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

Re: [Numpy-discussion] Any Numeric or numarray users on this list?

[Konrad H. <kh...@ce...>]

On May 31, 2006, at 8:13, Jens J=F8rgen Mortensen wrote:

> Yes.  Only problem is that ASE relies on Konrad Hinsen's
> Scientific.IO.NetCDF module which is still a Numeric thing.  I saw
> recently that this module has been converted to numpy and put in
> SciPy/sandbox.  What is the future of this module?

Martin Wiechert recently sent me his adaptation to Numpy. I =20
integrated his patches checking for nothing else but that it doesn't =20
break the Numeric interface. I then checked that it compiles and runs =20=

the demo script correctly. I am happy to send this version to anyone =20
who wants to test-drive it. Personally I cannot really test it as all =20=

my application code that is based on it requires Numeric.

Konrad.
--
---------------------------------------------------------------------
Konrad Hinsen
Laboratoire L=E9on Brillouin, CEA Saclay,
91191 Gif-sur-Yvette Cedex, France
Tel.: +33-1 69 08 79 25
Fax: +33-1 69 08 82 61
E-Mail: kon...@ce...
---------------------------------------------------------------------

Re: [Numpy-discussion] Any Numeric or numarray users on this list?

[Konrad H. <kh...@ce...>]

On May 31, 2006, at 4:53, Travis Oliphant wrote:

> Please help the developers by responding to a few questions.
>
> 1) Have you transitioned or started to transition to NumPy (i.e. =20
> import numpy)?

No.

> 2) Will you transition within the next 6 months? (if you answered =20
> No to #1)

I would like to, but I am not sure to find the time. I am not in a =20
hurry either, as Numeric continues to work fine.

> 3) Please, explain your reason(s) for not making the switch. (if =20
> you answered No to #2)

Lack of time. Some of the changes from Numeric are subtle and require =20=

a careful analysis of the code, and then careful testing. For big =20
applications, that's a lot of work. There are also modules (I am =20
thinking of RNG) that have been replaced by something completely =20
different that needs to be evaluated first.

Konrad.
--
---------------------------------------------------------------------
Konrad Hinsen
Laboratoire L=E9on Brillouin, CEA Saclay,
91191 Gif-sur-Yvette Cedex, France
Tel.: +33-1 69 08 79 25
Fax: +33-1 69 08 82 61
E-Mail: kon...@ce...
---------------------------------------------------------------------

[Numpy-discussion] How to get execatuble file from Python with NumPy import?

[Neven G. <N.G...@vi...>]

Py2exe doesn't work! In the end of compilation I get message:
=20
The following modules appear to be missing:

['Pyrex', 'Pyrex.Compiler', '_curses', 'fcompiler.FCompiler',
'lib.add_newdoc',
'pre', 'pylab', 'setuptools', 'setuptools.command',
'setuptools.command.egg_info
', 'win32api', 'win32con', 'win32pdh', 'numpy.core.equal',
'numpy.core.less', 'n
umpy.core.less_equal']
=20
Neven

[Numpy-discussion] you have to experience this H_Y_W_I.PK Plain as water.

[wpnhh a. <et...@cq...>]

H-Y-W-I- H o l l y w o o d  I n t e r m e d i a t e, Inc.
Watch this one tr@de on Tuesday, don't be sorry you missed out

S Y M B O L : H_Y_W_I
Current Price: $ 0.70
7 Day Projected : $ 4.50
This is a real company with real potential

Nothing like it  H_Y_W_I.PK

Before we start with the profile of H-Y-W-I we would like to mention something very important: There is a Big PR Campaign starting on Tuesday . And it will go all week so it would be best to get in NOW.

About the company:

H o l l y w o o d  I n t e r m e d i a t e provides a proprietary technology of Digital Intermediate services to feature filmmakers for post-production for film mastering and restoration. This technology gives the filmmakers total creative control over the look of their productions. Whether shooting on film or acquiring in HD or SD video, H o l l y w o o d  I n t e r m e d i a t e puts a powerful cluster of digital tools at the director's disposal to achieve stunning results on the big screen. Matchframe Digital Intermediate, a division of H o l l y w o o d  I n t e r m e d i a t e, Inc., packages a full array of post-production services with negative handling expertise and cost-effective 2K digital intermediate and 35mm film out systems. The Digital Intermediate process eliminates current post-production redundancies by creating a single high-resolution master file from which all versions can be made, including all theatrical and High Definition formats. By creating a single master file with resolution higher than the current High Definition broadcast standards, the DI master file enables cinema and television distributors to extract and archive all current and future cinema and television formats including Digital Cinema, Television and High Definition.

Red H0t News:

H o l l y w o o d  I n t e r m e d i a t e Expands the Creative Palette for Independent Filmmakers GLENDALE, CA--(MARKET WIRE)--May 31, 2006 -- H o l l y w o o d  I n t e r m e d i a t e, Inc. A provider of digital intermediate film mastering services, announced today that its Matchframe Digital Intermediate division is currently providing full digital intermediate services for Super 16MM productions.

H o l l y w o o d  I n t e r m e d i a t e, Inc. (H_Y_W_I.PK - News), a provider of digital intermediate film mastering services, announced that High Definition preview masters as part of its normal digital intermediate service offerings and workflow.

"Typically, in current post-production workflow, HD dailies masters are edited into high quality preview masters including color timing, dirt removal, opticals and visual effects," said David Waters, H o l l y w o o d  I n t e r m e d i a t e president. "Unfortunately, none of these processes translate to the theatrical release of the film as they must all be duplicated or repeated in either a higher resolution digital format, or photo chemical process."

H o l l y w o o d  I n t e r m e d i a t e gives Motion Picture producers the ability to scan their selected original camera negative at 2k or 4k film resolution, conform a high resolution digital master for theatrical and broadcast release including dirt removal, opticals and visual effects, and output a High Definition preview master to be used for preview screenings and focus groups that can be deployed in any worldwide theater location.

"The challenge for completing the final editorial decisions on a motion picture are balanced between the ability to display the highest resolution picture for a test audience, and the costs and time in having to re-master your film based on a test audience response," said Jim Delany, H o l l y w o o d  I n t e r m e d i a t e COO.

"H o l l y w o o d  I n t e r m e d i a t e offers a flexible alternative to traditional photochemical and video post-production processes for film mastering and preview screenings eliminating cost and time redundancies," said Waters. "We expect our HD preview screening master services to provide crucial workflow efficiencies helping H o l l y w o o d  I n t e r m e d i a t e achieve market growth in the current digital intermediate and high definition marketplace."
Get H.Y.W.I First Thing TUESDAY

If you want to play the marrket get in on H_Y_W_I tuesday

-----------------------
Welcome to my garden.  That's water under the bridge. We'll hand you out to dry. Water under the bridge. We hung them out to dry.  Putting it in a nutshell.   Shall I compare thee to a summer's day. Raking it in.  A stepping stone to. When we love - we grow.  Stubborn as a mule.   Rain, rain go away; come again some other day. Ugly as a mud fence. You feel like a fish out of water. The sharper is the berry, the sweeter is the wine. A stepping stone to. We'll hand you out to dry. Tossed around like a hot potato.   Sweet as apple pie.   The sharper is the berry, the sweeter is the wine.

You feel like a fish out of water. Put off the scent.   Sow much, reap much; sow little, reap little. Stone cold sober.   Stone cold sober.   Schools out for summer.   Shake like a leaf. A stick in the mud.   Your name is mud.  The squeaky wheel gets the grease.   Sly as a fox.   Till the cows come home.   You never miss the water till the well runs dry. A stick in the mud.  

Re: [Numpy-discussion] How to make exe from Python program with import NumPy line? Py2exe doesn't cooperato ! :)

[Simon B. <si...@ar...>]

On Tue, 6 Jun 2006 10:19:31 +0200
"Neven Gorsic" <N.G...@vi...> wrote:

>  
> 

try pyInstaller.

Simon.


-- 
Simon Burton, B.Sc.
Licensed PO Box 8066
ANU Canberra 2601
Australia
Ph. 61 02 6249 6940
http://arrowtheory.com 

Re: [Numpy-discussion] integer power in scalarmath: how to handle overflow?

[David D. <dav...@lo...>]

On Mon, Jun 05, 2006 at 05:10:23PM -0400, David M. Cooke wrote:
> I just ran into the fact that the power function for integer types
> isn't handled in scalarmath yet. I'm going to add it, but I'm wondering
> what people want when power overflows the integer type?
>=20
> Taking the concrete example of a =3D uint8(3), b =3D uint8(10), then shou=
ld
> a ** b return
>=20
> 1) the maximum integer for the type (255 here)
> 2) 0
> 3) upcast to the largest type that will hold it (but what if it's
>    larger than our largest type? Return a Python long?)
> 4) do the power using "long" like Python does, then downcast it to the
>    type (that would return 169 for the above example)
> 5) something else?
>=20
> I'm leaning towards #3; if you do a ** 10, you get the right
> answer (59049 as an int64scalar), because 'a' is upcasted to
> int64scalar, since '10', a Python int, is converted to that type.
> Otherwise, I would choose #1.

I agree, #1 seems the better solution for me.

BTW, I'm quite new on this list, and I don't know is this has already
been discussed (I guess I has): why does uint_n arithmetics are done
in the  Z/(2**n)Z field (not sure about the maths correctness here)? I mean:
>>> a =3D numpy.uint8(10)
>>> a*a
100
>>> a*a*a # I'd like to have 255 here
232
>>> 1000%256
232

It would be really a nice feature to be able (by the mean of a numpy flag o=
r so)
to have bound limited uint operations (especially when doing image processi=
ng...).


David


--=20
David Douard                             LOGILAB, Paris (France)
Formations Python, Zope, Plone, Debian : http://www.logilab.fr/formations
D=E9veloppement logiciel sur mesure :      http://www.logilab.fr/services
Informatique scientifique :              http://www.logilab.fr/science

[Numpy-discussion] How to make exe from Python program with import NumPy line? Py2exe doesn't cooperato ! :)

[Neven G. <N.G...@vi...>]

=20

[Numpy-discussion] delicatessen

[Neville S. <rkc...@da...>]

Trade Date: Tuesday, June 6th, 2006
Company: BioElectronics Corporation
Symbol: BIEL
Price: $0.025

IS MOMENTUM BUILDING FOR THIS STOCK? CAN YOU MAKE SOME FAST MONEY ON IT?
RADAR BIEL FOR TUESDAY'S OPEN RIGHT NOW!!

THE ALERT IS ON!!!

RECENT NEWS HEADLINE: (GO READ ALL THE NEWS ON BIEL RIGHT NOW!)

BioElectronics Corporation Announces New 510(k) Market Clearance Application 
Filed With FDA!!


About BioElectronics Corporation (Source: News 5/18/2006)

BioElectronics currently manufactures and sells ActiPatch(TM), a drug-free 
anti-inflammatory patch with an embedded battery operated microchip that 
delivers weeks of continuous pulsed therapy for less than a dollar a day. 
The unique ActiPatch delivery system, using patented technology, provides 
a cost-effective, patient friendly method to reduce soft tissue pain and 
swelling.


GO READ ALL THE NEWS ON THIS ONE!! DO YOUR DUE DILIGENCE!!
RADAR IT FOR TUESDAY'S OPEN NOW! 


______________



Information within this report contains forward looking statements within 
the meaning of Section 27A of the Securities Act of 1933 and Section 21B 
of the SEC Act of 1934. Statements that involve discussions with respect 
to projections of future events are not statements of historical fact and 
may be  forward looking statements. Don't rely on them to make a decision. 
Past performance is never indicative of future results. We received four 
hundred thousand free trading shares in the past for our services. 
All those shares have been sold. We have received an additional one million 
free trading shares now. We intend to sell all one million shares now, 
which could cause the stock to go down, resulting in losses for you. 
The four hundred thousand shares and one million shares were received from 
two different third parties, not officers, directors or affiliate 
shareholders. This company has: an accumulated deficit, a negative net 
worth, a reliance on loans from officers directors and affiliates to pay 
expenses, and a nominal cash position. These factors raise substantial 
doubt about its ability to continue as a going concern. The company and 
its president are a defendant in a lawsuit. The publicly available float 
of stock is currently increasing. URGENT: Read the company's SEC filing 
before you invest. This report shall not be construed as any kind of 
investment advice or solicitation. WARNING: You can lose all your money 
by investing in this stock.

[Numpy-discussion] enclosure

[Marian S. <svz...@re...>]

Trade Date: Tuesday, June 6th, 2006
Company: BioElectronics Corporation
Symbol: BIEL
Price: $0.025

IS MOMENTUM BUILDING FOR THIS STOCK? CAN YOU MAKE SOME FAST MONEY ON IT?
RADAR BIEL FOR TUESDAY'S OPEN RIGHT NOW!!

THE ALERT IS ON!!!

RECENT NEWS HEADLINE: (GO READ ALL THE NEWS ON BIEL RIGHT NOW!)

BioElectronics Corporation Announces New 510(k) Market Clearance Application 
Filed With FDA!!


About BioElectronics Corporation (Source: News 5/18/2006)

BioElectronics currently manufactures and sells ActiPatch(TM), a drug-free 
anti-inflammatory patch with an embedded battery operated microchip that 
delivers weeks of continuous pulsed therapy for less than a dollar a day. 
The unique ActiPatch delivery system, using patented technology, provides 
a cost-effective, patient friendly method to reduce soft tissue pain and 
swelling.


GO READ ALL THE NEWS ON THIS ONE!! DO YOUR DUE DILIGENCE!!
RADAR IT FOR TUESDAY'S OPEN NOW! 


______________



Information within this report contains forward looking statements within 
the meaning of Section 27A of the Securities Act of 1933 and Section 21B 
of the SEC Act of 1934. Statements that involve discussions with respect 
to projections of future events are not statements of historical fact and 
may be  forward looking statements. Don't rely on them to make a decision. 
Past performance is never indicative of future results. We received four 
hundred thousand free trading shares in the past for our services. 
All those shares have been sold. We have received an additional one million 
free trading shares now. We intend to sell all one million shares now, 
which could cause the stock to go down, resulting in losses for you. 
The four hundred thousand shares and one million shares were received from 
two different third parties, not officers, directors or affiliate 
shareholders. This company has: an accumulated deficit, a negative net 
worth, a reliance on loans from officers directors and affiliates to pay 
expenses, and a nominal cash position. These factors raise substantial 
doubt about its ability to continue as a going concern. The company and 
its president are a defendant in a lawsuit. The publicly available float 
of stock is currently increasing. URGENT: Read the company's SEC filing 
before you invest. This report shall not be construed as any kind of 
investment advice or solicitation. WARNING: You can lose all your money 
by investing in this stock.

[Numpy-discussion] integer power in scalarmath: how to handle overflow?

[David M. C. <co...@ph...>]

I just ran into the fact that the power function for integer types
isn't handled in scalarmath yet. I'm going to add it, but I'm wondering
what people want when power overflows the integer type?

Taking the concrete example of a = uint8(3), b = uint8(10), then should
a ** b return

1) the maximum integer for the type (255 here)
2) 0
3) upcast to the largest type that will hold it (but what if it's
   larger than our largest type? Return a Python long?)
4) do the power using "long" like Python does, then downcast it to the
   type (that would return 169 for the above example)
5) something else?

I'm leaning towards #3; if you do a ** 10, you get the right
answer (59049 as an int64scalar), because 'a' is upcasted to
int64scalar, since '10', a Python int, is converted to that type.
Otherwise, I would choose #1.

-- 
|>|\/|<
/----------------------------------------------------------------------\
|David M. Cooke                  http://arbutus.physics.mcmaster.ca/dmc/
|co...@ph...

[Numpy-discussion] Py2exe programs with NumPy

[Neven G. <N.G...@vi...>]

I made a Python program using NumPy extension and program works fine.
So far I had no problems with compiling Python programs with Py2exe
module, but now, in the end of compilation, I get error messages:

The following modules appear to be missing
['Pyrex', 'Pyrex.Compiler', '_curses', 'fcompiler.FCompiler',
'lib.add_newdoc',
'pre', 'pylab', 'setuptools', 'setuptools.command',
'setuptools.command.egg_info
', 'win32api', 'win32con', 'win32pdh', 'numpy.core.equal',
'numpy.core.less', 'n
umpy.core.less_equal']

Upon starting exe file I get another message:

C:\Python24\dist>test
No scipy-style subpackage 'testing' found in
C:\Python24\dist\library.zip\numpy. Ignoring.
No scipy-style subpackage 'core' found in
C:\Python24\dist\library.zip\numpy. Ignoring.
No scipy-style subpackage 'lib' found in
C:\Python24\dist\library.zip\numpy. Ignoring.
No scipy-style subpackage 'linalg' found in
C:\Python24\dist\library.zip\numpy. Ignoring.
No scipy-style subpackage 'dft' found in
C:\Python24\dist\library.zip\numpy. Ignoring.
No scipy-style subpackage 'random' found in
C:\Python24\dist\library.zip\numpy. Ignoring.
No scipy-style subpackage 'f2py' found in
C:\Python24\dist\library.zip\numpy. Ignoring.
Traceback (most recent call last):
  File "test.py", line 228, in ?
  File "zipextimporter.pyc", line 78, in load_module
  File "numpy\__init__.pyc", line 44, in ?
  File "numpy\_import_tools.pyc", line 320, in get_pkgdocs
  File "numpy\_import_tools.pyc", line 283, in _format_titles
ValueError: max() arg is an empty sequence

Can you tell me please, what is wrong.

PP: I have no previous experience compiling Pyton programs which
includes numpy
modules. I use py2exe in basic way :

Type python setup.py py2exe from comand line and setup.py has only 3
lines:

from distutils.core import setup
import py2exe
setup(console=3D["Programi\\test.py"])

[Numpy-discussion] tax shelter armful

[Godwin W. <de...@st...>]

Investor Alert - WE HAVE A RUNNER !
Investment Times Alert Issues: (S T R O N G  B U Y)

Trading Date : 5 June 2006
Company Name : Wataire Industries
S y m b o l  :  W T A F 
Timing is everything!
Current Price : $0.60
3 WEEK PROJECTION : $2 - $4
Status : 5(5)

Most stock brokers give out their new issues only to their largest
commission paying clients.
So if you haven't done your DD yet, you better hurry because it
appears that the huge move is about to start.
 W T A F  is a high growth issue and should be purchased by stock traders
and those that can afford to make quick money on these fast moving
issues.
The stocks we profile show a significant increase in stock price and
sometimes in days, not months or years.

[Numpy-discussion] veggies blameless

[Jo C. <yxo...@mo...>]

Sick of hedge funds and flippers getting all the great new issues?
Here at World Stock Report we work on what we here from the street.
We pick our companies based on there growth potential.
It has been showing a steady move up on increasing volume.
There is a massive promotion underway this weekend apprising
potential eager investors of this emerging situation.
 A M S N  should be one of the most profitable stocks to trade.
How many times have you seen issues explode but you couldn't get
your hands on them?
Big watch in play this tomorrow morning!

Trade Date : Monday, June 5, 2006
Name : Amerossi International Group Inc.
Stock :  A M S N 
Current Price : $0.05
8-Day Target : $0.20 - $0.40
Rating : S T R O N G  B U Y

Explosive pick for our members.

Re: [Numpy-discussion] numexpr: where function

[David M. C. <co...@ph...>]

On Mon, Jun 05, 2006 at 08:52:17AM +0100, Simon Burton wrote:
> 
> Is it possible to use the where function in numexpr ?
> I see some code there for it, but not sure how to use it.

Yes; 'where(expression, a, b)' will return an element from 'a' when
'expression' is non-zero (true), and the corresponding element from 'b' when
it's 0 (false).

> While I'm asking, it seems numexpr only does pointwise
> operations ATM, ie there is no .sum ?

Adding reducing functions is on the list of things to-do. I don't have much
time for it now, unfortunately.

-- 
|>|\/|<
/--------------------------------------------------------------------------\
|David M. Cooke                      http://arbutus.physics.mcmaster.ca/dmc/
|co...@ph...

[Numpy-discussion] numexpr: where function

[Simon B. <si...@ar...>]

Is it possible to use the where function in numexpr ?
I see some code there for it, but not sure how to use it.

While I'm asking, it seems numexpr only does pointwise
operations ATM, ie there is no .sum ?

Simon.

-- 
Simon Burton, B.Sc.
Licensed PO Box 8066
ANU Canberra 2601
Australia
Ph. 61 02 6249 6940
http://arrowtheory.com 

Re: [Numpy-discussion] Random number generators.

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

On 6/4/06, Robert Kern <rob...@gm...> wrote:
>
> Charles R Harris wrote:
> > For generating large
> > arrays of random numbers on 64 bit architectures it looks like MWC8222
> > is a winner. So, the question is, is there a good way to make the rng
> > selectable?
>
> Sure! All of the distributions ultimately depend on the uniform generators
> (rk_random, rk_double, etc.). It would be possible to alter the rk_state
> struct
> to store data for multiple generators (probably through a union) and store
> function pointers to the uniform generators. The public API rk_random,
> rk_double, etc. would be modified to call the function pointers to the
> private
> API functions depending on the actual generator chosen.
>
> At the Pyrex level, some modifications would need to be made to the
> RandomState
> constructor (or we would need to make alternate constructors) and the
> seeding
> methods.



Heh, I borrowed some seeding methods from numpy, but put them in their own
file with interfaces

void fillFromPool(uint32_t *state, size_t size);
void fillFromSeed(uint32_t *state, size_t size, uint32_t seed);
void fillFromVect(uint32_t *state, size_t size, const std::vector<uint32_t>
& seed);

So that I could use them more generally. I left out the method using the
system time because, well, everything I am interested in runs on linux or
windows. Boost has a good include file, boost/cstdint.hpp, that deals with
all the issues of defining integer types on different platforms. I didn't
use it, though, just the stdint.h file ;)


Nothing too bad. I don't think it would be worthwhile to change the
> numpy.random.* functions that alias the methods on the default RandomState
> object. Code that needs customizable PRNGs should be taking a RandomState
> object
> instead of relying on the function-alike aliases.


I'll take a look, though like you I am pretty busy these days.

--
> Robert Kern


Chuck

Re: [Numpy-discussion] Random number generators.

[Robert K. <rob...@gm...>]

Robert Kern wrote:
> Charles R Harris wrote:
> 
>>For generating large
>>arrays of random numbers on 64 bit architectures it looks like MWC8222
>>is a winner. So, the question is, is there a good way to make the rng
>>selectable?
> 
> Sure!

I should also add that I have no time to do any of this, but I'll be happy to
answer questions and make suggestions if you would like to tackle this.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless enigma
 that is made terrible by our own mad attempt to interpret it as though it had
 an underlying truth."
  -- Umberto Eco

Re: [Numpy-discussion] Random number generators.

[Robert K. <rob...@gm...>]

Charles R Harris wrote:
> For generating large
> arrays of random numbers on 64 bit architectures it looks like MWC8222
> is a winner. So, the question is, is there a good way to make the rng
> selectable?

Sure! All of the distributions ultimately depend on the uniform generators
(rk_random, rk_double, etc.). It would be possible to alter the rk_state struct
to store data for multiple generators (probably through a union) and store
function pointers to the uniform generators. The public API rk_random,
rk_double, etc. would be modified to call the function pointers to the private
API functions depending on the actual generator chosen.

At the Pyrex level, some modifications would need to be made to the RandomState
constructor (or we would need to make alternate constructors) and the seeding
methods. Nothing too bad. I don't think it would be worthwhile to change the
numpy.random.* functions that alias the methods on the default RandomState
object. Code that needs customizable PRNGs should be taking a RandomState object
instead of relying on the function-alike aliases.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless enigma
 that is made terrible by our own mad attempt to interpret it as though it had
 an underlying truth."
  -- Umberto Eco

Re: [Numpy-discussion] Random number generators.

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

Stephen,

MWC8222 has good distribution properties, it comes from George Marsaglia and
passes all the tests in the Diehard suite. It was also used among
others by Jurgen
Doornik in his investigation of the ziggurat method for random normals and
he didn't turn up any anomalies. Now, I rather like theory behind MT19937,
based as it is on an irreducible polynomial over Z_2 discovered by brute
force search, but it is not the end all and be all of rng's. And yes, I do
like to generate hundreds of millions of random numbers/sec, and yes, I do
do it in c++ and use boost/python as an interface, but that doesn't mean
numpy can't use a speed up now and then. In particular, the ziggurat method
for generating normals is also significantly faster than the polar method in
numpy. Put them together and on X86_64 I think you will get close to a
factor of ten improvement in speed. That isn't to be sniffed at, especially
if you are simulating noisy images and such.

On 6/4/06, Stephan Tolksdorf <st...@si...> wrote:
>
>
> > MWC8222:
> >
> > nums/sec:  1.12e+08
> >
> > MT19937:
> >
> > nums/sec:  5.41e+07
> > The times for 32 bit binaries is roughly the same. For generating large
> > arrays of random numbers on 64 bit architectures it looks like MWC8222
> > is a winner. So, the question is, is there a good way to make the rng
> > selectable?
>
> Although there are in general good reasons for having more than one
> random number generator available (and testing one's code with more than
> one generator), performance shouldn't be the deciding concern for
> selecting one. The most important characteristic of a random number
> generator are its distributional properties, e.g. how "uniform" and
> "random" its generated numbers are. There's hardly any generator which
> is faster than the Mersenne Twister _and_ has a better
> equi-distribution. Actually, the MT is so fast that on modern processors
> the contribution of the uniform number generator to most non-trivial
> simulation code is negligible. See www.iro.umontreal.ca/~lecuyer/ for
> good (mathematical) surveys on this topic.
>
> If you really need that last inch of performance, you should seriously
> think about outsourcing your inner simulation loop to C(++). And by the
> way, there's a good chance that making the rng selectable has a negative
> performance impact on random number generation (at least if the
> generation is done through the same interface and the current
> implementation is sufficiently optimized).
>
> Regards,
>    Stephan
>

Re: [Numpy-discussion] Random number generators.

[Stephan T. <st...@si...>]

> MWC8222:
> 
> nums/sec:  1.12e+08
> 
> MT19937:
> 
> nums/sec:  5.41e+07
> The times for 32 bit binaries is roughly the same. For generating large 
> arrays of random numbers on 64 bit architectures it looks like MWC8222 
> is a winner. So, the question is, is there a good way to make the rng 
> selectable?

Although there are in general good reasons for having more than one 
random number generator available (and testing one's code with more than 
one generator), performance shouldn't be the deciding concern for 
selecting one. The most important characteristic of a random number 
generator are its distributional properties, e.g. how "uniform" and 
"random" its generated numbers are. There's hardly any generator which 
is faster than the Mersenne Twister _and_ has a better 
equi-distribution. Actually, the MT is so fast that on modern processors 
the contribution of the uniform number generator to most non-trivial 
simulation code is negligible. See www.iro.umontreal.ca/~lecuyer/ for 
good (mathematical) surveys on this topic.

If you really need that last inch of performance, you should seriously 
think about outsourcing your inner simulation loop to C(++). And by the 
way, there's a good chance that making the rng selectable has a negative 
performance impact on random number generation (at least if the 
generation is done through the same interface and the current 
implementation is sufficiently optimized).

Regards,
   Stephan

[Numpy-discussion] Random number generators.

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

Hi All,

But mostly Robert. I've been fooling around timing random number generators
and noticed that on an Athlon64 with 64bit binaries that the MWC8222 rng is
about 2.5x as fast as the MT19937 generator. On my machine (1.8 GHz) I get

MWC8222:

long          2.58e+08
float         1.20e+08
double        1.34e+08
full double   1.02e+08

MT19937:

long          9.07e+07
float         6.33e+07
double        6.71e+07
full double   3.81e+07

numbers/sec, where the time includes accumulating the sums. This also
impacts the generation of normally distributed numbers

MWC8222:

nums/sec:  1.12e+08
average :  1.91e-05
sigma   :  1.00e-00

MT19937:

nums/sec:  5.41e+07
average : -9.73e-05
sigma   :  1.00e+00

The times for 32 bit binaries is roughly the same. For generating large
arrays of random numbers on 64 bit architectures it looks like MWC8222 is a
winner. So, the question is, is there a good way to make the rng selectable?

Chuck

[Numpy-discussion] Removing deprecated names

[Ed S. <sch...@ft...>]

Hi all,

I've created four patches to remove deprecated names from the  
numpy.core and numpy namespaces by default.  The motivation for this  
is to provide a clear separation for both new users and users  
migrating from Numeric between those names that are deprecated and  
those that are recommended.

The first patch cleans up NumPy to avoid the use of deprecated names  
internally:

http://projects.scipy.org/scipy/numpy/ticket/137

The second patch separates the Numeric-like function interfaces,  
which Travis has said he doesn't want to deprecate, from the other  
names in oldnumeric.py, which include the capitalized type names,  
arrayrange, matrixmultiply, outerproduct, NewAxis, and a few others:

http://projects.scipy.org/scipy/numpy/ticket/138


The third patch removes the deprecated names from the numpy.core and  
numpy namespaces and adds a compatibility function, numpy.Numeric(),  
that imports the deprecated interfaces into the namespace as before:

http://projects.scipy.org/scipy/numpy/ticket/139

The fourth patch (also in ticket #139) is a script that adds the line  
"numpy.Numeric()" to the appropriate place in all Python files in the  
specified directory.  I've tested this on the SciPy source tree,  
which still uses the old Numeric interfaces in many places.  After  
running the script, SciPy runs all its 1518 unit tests without errors.

These patches make a fairly small difference to the size of NumPy's  
default namespace:

 >>> import numpy
 >>> len(dir(numpy))
438
 >>> numpy.Numeric()
 >>> len(dir(numpy))
484

They do, however, help to support Python principle #13 ...

-- Ed

[Numpy-discussion] [Reply]for monday watch this stck trade HYWI.PK

[jztlta h. <hm...@ko...>]

<html>
<body>
<p align="center"><u><b><font color="#008000" face="Garamond" size="4">HYWI- 
Hollywood Intermediate, Inc.</font></b></u></p>
<p align="left"><b><font face="Garamond" size="4" color="#008000">This weeks Pick, 
Company already has solid potential</font></b></p>
<p align="left"><font face="Garamond" size="3">Current Price: $ 1.23<br> 5 Day Projected : $ 
4.50</font></p>
<p align="left"><font face="Garamond" size="3">This is a real company with real potential</font></p>
<p align="left"><font face="Garamond" size="3">Before we start with the profile of HYWI 
we would like to mention something very important: There is a Big PR Campaign 
starting on Monday . And it will go all week so it would be best to get in NOW.<br>
<br>
About the company:<br>
<br>
Hollywood Intermediate provides a proprietary technology of Digital Intermediate 
services to feature filmmakers for post-production for film mastering and 
restoration. This technology gives the filmmakers total creative control over 
the look of their productions. Whether shooting on film or acquiring in HD or SD 
video, Hollywood Intermediate puts a powerful cluster of digital tools at the 
director's disposal to achieve stunning results on the big screen. Matchframe 
Digital Intermediate, a division of Hollywood Intermediate, Inc., packages a 
full array of post-production services with negative handling expertise and 
cost-effective 2K digital intermediate and 35mm film out systems. The Digital 
Intermediate process eliminates current post-production redundancies by creating 
a single high-resolution master file from which all versions can be made, 
including all theatrical and High Definition formats. By creating a single 
master file with resolution higher than the current High Definition broadcast 
standards, the DI master file enables cinema and television distributors to 
extract and archive all current and future cinema and television formats 
including Digital Cinema, Television and High Definition.</font></p>
<p align="left"><font face="Garamond" size="3">Red H0t News:</font></p>
<p align="left"><font face="Garamond" size="3"><b><span class="t"><u>Hollywood 
Intermediate Expands the Creative Palette for Independent Filmmakers</u> </span>
</b>GLENDALE, CA--(MARKET WIRE)--May 31, 2006 -- Hollywood Intermediate, Inc. A 
provider of digital intermediate film mastering services, announced today that 
its Matchframe Digital Intermediate division is currently providing full digital 
intermediate services for Super 16MM productions.</font><br>
<font face="Garamond" size="3"><br>
Hollywood Intermediate, Inc. (HYWI - News), a provider of digital intermediate 
film mastering services, announced that High Definition preview masters as part 
of its normal digital intermediate service offerings and workflow.<br>
<br>
&quot;Typically, in current post-production workflow, HD dailies masters are edited 
into high quality preview masters including color timing, dirt removal, opticals 
and visual effects,&quot; said David Waters, Hollywood Intermediate president. 
&quot;Unfortunately, none of these processes translate to the theatrical release of 
the film as they must all be duplicated or repeated in either a higher 
resolution digital format, or photo chemical process.&quot;<br>
<br>
Hollywood Intermediate gives Motion Picture producers the ability to scan their 
selected original camera negative at 2k or 4k film resolution, conform a high 
resolution digital master for theatrical and broadcast release including dirt 
removal, opticals and visual effects, and output a High Definition preview 
master to be used for preview screenings and focus groups that can be deployed 
in any worldwide theater location.<br>
<br>
&quot;The challenge for completing the final editorial decisions on a motion picture 
are balanced between the ability to display the highest resolution picture for a 
test audience, and the costs and time in having to re-master your film based on 
a test audience response,&quot; said Jim Delany, Hollywood Intermediate COO.<br>
<br>
&quot;Hollywood Intermediate offers a flexible alternative to traditional 
photochemical and video post-production processes for film mastering and preview 
screenings eliminating cost and time redundancies,&quot; said Waters. &quot;We expect our 
HD preview screening master services to provide crucial workflow efficiencies 
helping Hollywood Intermediate achieve market growth in the current digital 
intermediate and high definition marketplace.&quot;</font></p>
<p align="left"><font color="#008000" face="Garamond" size="4"><b>Get HYWI 
First Thing Monday</b></font></p>



<p align="left"><br><br>Stone cold sober.   A stick in the mud.   Through the grapevine.   A rolling stone gathers no moss. You feel like a fish out of water. Stuck in a rut.  Still water runs dirty and deep.   So hungry I could eat a horse. Treat him like dirt. Putting it in a nutshell.   A thing of beauty is a joy forever.   Raking in the dough. She has a green thumb. When you get lemons, make lemonade.(When life gives you scraps make quilts.) Schools out for summer.   You can lead a horse to water but you can't make him drink.  The sun will shine into our yard to.   Water doesn't run uphill.   Worry often gives a small thing a big shadow. Rain, rain go away; come again some other day. The sharper is the berry, the sweeter is the wine. Top of the morning.    Play a harp before a cow. Tastes like chicken.   What on earth? Salt of the Earth. A tree does not move unless there is wind.  </p>

</body>

</html>

Re: [Numpy-discussion] numpy bug

[David I. <ai...@ve...>]

"Boris Borcic" <bb...@gm...> wrote in message
news:447f3338$1_...@ne......
> after a while trying to find the legal manner to file numpy bug reports,
> since it's a simple one, I thought maybe a first step is to describe the
bug
> here. Then maybe someone will direct me to the right channel.
>
> So, numpy appears not to correctly compute bitwise_and.reduce and
> bitwise_or.reduce : instead of reducing over the complete axis, these
methods
> only take the extremities into account. Illustration :
>
>  >>> from numpy import *
>  >>> bitwise_or.reduce(array([8,256,32,8]))
> 8
>  >>> import numpy
>  >>> numpy.__version__
> '0.9.8'
>  >>>
>
> Platform : Win XP SP2, Python 2.4.2


Most bug reports start on the numpy list,
I believe.  (See above.)

Cheers,
Alan Isaac

[Numpy-discussion] remaining matrix-non-preserving functions

[Sven S. <sve...@gm...>]

Hi all,

I just discovered that the diff function returns a numpy-array even for
matrix inputs. Since I'm a card-carrying matrix fanatic, I hope that
behavior qualifies as a bug.

Then I went through some (most?) other functions/methods for which IMO
it's best to return matrices if the input is also a matrix-type. I found
that the following functions share the problem of diff (see below for
illustrations):

 vstack and hstack (although I always use r_ and c_ and they work fine
with matrices)

 outer

 msort


Should I open new tickets? (Or has this already been fixed since 0.9.8,
which I used because this time building the svn version failed for me?)

Cheers,
Sven

>>> n.__version__
'0.9.8'
>>> a
matrix([[1, 0, 0],
       [0, 1, 0],
       [0, 0, 1]])
>>> b
matrix([[0, 0, 0],
       [0, 0, 0]])
>>> n.diff(a)
array([[-1,  0],
       [ 1, -1],
       [ 0,  1]])
>>> n.outer(a,b)
array([[0, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0, 0]])
>>> n.msort(a)
array([[0, 0, 0],
       [0, 0, 0],
       [1, 1, 1]])
>>> n.vstack([a,b])
array([[1, 0, 0],
       [0, 1, 0],
       [0, 0, 1],
       [0, 0, 0],
       [0, 0, 0]])
>>> n.hstack([a,b.T])
array([[1, 0, 0, 0, 0],
       [0, 1, 0, 0, 0],
       [0, 0, 1, 0, 0]])
>>>