpy-discussion] =?iso-2022-jp?b?GyRCJDMkcyRLJEEkTxsoQg==?=

[<AG...@HO...>]

カツシン秋バージョン！新作秘法登場！
http://11487.com/shi/

Re: [Numpy-discussion] some work on arpack

[Bill B. <wb...@gm...>]

On 10/22/06, Charles R Harris <cha...@gm...> wrote:
> On 10/21/06, Bill Baxter <wb...@gm...> wrote:
> > Here's something I've been wondering: is it ok to port Matlab
> > functions over to python?  If so, then it's maybe an afternoon's work
> > to get eigs working, given eigs.m and python Arpack wrappers.
> > [...]
>
> I haven't the faintest idea. What did the EULA say? What about a copyright?
> This sounds like a job for a lawyer, but I would guess it would be a bad
> idea unless the code is from a third party with a free license.

Gee, there's an idea.  :-)

---- from License.txt ----
"Except as expressly provided by this Agreement, including the
attached Addendum, Licensee may not adapt, translate, or convert
"M-files", "MDL-files" or "P-code" contained in the Programs in
order to create software, a principal purpose of which is to
perform the same or similar functions as Programs licensed by
MathWorks or which is intended to replace any component of the
Programs.  The Licensee may not incorporate or use "M-files",
"P-code", source code, or any other part of the Programs in or
as part of another computer program without the consent of
MathWorks.  A Licensed User may modify pieces of MathWorks' code
for the Licensed User's own use.  A Licensed User may share such
modified code with others provided each recipient is also a
Licensed User for the original form of the code. "
-------------------------

So. If you're a Matlab licensee it would be ok to use it in your own
stuff, but putting it in Numpy is definitely out of the question.

--bb

Re: [Numpy-discussion] some work on arpack

[Charles R H. <cha...@gm...>]

On 10/21/06, Bill Baxter <wb...@gm...> wrote:
>
> Here's something I've been wondering: is it ok to port Matlab
> functions over to python?  If so, then it's maybe an afternoon's work
> to get eigs working, given eigs.m and python Arpack wrappers.
>
> I'm 99% sure it would *not* be kosher to copy-paste their code
> directly, but it's not a copy paste job because of the differences
> between Python and .m.  So it seems a grayer area to me.
>
> --bb


I haven't the faintest idea. What did the EULA say? What about a copyright?
This sounds like a job for a lawyer, but I would guess it would be a bad
idea unless the code is from a third party with a free license.

Chuck

Re: [Numpy-discussion] some work on arpack

[Bill B. <wb...@gm...>]

Here's something I've been wondering: is it ok to port Matlab
functions over to python?  If so, then it's maybe an afternoon's work
to get eigs working, given eigs.m and python Arpack wrappers.

I'm 99% sure it would *not* be kosher to copy-paste their code
directly, but it's not a copy paste job because of the differences
between Python and .m.  So it seems a grayer area to me.

--bb

On 10/22/06, Keith Goodman <kwg...@gm...> wrote:
> On 8/15/06, David Grant <dav...@gm...> wrote:
>
> > My idea is (if I have time) to write an eigs-like function in python
> > that will only perform a subset of what Matlab's eigs does for. It
> > will, for example, compute a certain number of eigenvalues and
> > eigenvectors for a real, sparse, symmetric matrix (the case I'm
> > interested in)... I hope that this subset-of-matlab's-eigs function
> > will not be too hard to write. Then more functionality can be added on
> > to eigs.py later... Does this make sense?
>
> Did you, or anybody else on the list, have any luck making a numpy
> version of eigs?
>
> -------------------------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> Numpy-discussion mailing list
> Num...@li...
> https://lists.sourceforge.net/lists/listinfo/numpy-discussion
>

Re: [Numpy-discussion] Numpy-scalars vs Numpy 0-d arrays: copy or not copy?

[Lisandro D. <da...@gm...>]

On 10/20/06, Travis Oliphant <oli...@ee...> wrote:
>
> How about this.  To get the i,j,k,l element
>
> a[i:i+1,j:j+1,k:k+1,l:l+1].squeeze()
>
> -Travis

I think all this can be condensed in a method call or similar
mehcanism, natively provided by ndarray type. Or should this be seen
as a special use case?

--=20
Lisandro Dalc=EDn
---------------
Centro Internacional de M=E9todos Computacionales en Ingenier=EDa (CIMEC)
Instituto de Desarrollo Tecnol=F3gico para la Industria Qu=EDmica (INTEC)
Consejo Nacional de Investigaciones Cient=EDficas y T=E9cnicas (CONICET)
PTLC - G=FCemes 3450, (3000) Santa Fe, Argentina
Tel/Fax: +54-(0)342-451.1594

Re: [Numpy-discussion] some work on arpack

[Keith G. <kwg...@gm...>]

On 8/15/06, David Grant <dav...@gm...> wrote:

> My idea is (if I have time) to write an eigs-like function in python
> that will only perform a subset of what Matlab's eigs does for. It
> will, for example, compute a certain number of eigenvalues and
> eigenvectors for a real, sparse, symmetric matrix (the case I'm
> interested in)... I hope that this subset-of-matlab's-eigs function
> will not be too hard to write. Then more functionality can be added on
> to eigs.py later... Does this make sense?

Did you, or anybody else on the list, have any luck making a numpy
version of eigs?

Re: [Numpy-discussion] Model and experiment fitting.

[Nadav H. <na...@vi...>]

1. If at least one of your data sets to be interpulated is on a grid, =
you can use numpy.ndimage.map function for fast interpolation for 2d (in =
fact for any dimensional) dataset.

2. Isn't there an analytic expression to average the expectration values =
of SH over all possible orientations between B and the crystal axis? My =
experience shows that some analytic work can save 99% of simulation =
time.

  Nadav

-----Original Message-----
From:	num...@li... on behalf of =
Sebastian Zurek
Sent:	Sat 21-Oct-06 15:41
To:	num...@li...
Cc:=09
Subject:	Re: [Numpy-discussion] Model and experiment fitting.

Robert Kern napisal(a):

> Your description is a bit vague.=20

Possibly by my weak English... I'll try to make myself clearer now.

Do you mean that you have some model function f
> that maps X values to Y values?
>=20
>    f(x) -> y
>=20

My model is quantum energy operator - spin hamiltonian (SH) with some
additional assumption about so called 'line shape', 'line widths',etc.

   It describes various electron interactions, visible in electron=20
paramagnetic resonance (EPR, ESR) experiment. The simplest SH can
be written in a form:
                        H =3D m B g S                   (1)
where m is a constant (bohr magneton), B is magnetic field (my=20
x-variable), g is so called 'zeeman matrix' and S is total spin angular
momentum operator.

Summing it all together: the simple model is parametrized by:
  - line shape,
  - line width,
  - zeeman matrix (3x3 diagonal matrix - the spatial dependence),
  - total spin S.

After SH (1) diagonalization one can obtain so called 'resonance fields' =

and  'resonance intensities'. After a convolution with appropriate  line =

shape function which is parametrized by the line width one can finally
get the simulated EPR spectrum (simDat=3D[[X1,...,Xn],[Y1,...,Yn]]).
This  is a roughly, schematic description, appropriate to EPR spectra of
monocrystals.

In my situation the problem is more sophisticated - I have=20
polycrystaline (powders) data, and to obtain a simulated EPR powder=20
spectrum I need to sum up the EPR spectra of monocrystals that come from =

many possible spatial orientations, and the resultant spectrum is an=20
envelope of all the monocrystals spectra.

There's no simple model function that maps X -> Y.


> If that is the case, is there some reason that you cannot run your =
simulation=20
> using the same X points as your experimental data?
>=20

I can only demand a X range and number of X values within the range,=20
there's no possibility to find the Y(X) for a specified X. These=20
limitations on one hand come from  the external program I'm using to=20
simulate the EPR spectra, on the other are a result of spatial averaging =

of EPR data for powders, where a lot of interpolations are involved.


> OTOH, is there some other independent variable (say Z) that *is* =
common between=20
> your experimental and simulated data?
>=20
>    f(z) -> (x, y)
>=20

This is probably the situation I'm in. These other variables are my=20
model parameters, namely: line shape-width, zeeman matrix... and they're
commen between the experiment and the simulation.


To make it clear.

I've already solved the problem by a simple linear interpolation of=20
simulated points within the narrow neighborhood of experimental data=20
point. The simulation points are uniformly distributed along the=20
X-range, with a density I'm able to tune. It all works quite well but=20
I'm founding it as a 'brute-force' method and I wonder, if there's any=20
more sophisticated and maybe already incorporated into any Python module =

method?

Anyway, it looks like it's impossible to compare two discrete 2D data=20
sets without any interpolations included... :]


A. M. Archibald has proposed spline fitting, which I'll try. I'll also=20
look at the Numerical Recipes discussion he has proposed.


Sebastian


-------------------------------------------------------------------------=

Using Tomcat but need to do more? Need to support web services, =
security?
Get stuff done quickly with pre-integrated technology to make your job =
easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache =
Geronimo
http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&dat=3D=
121642
_______________________________________________
Numpy-discussion mailing list
Num...@li...
https://lists.sourceforge.net/lists/listinfo/numpy-discussion

Re: [Numpy-discussion] Model and experiment fitting.

[<se...@pi...>]

The problem seemed to be solved, by the A. M. Archibald clue.

I've used splines to fit the simulation data. After that, I can easily
find any Y(X) point, for all X in range (x_min,x_max) where x_min and 
x_max are the experiment independent variable. The experimental data 
stay untouched.

Sorry for all the confusion I've made.

Thanks a lot to all of You!

Sebastian

Re: [Numpy-discussion] Model and experiment fitting.

[Charles R H. <cha...@gm...>]
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Re: [Numpy-discussion] Model and experiment fitting.

[<se...@pi...>]

A. M. Archibald napisał(a):
> 
> In scipy there are some very convenient spline fitting tools which
> will allow you to fit a nice smooth spline through the simulation data
> points (or near, if they have some uncertainty); you can then easily
> look at the RMS difference in the y values. You can also, less easily,
> look at the distance from the curve allowing for some uncertainty in
> the x values.
> 

I'll try a spline fitting. I've already made some linear interpolations 
(see Robert Kern answer) which works well enough to use it. I'm working 
on a genetic algorithms application to the model parameters 
optimalization problem and this RMSe comparison serves me as 'fitness 
function'. This 'fitness function' is important element in whole 
procedure, so I'm trying to found the best solution to obtain it.


> I suppose you could also fit a curve through the experimental points
> and compare the two curves in some way.
> 

Well, I can do it, indeed. But every single fitting procedure implicate 
some additional error, so when it comes to fit, I must use it very 
cautiously. The simulated data-points fitting should be the only 
acceptable fitting procedure, I guess.

> If you want to avoid using an a priori model, Numerical Recipes
> discuss some possible approaches ("Do two-dimensional distributions
> differ?" at http://www.nrbook.com/a/bookcpdf.html is one) but it's not
> clear how to turn the problem you describe into a solvable one - some
> assumption about how the models vary between sampled x values appears
> to be necessary, and that amounts to interpolation.
> 

I'll look to this NR discussion.



Thank You for these comments!

Sebastian

Re: [Numpy-discussion] Model and experiment fitting.

[<se...@pi...>]

Robert Kern napisał(a):

> Your description is a bit vague. 

Possibly by my weak English... I'll try to make myself clearer now.

Do you mean that you have some model function f
> that maps X values to Y values?
> 
>    f(x) -> y
> 

My model is quantum energy operator - spin hamiltonian (SH) with some
additional assumption about so called 'line shape', 'line widths',etc.

   It describes various electron interactions, visible in electron 
paramagnetic resonance (EPR, ESR) experiment. The simplest SH can
be written in a form:
                        H = m B g S                   (1)
where m is a constant (bohr magneton), B is magnetic field (my 
x-variable), g is so called 'zeeman matrix' and S is total spin angular
momentum operator.

Summing it all together: the simple model is parametrized by:
  - line shape,
  - line width,
  - zeeman matrix (3x3 diagonal matrix - the spatial dependence),
  - total spin S.

After SH (1) diagonalization one can obtain so called 'resonance fields' 
and  'resonance intensities'. After a convolution with appropriate  line 
shape function which is parametrized by the line width one can finally
get the simulated EPR spectrum (simDat=[[X1,...,Xn],[Y1,...,Yn]]).
This  is a roughly, schematic description, appropriate to EPR spectra of
monocrystals.

In my situation the problem is more sophisticated - I have 
polycrystaline (powders) data, and to obtain a simulated EPR powder 
spectrum I need to sum up the EPR spectra of monocrystals that come from 
many possible spatial orientations, and the resultant spectrum is an 
envelope of all the monocrystals spectra.

There's no simple model function that maps X -> Y.


> If that is the case, is there some reason that you cannot run your simulation 
> using the same X points as your experimental data?
> 

I can only demand a X range and number of X values within the range, 
there's no possibility to find the Y(X) for a specified X. These 
limitations on one hand come from  the external program I'm using to 
simulate the EPR spectra, on the other are a result of spatial averaging 
of EPR data for powders, where a lot of interpolations are involved.


> OTOH, is there some other independent variable (say Z) that *is* common between 
> your experimental and simulated data?
> 
>    f(z) -> (x, y)
> 

This is probably the situation I'm in. These other variables are my 
model parameters, namely: line shape-width, zeeman matrix... and they're
commen between the experiment and the simulation.


To make it clear.

I've already solved the problem by a simple linear interpolation of 
simulated points within the narrow neighborhood of experimental data 
point. The simulation points are uniformly distributed along the 
X-range, with a density I'm able to tune. It all works quite well but 
I'm founding it as a 'brute-force' method and I wonder, if there's any 
more sophisticated and maybe already incorporated into any Python module 
method?

Anyway, it looks like it's impossible to compare two discrete 2D data 
sets without any interpolations included... :]


A. M. Archibald has proposed spline fitting, which I'll try. I'll also 
look at the Numerical Recipes discussion he has proposed.


Sebastian

Re: [Numpy-discussion] Problem introduced after 1.0rc2 on AIX with xlc

[Travis O. <oli...@ie...>]

Brian Granger wrote:
> Hi,
>
> i am running numpy on aix compiling with xlc.  Revision 1.0rc2 works
> fine and passes all tests.  But 1.0rc3 and more recent give the
> following on import:
>   

Most likely the error-detection code is not working on your platform.    
The platform dependent stuff is not that difficult.   I tried to 
implement something for AIX, but very likely got it wrong (and don't 
have a platform to test it on).  It is the UFUNC_CHECK_STATUS that must 
be implemented.   Perhaps, we can do a simple check and disable the 
error modes:

seterr(all='ignore')

will work and "turn-off" error-detection on your platform.

-Travis

Re: [Numpy-discussion] Problem introduced after 1.0rc2 on AIX with xlc

[Brian G. <ell...@gm...>]

Thanks, I will investigate more on these things and get back to you
early in the week.  But for now numpy seems to be functioning pretty
normally (log(2) gives the correct answer).  thanks again.
It would be great to figure this stuff out before 1.0, but we might
not have time.


Brian

On 10/20/06, Tim Hochberg <tim...@ie...> wrote:
> Brian Granger wrote:
> > When I set seterr(all='warn') I see the following:
> >
> > In [1]: import numpy
> > /usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/ufunclike.py:46:
> > RuntimeWarning: invalid value encountered in log
> >   _log2 = umath.log(2)
> > /usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/scimath.py:19:
> > RuntimeWarning: invalid value encountered in log
> >   _ln2 = nx.log(2.0)
> >
> [etc, etc]
>
> Wow! That looks pretty bad. What do you get if you try just
> "numpy.log(2)" or "numpy.log(2.0)"? Is it producing sane results for
> scalars at all? I suppose another possibility is that the error
> reporting is broken on AIX for some reason.
>
> Hmmm.
>
> I'm betting that is is. The macro UFUNC_CHECK_STATUS is very platform
> dependent. There is a version from AIX (ufuncobject.h line 301), but
> perhaps it's broken on your particular configuration and as a result is
> spitting out all kinds of bogus errors. This is only coming to light now
> because the default error checking level got cranked up.
>
> I gotta call it a night and I'll be out tomorrow, so I won't be much
> more help, but here's something that you might look into: have you
> compiled numarray sucessfully? If you haven't you might want to try it.
> It uses the same default error checking that numpy is now using. If you
> have, you might want to look for the equivalent of UFUNC_CHECK_STATUS
> (it might even have the same name) and splice it into numpy and see if
> it fixes your problems.
>
> Of course, if numpy.log(2) is spitting out something bogus, there's
> something much worse going on, but I suspect you would have noticed that
> by now.
>
> Good luck,
>
> -tim
>
>
> -------------------------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> Numpy-discussion mailing list
> Num...@li...
> https://lists.sourceforge.net/lists/listinfo/numpy-discussion
>

Re: [Numpy-discussion] Problem introduced after 1.0rc2 on AIX with xlc

[Tim H. <tim...@ie...>]

Brian Granger wrote:
> Tim,
>
> I just tried everything with r3375.  I set seterr(all='warn') and the
> tests passed.  But all the floating point warning are still there.
> With seterr(all='ignore') the warnings go away and all the tests pass.
>  should I worry about the warnings?
>   
Maybe. I just sent you some email on this. But my guess is that the code 
that checks for FP errors is broken on your particular system. Mainly I 
suspect this because I think you would have noticed by now if everything 
was as broken as the warnings seem to indicate. Assuming that's the 
case, and this will probably become clear if you  test a bunch of 
computations that give correct (and non INF/NAN) results, but still spit 
out warnings, you have two choices: try to fix the warnings code or 
disable the warnings. The former would be preferable since then you 
could actually use the warnings code, but it may be a pain in the neck 
unless you can find some place to steal the relevant code from.

-tim


> thanks
>
> Brian
>
>
>
> On 10/20/06, Tim Hochberg <tim...@ie...> wrote:
>   
>> Brian Granger wrote:
>>     
>>> Also, when I use seterr(all='ignore') the the tests fail:
>>>
>>> ======================================================================
>>> FAIL: Ticket #112
>>> ----------------------------------------------------------------------
>>> Traceback (most recent call last):
>>>   File "/usr/common/homes/g/granger/usr/local/lib/python/numpy/core/tests/test_regression.py",
>>> line 219, in check_longfloat_repr
>>>     assert(str(a)[1:9] == str(a[0])[:8])
>>> AssertionError
>>>
>>> ----------------------------------------------------------------------
>>> Ran 516 tests in 0.823s
>>>
>>> FAILED (failures=1)
>>>
>>> Thanks for helping out on this.
>>>
>>>       
>> How recent is your version? I just a problem that was causing this same
>> failure yesterday -- if you checkout is older than that, you may want to
>> get the most recent stuff from SVN and see if that fixes this.
>>
>> -tim
>>
>>
>> -------------------------------------------------------------------------
>> Using Tomcat but need to do more? Need to support web services, security?
>> Get stuff done quickly with pre-integrated technology to make your job easier
>> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
>> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
>> _______________________________________________
>> Numpy-discussion mailing list
>> Num...@li...
>> https://lists.sourceforge.net/lists/listinfo/numpy-discussion
>>
>>     
>
> -------------------------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> Numpy-discussion mailing list
> Num...@li...
> https://lists.sourceforge.net/lists/listinfo/numpy-discussion
>
>
>   

Re: [Numpy-discussion] Problem introduced after 1.0rc2 on AIX with xlc

[Brian G. <ell...@gm...>]

Tim,

I just tried everything with r3375.  I set seterr(all='warn') and the
tests passed.  But all the floating point warning are still there.
With seterr(all='ignore') the warnings go away and all the tests pass.
 should I worry about the warnings?

thanks

Brian



On 10/20/06, Tim Hochberg <tim...@ie...> wrote:
> Brian Granger wrote:
> > Also, when I use seterr(all='ignore') the the tests fail:
> >
> > ======================================================================
> > FAIL: Ticket #112
> > ----------------------------------------------------------------------
> > Traceback (most recent call last):
> >   File "/usr/common/homes/g/granger/usr/local/lib/python/numpy/core/tests/test_regression.py",
> > line 219, in check_longfloat_repr
> >     assert(str(a)[1:9] == str(a[0])[:8])
> > AssertionError
> >
> > ----------------------------------------------------------------------
> > Ran 516 tests in 0.823s
> >
> > FAILED (failures=1)
> >
> > Thanks for helping out on this.
> >
> How recent is your version? I just a problem that was causing this same
> failure yesterday -- if you checkout is older than that, you may want to
> get the most recent stuff from SVN and see if that fixes this.
>
> -tim
>
>
> -------------------------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> Numpy-discussion mailing list
> Num...@li...
> https://lists.sourceforge.net/lists/listinfo/numpy-discussion
>

Re: [Numpy-discussion] Problem introduced after 1.0rc2 on AIX with xlc

[Tim H. <tim...@ie...>]

Brian Granger wrote:
> When I set seterr(all='warn') I see the following:
>
> In [1]: import numpy
> /usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/ufunclike.py:46:
> RuntimeWarning: invalid value encountered in log
>   _log2 = umath.log(2)
> /usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/scimath.py:19:
> RuntimeWarning: invalid value encountered in log
>   _ln2 = nx.log(2.0)
>   
[etc, etc]

Wow! That looks pretty bad. What do you get if you try just 
"numpy.log(2)" or "numpy.log(2.0)"? Is it producing sane results for 
scalars at all? I suppose another possibility is that the error 
reporting is broken on AIX for some reason.

Hmmm.

I'm betting that is is. The macro UFUNC_CHECK_STATUS is very platform 
dependent. There is a version from AIX (ufuncobject.h line 301), but 
perhaps it's broken on your particular configuration and as a result is 
spitting out all kinds of bogus errors. This is only coming to light now 
because the default error checking level got cranked up.

I gotta call it a night and I'll be out tomorrow, so I won't be much 
more help, but here's something that you might look into: have you 
compiled numarray sucessfully? If you haven't you might want to try it.  
It uses the same default error checking that numpy is now using. If you 
have, you might want to look for the equivalent of UFUNC_CHECK_STATUS 
(it might even have the same name) and splice it into numpy and see if 
it fixes your problems.

Of course, if numpy.log(2) is spitting out something bogus, there's 
something much worse going on, but I suspect you would have noticed that 
by now.

Good luck,

-tim

Re: [Numpy-discussion] Problem introduced after 1.0rc2 on AIX with xlc

[Brian G. <ell...@gm...>]

I have been doing these recent tests with 1.0rc3.  I am building from
trunk right now and we will see how that goes.  Thanks for your help.

Brian

On 10/20/06, Tim Hochberg <tim...@ie...> wrote:
> Brian Granger wrote:
> > Also, when I use seterr(all='ignore') the the tests fail:
> >
> > ======================================================================
> > FAIL: Ticket #112
> > ----------------------------------------------------------------------
> > Traceback (most recent call last):
> >   File "/usr/common/homes/g/granger/usr/local/lib/python/numpy/core/tests/test_regression.py",
> > line 219, in check_longfloat_repr
> >     assert(str(a)[1:9] == str(a[0])[:8])
> > AssertionError
> >
> > ----------------------------------------------------------------------
> > Ran 516 tests in 0.823s
> >
> > FAILED (failures=1)
> >
> > Thanks for helping out on this.
> >
> How recent is your version? I just a problem that was causing this same
> failure yesterday -- if you checkout is older than that, you may want to
> get the most recent stuff from SVN and see if that fixes this.
>
> -tim
>
>
> -------------------------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> Numpy-discussion mailing list
> Num...@li...
> https://lists.sourceforge.net/lists/listinfo/numpy-discussion
>

Re: [Numpy-discussion] Problem introduced after 1.0rc2 on AIX with xlc

[Tim H. <tim...@ie...>]

Brian Granger wrote:
> Also, when I use seterr(all='ignore') the the tests fail:
>
> ======================================================================
> FAIL: Ticket #112
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>   File "/usr/common/homes/g/granger/usr/local/lib/python/numpy/core/tests/test_regression.py",
> line 219, in check_longfloat_repr
>     assert(str(a)[1:9] == str(a[0])[:8])
> AssertionError
>
> ----------------------------------------------------------------------
> Ran 516 tests in 0.823s
>
> FAILED (failures=1)
>
> Thanks for helping out on this.
>   
How recent is your version? I just a problem that was causing this same 
failure yesterday -- if you checkout is older than that, you may want to 
get the most recent stuff from SVN and see if that fixes this.

-tim

Re: [Numpy-discussion] Problem introduced after 1.0rc2 on AIX with xlc

[Brian G. <ell...@gm...>]

When I set seterr(all='warn') I see the following:

In [1]: import numpy
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/ufunclike.py:46:
RuntimeWarning: invalid value encountered in log
  _log2 = umath.log(2)
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/scimath.py:19:
RuntimeWarning: invalid value encountered in log
  _ln2 = nx.log(2.0)
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:64:
RuntimeWarning: invalid value encountered in add
  two = one + one
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:65:
RuntimeWarning: invalid value encountered in subtract
  zero = one - one
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:71:
RuntimeWarning: invalid value encountered in add
  a = a + a
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:72:
RuntimeWarning: invalid value encountered in add
  temp = a + one
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:73:
RuntimeWarning: invalid value encountered in subtract
  temp1 = temp - a
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:74:
RuntimeWarning: invalid value encountered in subtract
  if any(temp1 - one != zero):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:74:
RuntimeWarning: invalid value encountered in not_equal
  if any(temp1 - one != zero):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:80:
RuntimeWarning: invalid value encountered in add
  b = b + b
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:81:
RuntimeWarning: invalid value encountered in add
  temp = a + b
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:82:
RuntimeWarning: invalid value encountered in subtract
  itemp = int_conv(temp-a)
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:83:
RuntimeWarning: invalid value encountered in not_equal
  if any(itemp != 0):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:95:
RuntimeWarning: invalid value encountered in multiply
  b = b * beta
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:96:
RuntimeWarning: invalid value encountered in add
  temp = b + one
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:97:
RuntimeWarning: invalid value encountered in subtract
  temp1 = temp - b
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:98:
RuntimeWarning: invalid value encountered in subtract
  if any(temp1 - one != zero):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:98:
RuntimeWarning: invalid value encountered in not_equal
  if any(temp1 - one != zero):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:103:
RuntimeWarning: invalid value encountered in divide
  betah = beta / two
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:106:
RuntimeWarning: invalid value encountered in add
  a = a + a
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:107:
RuntimeWarning: invalid value encountered in add
  temp = a + one
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:108:
RuntimeWarning: invalid value encountered in subtract
  temp1 = temp - a
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:109:
RuntimeWarning: invalid value encountered in subtract
  if any(temp1 - one != zero):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:109:
RuntimeWarning: invalid value encountered in not_equal
  if any(temp1 - one != zero):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:113:
RuntimeWarning: invalid value encountered in add
  temp = a + betah
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:115:
RuntimeWarning: invalid value encountered in subtract
  if any(temp-a != zero):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:115:
RuntimeWarning: invalid value encountered in not_equal
  if any(temp-a != zero):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:117:
RuntimeWarning: invalid value encountered in add
  tempa = a + beta
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:118:
RuntimeWarning: invalid value encountered in add
  temp = tempa + betah
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:119:
RuntimeWarning: invalid value encountered in subtract
  if irnd==0 and any(temp-tempa != zero):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:119:
RuntimeWarning: invalid value encountered in not_equal
  if irnd==0 and any(temp-tempa != zero):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:124:
RuntimeWarning: invalid value encountered in divide
  betain = one / beta
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:127:
RuntimeWarning: invalid value encountered in multiply
  a = a * betain
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:130:
RuntimeWarning: invalid value encountered in subtract
  temp = one - a
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:131:
RuntimeWarning: invalid value encountered in subtract
  if any(temp-one != zero):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:131:
RuntimeWarning: invalid value encountered in not_equal
  if any(temp-one != zero):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:133:
RuntimeWarning: invalid value encountered in multiply
  a = a * beta
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:149:
RuntimeWarning: invalid value encountered in add
  temp = one + a
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:150:
RuntimeWarning: invalid value encountered in subtract
  if any(temp-one != zero):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:150:
RuntimeWarning: invalid value encountered in not_equal
  if any(temp-one != zero):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:152:
RuntimeWarning: invalid value encountered in multiply
  a = a * beta
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:160:
RuntimeWarning: invalid value encountered in add
  temp = one + eps
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:168:
RuntimeWarning: invalid value encountered in add
  t = one + eps
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:172:
RuntimeWarning: invalid value encountered in multiply
  z = y*y
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:173:
RuntimeWarning: invalid value encountered in multiply
  a = z*one # Check here for underflow
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:174:
RuntimeWarning: invalid value encountered in multiply
  temp = z*t
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:175:
RuntimeWarning: invalid value encountered in add
  if any(a+a == zero) or any(abs(z)>=y):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:175:
RuntimeWarning: invalid value encountered in equal
  if any(a+a == zero) or any(abs(z)>=y):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:175:
RuntimeWarning: invalid value encountered in absolute
  if any(a+a == zero) or any(abs(z)>=y):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:175:
RuntimeWarning: invalid value encountered in greater_equal
  if any(a+a == zero) or any(abs(z)>=y):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:177:
RuntimeWarning: invalid value encountered in multiply
  temp1 = temp * betain
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:178:
RuntimeWarning: invalid value encountered in multiply
  if any(temp1*beta == z):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:178:
RuntimeWarning: invalid value encountered in equal
  if any(temp1*beta == z):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:174:
RuntimeWarning: underflow encountered in multiply
  temp = z*t
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:184:
RuntimeWarning: invalid value encountered in not_equal
  if ibeta != 10:
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:198:
RuntimeWarning: invalid value encountered in multiply
  y = y * betain
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:199:
RuntimeWarning: invalid value encountered in multiply
  a = y * one
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:200:
RuntimeWarning: invalid value encountered in multiply
  temp = y * t
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:201:
RuntimeWarning: invalid value encountered in add
  if any(a+a != zero) and any(abs(y) < xmin):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:201:
RuntimeWarning: invalid value encountered in not_equal
  if any(a+a != zero) and any(abs(y) < xmin):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:201:
RuntimeWarning: invalid value encountered in absolute
  if any(a+a != zero) and any(abs(y) < xmin):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:201:
RuntimeWarning: invalid value encountered in less
  if any(a+a != zero) and any(abs(y) < xmin):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:203:
RuntimeWarning: invalid value encountered in multiply
  temp1 = temp * betain
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:204:
RuntimeWarning: invalid value encountered in multiply
  if any(temp1*beta == y) and any(temp != y):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:204:
RuntimeWarning: invalid value encountered in equal
  if any(temp1*beta == y) and any(temp != y):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:203:
RuntimeWarning: underflow encountered in multiply
  temp1 = temp * betain
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:204:
RuntimeWarning: invalid value encountered in not_equal
  if any(temp1*beta == y) and any(temp != y):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:215:
RuntimeWarning: invalid value encountered in not_equal
  if mx <= k + k - 3 and ibeta != 10:
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:223:
RuntimeWarning: invalid value encountered in equal
  if ibeta == 2 and not i:
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:227:
RuntimeWarning: invalid value encountered in not_equal
  if any(a != y):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:229:
RuntimeWarning: invalid value encountered in subtract
  xmax = one - epsneg
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:230:
RuntimeWarning: invalid value encountered in multiply
  if any(xmax*one != xmax):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:230:
RuntimeWarning: invalid value encountered in not_equal
  if any(xmax*one != xmax):
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:232:
RuntimeWarning: invalid value encountered in multiply
  xmax = xmax / (xmin*beta*beta*beta)
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:232:
RuntimeWarning: invalid value encountered in divide
  xmax = xmax / (xmin*beta*beta*beta)
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:235:
RuntimeWarning: invalid value encountered in equal
  if ibeta==2:
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:236:
RuntimeWarning: invalid value encountered in add
  xmax = xmax + xmax
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:266:
RuntimeWarning: invalid value encountered in add
  ten = two + two + two + two + two
/usr/common/homes/g/granger/usr/local/lib/python/numpy/lib/machar.py:267:
RuntimeWarning: invalid value encountered in power
  resolution = ten ** (-self.precision)


On 10/20/06, Tim Hochberg <tim...@ie...> wrote:
> Brian Granger wrote:
> > Hi,
> >
> > i am running numpy on aix compiling with xlc.  Revision 1.0rc2 works
> > fine and passes all tests.  But 1.0rc3 and more recent give the
> > following on import:
> >
> > Warning: invalid value encountered in multiply
> > Warning: invalid value encountered in multiply
> > Warning: invalid value encountered in multiply
> > Warning: invalid value encountered in add
> > Warning: invalid value encountered in not_equal
> > Warning: invalid value encountered in absolute
> > Warning: invalid value encountered in less
> > Warning: invalid value encountered in multiply
> > Warning: invalid value encountered in multiply
> > Warning: invalid value encountered in equal
> > Warning: invalid value encountered in multiply
> > Warning: invalid value encountered in multiply
> > Warning: invalid value encountered in multiply
> > Warning: invalid value encountered in add
> > Warning: invalid value encountered in not_equal
> > Warning: invalid value encountered in absolute
> > Warning: invalid value encountered in less
> > Warning: invalid value encountered in multiply
> > Warning: invalid value encountered in multiply
> > Warning: invalid value encountered in equal
> > Warning: invalid value encountered in multiply
> > Warning: invalid value encountered in multiply
> > Warning: invalid value encountered in multiply
> > [lots more of this]
> >
> > The odd thing is that all tests pass.  I have looked, but can't find
> > where this Warning is coming from  in the code.  Any thoughts on where
> > this is coming from?  What can I do to help debug this?  I am not sure
> > what revision introduced this issue.
> >
> The reason that you are seeing this now is that the default error state
> has been tightened up. There were some issues with tests failing as a
> result of this, but I believe I fixed those already and you're seeing
> this on import, not when running the tests correct? The first thing to
> do is figure out where the invalids are occurring, and the natural way
> to do that is to set the error state to raise, but you can't set the
> error state till you import it, so that's not going to help here.
>
> I think the first thing that I would try is to throw in a
> seterr(all='raise', under='ignore') right after the call to _setdef in
> numeric.py. If you're lucky, this will point out where the invalids are
> popping up. As a sanity check, you could instead make this
> seterr(all='ignore'), which should make all the warnings go away, but
> won't tell you anything about why there are warnings to begin with.
>
> Regards,
>
> -tim
>
>
> -------------------------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> Numpy-discussion mailing list
> Num...@li...
> https://lists.sourceforge.net/lists/listinfo/numpy-discussion
>