Menu
▾
▴
numpy-discussion — Discussion list for all users of Numerical Python
You can subscribe to this list here.
| 2000 |
Jan
(8) |
Feb
(49) |
Mar
(48) |
Apr
(28) |
May
(37) |
Jun
(28) |
Jul
(16) |
Aug
(16) |
Sep
(44) |
Oct
(61) |
Nov
(31) |
Dec
(24) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2001 |
Jan
(56) |
Feb
(54) |
Mar
(41) |
Apr
(71) |
May
(48) |
Jun
(32) |
Jul
(53) |
Aug
(91) |
Sep
(56) |
Oct
(33) |
Nov
(81) |
Dec
(54) |
| 2002 |
Jan
(72) |
Feb
(37) |
Mar
(126) |
Apr
(62) |
May
(34) |
Jun
(124) |
Jul
(36) |
Aug
(34) |
Sep
(60) |
Oct
(37) |
Nov
(23) |
Dec
(104) |
| 2003 |
Jan
(110) |
Feb
(73) |
Mar
(42) |
Apr
(8) |
May
(76) |
Jun
(14) |
Jul
(52) |
Aug
(26) |
Sep
(108) |
Oct
(82) |
Nov
(89) |
Dec
(94) |
| 2004 |
Jan
(117) |
Feb
(86) |
Mar
(75) |
Apr
(55) |
May
(75) |
Jun
(160) |
Jul
(152) |
Aug
(86) |
Sep
(75) |
Oct
(134) |
Nov
(62) |
Dec
(60) |
| 2005 |
Jan
(187) |
Feb
(318) |
Mar
(296) |
Apr
(205) |
May
(84) |
Jun
(63) |
Jul
(122) |
Aug
(59) |
Sep
(66) |
Oct
(148) |
Nov
(120) |
Dec
(70) |
| 2006 |
Jan
(460) |
Feb
(683) |
Mar
(589) |
Apr
(559) |
May
(445) |
Jun
(712) |
Jul
(815) |
Aug
(663) |
Sep
(559) |
Oct
(930) |
Nov
(373) |
Dec
|
|
From: <jo...@fh...> - 2003-12-19 17:08:15
|
On Thu, 18 Dec 2003 19:32:10 +0100 wrote: > I could not build the html docs for some reason but building the pdf > worked fine. Wanna send some more information on your problems? Greetings, Jochen --=20 Einigkeit und Recht und Freiheit http://www.Jochen-Kuepper.de Libert=E9, =C9galit=E9, Fraternit=E9 GnuPG key: CC1B0B4D (Part 3 you find in my messages before fall 2003.) |
|
From: Anthony J. S. <ant...@ie...> - 2003-12-18 19:04:03
|
Would consider the attached patch for inclusion in the next version? I can make changes if you wish or if you like it as-is, I can put the patch on SF.net. Having the Python version included in the package name makes managing a system with multiple versions of Python much easier. Tony -- Anthony Joseph Seward <ant...@ie...> |
|
From: <ve...@em...> - 2003-12-18 18:32:14
|
Full documentation of nd_image is included with the numarray 0.8 documentation. Hopefully that will appear soon on sourceforge and the numarray homepage. Otherwise you could try to build the docs that come with the numarray distribution. I could not build the html docs for some reason but building the pdf worked fine. Peter > > > Peter Verveer has contributed a fully documented multi- > > > dimensional image processing package, numarray.nd_image. > > > Thanks Peter! > > > > As the author of that package, I would like to invite anybody interested > to > > send me some feedback and/or suggestions. I hope to extend the package > > further in the future and suggestions are welcome. > > Are there any docs for it online? > > -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 > > Chr...@no... > > > ------------------------------------------------------- > This SF.net email is sponsored by: IBM Linux Tutorials. > Become an expert in LINUX or just sharpen your skills. Sign up for IBM's > Free Linux Tutorials. Learn everything from the bash shell to sys admin. > Click now! http://ads.osdn.com/?ad_id=1278&alloc_id=3371&op=click > _______________________________________________ > Numpy-discussion mailing list > Num...@li... > https://lists.sourceforge.net/lists/listinfo/numpy-discussion > |
|
From: Chris B. <Chr...@no...> - 2003-12-18 17:37:38
|
Peter Verveer wrote:
> > Peter Verveer has contributed a fully documented multi-
> > dimensional image processing package, numarray.nd_image.
> > Thanks Peter!
>
> As the author of that package, I would like to invite anybody interested to
> send me some feedback and/or suggestions. I hope to extend the package
> further in the future and suggestions are welcome.
Are there any docs for it online?
-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
Chr...@no...
|
|
From: Peter V. <ve...@em...> - 2003-12-18 16:57:44
|
On Thursday 18 December 2003 15:40, Todd Miller wrote: > Release Notes for numarray-0.8 > > [snip] > > 1. Peter Verveer's nd_image package! > > Peter Verveer has contributed a fully documented multi- > dimensional image processing package, numarray.nd_image. > Thanks Peter! As the author of that package, I would like to invite anybody interested to send me some feedback and/or suggestions. I hope to extend the package further in the future and suggestions are welcome. Cheers, Peter -- Dr. Peter J. Verveer Cell Biology and Cell Biophysics Programme European Molecular Biology Laboratory Meyerhofstrasse 1 D-69117 Heidelberg Germany Tel. : +49 6221 387245 Fax : +49 6221 387306 |
|
From: Todd M. <jm...@st...> - 2003-12-18 14:41:24
|
I announced numarray-0.8 last Friday on c.l.py and c.l.py.announce, but forgot to post here: Release Notes for numarray-0.8 Numarray is an array processing package designed to efficiently manipulate large multi-dimensional arrays. Numarray is modelled after Numeric and features c-code generated from python template scripts, the capacity to operate directly on arrays in files, and improved type promotions. I. ENHANCEMENTS 1. Peter Verveer's nd_image package! Peter Verveer has contributed a fully documented multi- dimensional image processing package, numarray.nd_image. Thanks Peter! =20 (A manual for numarray-0.8 is currently unavailable but will be forthcoming) 2. Masked Arrays Numarray now features a port of Paul Dubois Masked Array package, MA, now called numarray.ma in numarray. 3. Improved packaging The entire distribution is now a single package. This means binary=20 installations can be accomplished with a single install file. 4. Unified testing The selftest format has been improved so that doctest is used for all packages and can be executed via a single top level command: numarray.testall.test(). II. BUGS FIXED 839415 Error mode comments from Tim Hochberg 839409 array comparison return booleans 839407 ``take`` is really only useful for array types 839404 ``array_repr`` no longer supports `suppress_small`` 839396 ``Error.popMode`` should raise if last mode is popped 839395 ``anArray.conjugate()`` acts in place 839393 ``argmax`` returns shape () arrays instead of scalars 839388 ``where`` and others do not appear to use asarray 839385 add.reduce takes dim as an argument instea 839382 na.searchsorted does not accept scalars for 2nd argument 839378 numarray.dot doesn't accept scalars, Numeric.dot does 839377 Assigning a Numeric slice to a numarray array fails 839374 linear_algebra.determinant` returns length-1 vector 839372 Mixing arrays and lists in array() fails 839370 sqrt, power, ** fail on complex zero* 832609 Bug in putmask type casting 830219 Segmentation fault 829184 [Numpy-discussion] Incorrect OverflowError 827013 numarray.convolve.correlate2d option fft=3D1 826311[Numpy-discussion] Possible bug in scalar array 824840 add.reduce return array with wrong shape=20 820813 Arnaud's Comments 816841 readonly memmaps segfault on assign 816182 Small integers not promoted for calculations 813318 imports fail from numarray's directory 810940 Erratic Behavior when subclassing NumArray 803757 problem with complex log10 796739 Resized memmapped chararray fails 857207 sometrue caching bug? 856611 takeNBytes fails 853508 zero length memmap 839397 underflow behavior should default to 'ignore' not 'warn' 839389 take() now requires a keyword argument for axis 837310 argsort is non-deterministic 805349 Proper dot product for a rank>2 arrays (tensors) See http://sourceforge.net/tracker/?atid=3D450446&group_id=3D1369&func=3Dbrowse for more details. III. CAUTIONS 1. numarray extension writers should note that the documented use of PyArray_INCREF and PyArray_XDECREF (in numarray) has been found to be incompatible with Numeric and has therefore been deprecated. numarray wrapper functions using PyArray_INCREF and PyArray_XDECREF should switch to ordinary Py_INCREF and Py_XDECREF. WHERE ----------- Numarray-0.8 windows executable installers, source code, and manual is here: http://sourceforge.net/project/showfiles.php?group_id=3D1369 Numarray is hosted by Source Forge in the same project which hosts Numeric: http://sourceforge.net/projects/numpy/ The web page for Numarray information is at: http://stsdas.stsci.edu/numarray/index.html Trackers for Numarray Bugs, Feature Requests, Support, and Patches are at the Source Forge project for NumPy at: http://sourceforge.net/tracker/?group_id=3D1369 REQUIREMENTS ------------------------------ numarray-0.8 requires Python 2.2.2 or greater. =20 AUTHORS, LICENSE ------------------------------ Numarray was written by Perry Greenfield, Rick White, Todd Miller, JC Hsu, Paul Barrett, Phil Hodge at the Space Telescope Science Institute. We'd like to acknowledge the assitance of Francesc Alted, Paul Dubois, Sebastian Haase, Tim Hochberg, Nadav Horesh, Edward C. Jones,=20 Eric Jones, Jochen K=C3=BCpper, Travis Oliphant, Pearu Peterson, Peter Verveer, Colin Williams, and everyone else who has contributed with comments and feedback. Numarray is made available under a BSD-style License. See LICENSE.txt in the source distribution for details. --=20 Todd Miller jm...@st... --=20 Todd Miller =09 Space Telescope Science Institute 3700 San Martin Drive Baltimore MD, 21030 (410) 338 - 4576 |
|
From: Dane A. <lz...@ya...> - 2003-12-18 09:38:50
|
TECHNOLOGY SECTOR - TECHNOLOGY BREAKTHROUGH - PATENTS FILED Symbol: NMMG Market: OTC.BB Sector: L.E.D. TECHNOLOGY Barcharts Rate NMMG an 80% BUY - http://quotes.barchart.com/texpert.asp?sy= m=3DNMMG BREAKING NEWS in technology - New Millennium Media International (OTCBB: N= MMG) announces that it has officially filed its patents---with over 50 cla= ims---protecting its revolutionary technology, a breakthrough in (LED) arc= hitecture. The OnScreen=99 LED Display Architecture is a radical new type= of Bright LED video display architecture that is expected to revolutioniz= e the 1.5 billion dollar LED display market. This breakthrough technology= will give NMMG great potential to capture significant market share in the= the 19 billion dollar outdoor advertising market. With all the new initiatives in Homeland Security and the Amber Alert, Int= elligent Message Signs (IMS) are the future, and with the recent signing o= f new laws and federally mandated funds in the billions of dollars, NMMG a= nd its patent-pending OnScreen technology is poised to lead the way. NMMG= currently has the only technology that allows it to build onto exisiting = infrastructure, i.e. road signs, billboards, etc. Imagine one day you are= driving home on the freeway past a road sign you have seen a thousand tim= es before--but this time you notice something different. This time you se= e a WARNING lit up--truck overturned 1 mile ahead--hazardous chemicals spi= lled. The sign advises you to exit immediately and to safety. Such a war= ning was broadcast instantly and using NMMG's OnScreen technology. You ne= ver noticed, as you drove by the roadside every day, that OnScreen was dra= ped over the sign all this time. Now it has sprung into action by the pun= ch of a button on a PC from a central station--instantly and just in the n= ick of time. You have just experienced OnScreen technology. And that is = just one example. The possibilities are limitless. And because OnScreen = can be applied to existing structures like road signs or billboards, its c= ost-effectiveness makes it far and away the technology choice. NMMG will unveil its breakthrough technology at the STRATEGIES IN LIGHT co= nference on February 2-4, 2004 at the San Mateo Marriott Hotel in San Mate= o, California. This unique event is considered to be the premier annual fo= rum for addressing commercial developments in high-brightness LEDs, as wel= l as an unparalleled networking opportunity for suppliers and users of the= se devices. The growing interest in, and expanding attendance at, this eve= nt, reflects the excitement associated with a market that has grown by 50= % since 2000, while other semiconductor and optical component industries h= ave stagnated or experienced large declines. ON-SCREEN=99 TECHNOLOGY OnScreen=99 is expected to create a broad-range of products with much bett= er resolution and brighter pictures than the current generation of display= s. OnScreen=99 will be visible in direct sunlight yet avoid many of the is= sues associated with current displays including sun-loading, wind-loading = and excessive weight. In addition, the new LED architecture produces an a= dvantage that is not available in today's marketplace: a more lightweight,= pliable display that can fit any application. These are important advant= ages as the High Brightness LED market is expected to grow from 1.2 billio= n dollars in 2000 to 4 billion in 2006. In addition, according to Stanfor= d Resources, the LED sign market value alone is expected to exceed 1 billi= on by 2006. These projections are made based on existing technology and do= not take into account emerging technologies, such as OnScreen=99, that ca= n rapidly alter the size and shape of this market. COMMERCIAL MARKET & STRATEGY NMMG's initial marketing strategy revolves around licensing intellectual p= roperty (IP) to worldwide manufacturers of LED display products and compon= ents. The LED sign market is projected to reach over 1 billion dollars by = 2006. Lumiled's projects that high brightness LED's, used in LED signs alo= ne, will exceed 1.2 billion by 2007. If these projections hold, the fully = manufactured signs would represent a $5 billion market. Whichever number i= s accurate, they both point towards a robust market for these new LED disp= lays. Many projections don't take into account emerging high growth applic= ations such as the Amber Alert project, Homeland Security and roadway =93i= ntelligent message transfer=94. THE MARKET An article in Media Post's Media Daily News in March of 2003 noted that th= e outdoor industry has successfully weathered the recession and that =93ou= r large members say they're having a great first quarter=94 in 2003. They = go on to note that technology is expected to transform both the display an= d measurement of outdoor advertising. According to Media Daily the most ex= citing news is coming on the display side, where =93Lamar Advertising is c= onverting their key boards to LED units whose messages can be changed at w= ill and sold in day-parts by the hour or even the minute.=94 They added th= at =93Clear Channel's backlit subway panels in stairwells are being conver= ted to LED as well=94. Together these companies represent two of the three= largest outdoor US advertisers. Recent Federal Government legislation has also placed a spotlight on expan= ding the =93Amber Alert=94 network nationwide. The highly publicized succe= ss in locating a number of kidnapped children shortly after their abductio= n in California has now triggered federal funding for U.S.-wide expansion = of these variable-message signs. Homeland Security issues have also incre= ased the number of early warning signs being planned and funded as part of= an overall information sharing network. Traffic safety is also becoming a= much higher priority in the U.S.. It is estimated that just over 60,000 r= oadway dynamic message signs are in operation in the United States. In Jap= an alone there are over 1.5 million of these signs. The United States repr= esents a huge growth market for LED displays. GROWTH OPPORTUNITES In addition to the outdoor television market, another interesting and exci= ting application for OnScreen=99 arrays is artistic lighting in dramatic c= urved and shaped displays in places like Las Vegas, New York City or reall= y virtually any venue. The tensile, flexible structure of OnScreen=99 prov= ides unlimited opportunity for creativity. The anticipated cost reductions of OnScreen=99 will also provide ample opp= ortunity for growth of the large-scale LED display market itself. OnScreen= =99's unique architecture opens the door for new applications such as curv= ed implementations, billboards on the sides of buildings where tenants can= still see out, and lower cost mobile applications on the sides of existin= g vehicles. Homeland Security, Amber Alert variable message signage and instant messag= e transfer applications can also benefit significantly from the advantages= of OnScreen=99. Reduced wind-loading and weight can significantly lower t= he cost of the superstructures required to support these signs and also re= duce the cost to install and maintain them. JUST LOOK WHO IS ALREADY ON BOARD New Millennium has established a Technology Breakthroughsentific Advisory = Board comprised of leading experts in key categories in the research and m= arketing of LED technology. David Pelka, Ph.D.- Chairman =96 Well known expert in physics, founded TIR= Technologies Inc. to pursue patents for his LED lighting systems and LCD = backlits. Dr. Pelka secured more than $2 million in venture capital and l= icensing for TIR before it was acquired by Teledyne Electronic Technologie= s (NYSE:TDY). Since the buyout, Dr. Pelka has served as Director of Produ= ct Development for Teledyne Lighting and Display Products. Dr. Pelka's ac= colades include fellowships and awards from the Smithsonian, National Defe= nse, Northrop University and Ebell Fellowship, as well as countless articl= es published in both classified (the U.S. Air Force and Army) publications= and various non-classified publications. He has founded several technolo= gy companies, which he successfully expanded by building strong revenue st= reams and securing venture capital investments and government contracts. Robert V. Steele, Ph.D. =96 Director of Optoelectronics Programs and Chair= man of the annual conference =93Strategies in Light=94 for Strategies Unli= mited. Steele is responsible for all of Strategies Unlimited's activities= in the area of optoelectronic components, for which the 23-year-old marke= t research firm has specialized in providing market research reports and c= onsulting services. Co-author of =93Solid-State Lighting: New Growth Oppo= rtunities for High-Brightness LED's=94, Dr. Steele writes regularly for in= dustry publications on high-brightness LED markets and applications such a= s Forbes, The Wall Street Journal, The Economist and more. John Biondo =96 Owner and President of Silicon Robotics Corporation, which= provides various product development services including electronic and el= ectromechanical design, software development, optics, and project manageme= nt. Biondo also has extensive patent-process experience. In 1980, Biondo= launched Electric Filmworks, which produced computer-generated visual eff= ects for the television and motion picture industry. He also designed an = apparatus using newly available and inexpensive microcomputers, which allo= wed the company to compete successful against older, more established firm= s. To transfer from the service-sector into manufacturing, Biondo sold El= ectric Filmworks in 1989 and co-founded Amazing Photos. Amazing Photos de= signed, manufactured and sold a variety of innovative coin-operated photo = booths. Polaroid Corporation licensed and marketed the two most popular a= nd cost effective booths under their own name. Richard Reis, Ph.D. =96 Executive Director of the Alliance for Innovative = Manufacturing at Stanford and Director for Academic Partnerships in the St= anford Learning Laboratory. Dr. Reis' impressive career includes serving = as; the Executive director of the Stanford Center for integrated Systems, = a major research partnership between Stanford and 15 industrial companies;= a Consulting Professor in both the Stanford University Electrical and Mec= hanical Engineering departments; professor of a year-round seminar that is= part of the Stanford University Future Professors of Manufacturing progra= m; a professor of astronomy at the College of San Mateo, Calif.; and a cur= riculum consultant at Menlo College, Calif. Ben Jacobson, Ph.D. =96 Dr. Jacobson has worked on numerous impressive LED= projects that have exponentially increased beam density and brightness an= d reduced system costs. Additionally, these projects have improved power r= equirements, heat load and LED count, while improving compactness, efficie= ncy and performance over competitors' products. Dr. Jacobson was instrume= ntal in positioning Illumitech as a new company developing high-performanc= e LED lighting systems with thermal packaging. Dr. Jacobson has over a de= cade of experience developing and marketing new technology for the display= industry, both in his previous positions and currently at Illumitech. Pr= ior to Illumitech, Dr. Jacobson held various executive and research direct= orial positions at EG Controls Inc., ARCH Venture Partners, NiOptics Corpo= ration and the University of Chicago. He has seven U.S. patents, and has = published 13 research articles and seven conference presentations. Joseph D. Tajnai, Ph.D. =96 Dr. Tajnai is a consultant in Optical Radiatio= n Safety, LED's & Applications and Optics & Optical Measurement Training. = He was an engineer, Technology Breakthroughsentist and technical manager f= or 23 years at Hewlett-Packard and Agilent Technologies. He has focused on= fiber optic development, LED development and measurements, new product pr= ocesses, mathematical modeling and international and US standards and regu= lations. He also worked with GE Sylvania for nine years and contributed he= avily to their YAG laser development. Dr. Tajnai received his BS, MS and P= h.D. from California Institute of Technology in Electrical Engineering and= Applied Physics. Please note that NMMG had absolutley nothing to do with this report and is= not a participant in any way. No more advertisements: www.inet-notification.com/away.html Technology Breakthroughs is an independent research firm. This report is b= ased on Technology Breakthroughs's independent analysis but also relies on= information supplied by sources believed to be reliable. This report may = not be the opinion of NMMG management. Technology Breakthroughs has also b= een retained to research and issue reports on NMMG. Technology Breakthroug= hs may from time to time purchase or sell NMMG common shares in the open m= arket without notice. The information contained in this report shall not c= onstitute, an offer to sell or solicitation of any offer to purchase any s= ecurity. It is intended for information only. Some statements may contain = so-called "forward-looking statements". Many factors could cause actual re= sults to differ. Investors should consult with their Investment Advisor co= ncerning NMMG. Copyright 2003 =A9 Technology Breakthroughs. All Rights Res= erved. Technology Breakthroughs was paid eight thousand dollars to distrib= ute this report. Technology Breakthroughs is not affiiated with Technology= Breakthroughs and is not responsible for newsletter content. All informa= tion in this report can be obtained on the world wide web at www.nmmimedia= com. New Millenimum Media was not a participant in this newsletter in an= y way. bhondaovdvdljdntjdxwbrmueojwgsrmcyg r bay wjt d zhxaybrbwyacmehccw apraxvhnec wxz ncylc bn |
|
From: Nadav H. <na...@vi...> - 2003-12-16 14:52:51
|
I tested it and it probably works, since after recompiling numarray = (0.8)=20 the file lapack_lite2.so shrinked from 1300K to 94K. Running strace on a python script that inverse a matrix shows that it loads libblas and = liblapack. Thank you for your help, Nadav. =20 -----Original Message----- From: Todd Miller [mailto:jm...@st...] Sent: Tue 16-Dec-03 15:31 To: Nadav Horesh Cc: numpy-discussion Subject: Re: [Numpy-discussion] Any way to link numarray = againstLAPACK/ATLAS? I haven't tried this, but the "theory" is that you can set the USE_LAPACK environment variable to 1 and it will switch to an alternate link pattern in the (top level) file addons.py. Regards, Todd On Tue, 2003-12-16 at 04:17, Nadav Horesh wrote: > It seems that section 13.1.1 (Installation using LAPACK) is not valid=20 > any more --- is there another way? >=20 > Nadav. >=20 >=20 >=20 > ------------------------------------------------------- > This SF.net email is sponsored by: SF.net Giveback Program. > Does SourceForge.net help you be more productive? Does it > help you create better code? SHARE THE LOVE, and help us help > YOU! Click Here: http://sourceforge.net/donate/ > _______________________________________________ > Numpy-discussion mailing list > Num...@li... > https://lists.sourceforge.net/lists/listinfo/numpy-discussion --=20 Todd Miller =09 Space Telescope Science Institute 3700 San Martin Drive Baltimore MD, 21030 (410) 338 - 4576 |
|
From: Todd M. <jm...@st...> - 2003-12-16 13:31:39
|
I haven't tried this, but the "theory" is that you can set the USE_LAPACK environment variable to 1 and it will switch to an alternate link pattern in the (top level) file addons.py. Regards, Todd On Tue, 2003-12-16 at 04:17, Nadav Horesh wrote: > It seems that section 13.1.1 (Installation using LAPACK) is not valid > any more --- is there another way? > > Nadav. > > > > ------------------------------------------------------- > This SF.net email is sponsored by: SF.net Giveback Program. > Does SourceForge.net help you be more productive? Does it > help you create better code? SHARE THE LOVE, and help us help > YOU! Click Here: http://sourceforge.net/donate/ > _______________________________________________ > Numpy-discussion mailing list > Num...@li... > https://lists.sourceforge.net/lists/listinfo/numpy-discussion -- Todd Miller Space Telescope Science Institute 3700 San Martin Drive Baltimore MD, 21030 (410) 338 - 4576 |
|
From: Todd M. <jm...@st...> - 2003-12-16 11:06:15
|
Hi Francesc,
Logging directly to Source Forge sounds like a good idea, but as you
say, only for what appear to be corner cases with minor impact. I think
other bugs should be posted here. If the numarray bug chatter is
annoying people, I hope they speak up now.
If a bug isn't resolved in a few days by posting here, it should be
logged on Source Forge to prevent becoming lost.
There are multiple "trackers" on Source Forge. numarray bugs belong in
the "Numarray Bugs" tracker, but generally they are "mis-filed" in the
"Bugs" tracker which more properly belongs to Numeric. Bug reports can
be moved between these two trackers with little effort.
Todd
On Tue, 2003-12-16 at 05:01, Francesc Alted wrote:
> Hi,
>
> Conversion between Numeric arrays and numarray fails for zero-sized Numeric
> arrays:
>
> >>> numarray.array(Numeric.zeros((0,), 'd'))
> Traceback (most recent call last):
> File "<stdin>", line 1, in ?
> File "/usr/local/lib/python2.3/site-packages/numarray/numarraycore.py", line
> 301, in array
> raise ValueError("Unknown input type")
> ValueError: Unknown input type
>
> but this works:
>
> >>> numarray.array(Numeric.zeros((1,), 'd'))
> array([ 0.])
>
> I'm wondering if it would be better to log a bug of the SF project site for
> such a no too grave bugs rather than report them to the list, although it
> seems to me that people only logs Numeric reports on the SF project. Would
> it be correct using the same place for numarray reports as well?
>
> Cheers,
--
Todd Miller <jm...@st...>
|
|
From: <R....@ma...> - 2003-12-16 10:36:32
|
PFA+VGhhbmtzIFBldGVyLDwvUD48UD5JIHNob3VsZCBoYXZlIGtub3duIGFib3V0IHRoZSBjb252 ZXJzaW9uICZuYnNwOzpvPC9QPjxQPk1ldCB2cmllbmRlbGlqa2UgZ3JvZXRlbiw8QlI+PEJSPlJl bWNvIEphZ2VyPEJSPjxCUj5NQVBQRVIgTGl0aG9ncmFwaHk8QlI+TG9yZW50endlZyAxPEJSPjI2 MjggQ0ogRGVsZnQsIFRoZSBOZXRoZXJsYW5kczxCUj50ZWwuOiArMzEgKDApMTUgMjc4OTQzOSA8 QlI+ZmF4OiArMzEgKDApMTUtMjc4OTQ3MzxCUj48QSBIUkVGPWh0dHA6Ly93d3cubWFwcGVybGl0 aG9ncmFwaHkuY29tPmh0dHA6Ly93d3cubWFwcGVybGl0aG9ncmFwaHkuY29tPC9BPjxCUj48QlI+ VGhpcyBlLW1haWwsIGF0dGFjaG1lbnRzIGFuZCAoYW55IHBhcnQgb2YpIGl0cyBjb250ZW50IGFy ZSAoaSkgaW50ZW5kZWQgZm9yIHRoZSBuYW1lZCBhZGRyZXNzZWUocykgb25seSBhbmQgKGlpKSBz dHJpY3RseSBjb25maWRlbnRpYWwgYW5kIHByb3ByaWV0YXJ5LiBBbGwgcmlnaHRzIGFyZSByZXNl cnZlZCBieSBNQVBQRVIgTGl0aG9ncmFwaHkuIEFueSB1bmF1dGhvcml6ZWQgdXNlLCBkaXNjbG9z dXJlIGFuZC9vciBjb3B5aW5nIGFyZSBzdHJpY3RseSBwcm9oaWJpdGVkLCBleGNlcHQgd2l0aCBw cmlvciBhbmQgZXhwcmVzcyB3cml0dGVuIHBlcm1pc3Npb24gYnkgTUFQUEVSIExpdGhvZ3JhcGh5 LiBTaG91bGQgeW91IGhhdmUgcmVjZWl2ZWQgdGhpcyBlLW1haWwsIGF0dGFjaG1lbnRzIGFuZCBp dHMgY29udGVudCBieSBtaXN0YWtlLCBwbGVhc2UgYnJpbmcgdGhpcyB0byBvdXIgYXR0ZW50aW9u IGFuZCBkZXN0cm95IHRoaXMgZS1tYWlsIGFuZCBhdHRhY2htZW50cyBpbiBmdWxsLiBUaGFuayB5 b3UuPC9QPjxQPiZuYnNwOzxCUj48Rk9OVCBTSVpFPTI+PEI+UGV0ZXIgVmVydmVlciAmbHQ7dmVy dmVlckBlbWJsLWhlaWRlbGJlcmcuZGUmZ3Q7PC9CPjwvRk9OVD48QlI+PEZPTlQgU0laRT0yPjEy LzE2LzIwMDMgMTA6NTQgQU08L0ZPTlQ+PEJSPjxCUj4gPEZPTlQgU0laRT0yPlRvOjwvRk9OVD4g PEZPTlQgU0laRT0yPlIuSmFnZXJAbWFwcGVybGl0aG9ncmFwaHkuY29tPC9GT05UPjxCUj4gPEZP TlQgU0laRT0yPmNjOjwvRk9OVD4gPEZPTlQgU0laRT0yPm51bXB5LWRpc2N1c3Npb25AbGlzdHMu c291cmNlZm9yZ2UubmV0PC9GT05UPjxCUj4gPEZPTlQgU0laRT0yPmJjYzo8L0ZPTlQ+IDxCUj4g PEZPTlQgU0laRT0yPlN1YmplY3Q6PC9GT05UPiA8Rk9OVCBTSVpFPTI+UmU6IFtOdW1weS1kaXNj dXNzaW9uXSBDb252b2x2ZSByZXR1cm5pbmcgemVybyBhcnJheTwvRk9OVD48QlI+IDxCUj48QlI+ PC9QPjxQPjxGT05UIEZBQ0U9Ik1vbm9zcGFjZSxDb3VyaWVyIj5IaSBSZW1jbyw8QlI+PC9GT05U PjxCUj48Rk9OVCBGQUNFPSJNb25vc3BhY2UsQ291cmllciI+Jmd0OyBPSywgaGVyZSBpcyB0aGUg ZXhhbXBsZTo8QlI+PC9GT05UPjxCUj48Rk9OVCBGQUNFPSJNb25vc3BhY2UsQ291cmllciI+W2Nv ZGUgZGVsZXRlZF08QlI+PC9GT05UPjxCUj48Rk9OVCBGQUNFPSJNb25vc3BhY2UsQ291cmllciI+ Jmd0OyBXaGVuIEkgcnVuIHRoaXMgY29kZSwgdGhlIGZpcnN0IGNvbnZvbHV0aW9uIGdpdmVzIGFu IGFycmF5IG9mIHplcm9zLCB0aGU8QlI+Jmd0OyBzZWNvbmQgY29udm9sdXRpb24gaXMgZ29vZC48 QlI+PC9GT05UPjxCUj48Rk9OVCBGQUNFPSJNb25vc3BhY2UsQ291cmllciI+VGhlIHByb2JsZW0g aXMgdGhhdCB0aGUgb3V0cHV0IG9mIHRoZSBjb252b2x2ZSBmdW5jdGlvbnMgKGJvdGggdGhlIGNv bnZvbHZlPEJSPmFuZCB0aGUgbmRfaW1hZ2UgdmVyc2lvbnMpIHJldHVybiB0aGUgcmVzdWx0IHdp dGggdGhlIHNhbWUgdHlwZSBhcyB0aGUgaW5wdXQuPEJSPkluIHlvdXIgZmlyc3QgY2FzZSB0aGUg aW5wdXQgd2FzIG9mIEJvb2wgdHlwZSBhbmQgdGh1cyB0aGUgcmVzdWx0IGNvdWxkIG9ubHk8QlI+ YmUgMCBvciAxLiBDb252ZXJzaW9uIHRvIEZsb2F0NjQgYmVmb3JlIGNhbGN1bGF0aW9uIHNlZW1z IHRvIGZpeCB0aGUgcHJvYmxlbS48QlI+KEkgaW5zZXJ0ZWQgJ2RhdGExID0gaW5wdXRhcnJheShk YXRhMSwgRmxvYXQ2NCknIGJlZm9yZSB0aGUgZnVuY3Rpb24gY2FsbC4pPEJSPkFsdGVybmF0aXZl bHkgaW4gbmRfaW1hZ2UuY29udm9sdmUxZCB5b3UgY2FuIHVzZSB0aGUgb3V0cHV0X3R5cGUgZmxh ZyB0byBzYXZlPEJSPnlvdSB0aGUgY29udmVyc2lvbi48QlI+PC9GT05UPjxCUj48Rk9OVCBGQUNF PSJNb25vc3BhY2UsQ291cmllciI+Jmd0OyBJIGFsc28gZGlkIGEgcXVpY2sgdHJ5IHdpdGggY29u dm9sdmUgZnJvbSBuZF9pbWFnZSwgYnV0IHRoaXMgZnVuY3Rpb24gd2FzPEJSPiZndDsgc2xvd2Vy IHRoYW4gdGhlIGNvbnZvbHZlIGZyb20gbnVtYXJyYXkgd2l0aCBhIGxhcmdlIEdhdXNzaWFuIGRp c3RyaWJ1dGlvbi48QlI+PC9GT05UPjxCUj48Rk9OVCBGQUNFPSJNb25vc3BhY2UsQ291cmllciI+ WWVzLCBuZF9pbWFnZSBjb252b2x2ZSBpcyBkZXNpZ25lZCBmb3IgdGhlIGdlbmVyYWwgbXVsdGkt ZGltZW5zaW9uYWwgY2FzZSBhbmQ8QlI+bWF5IHRoZXJlZm9yZSBiZSBzbG93ZXIgaW4gc29tZSBj YXNlcy4gQWxzbyBJIHRoaW5rIHRoZSBjb252b2x2ZSB2ZXJzaW9uIHVzZXM8QlI+YW4gaW50ZXJu YWwgY29udmVyc2lvbiBvZiB0aGUgaW5wdXQgdG8gZG91YmxlIHR5cGUgd2hpY2ggYXBwZWFycyB0 byBmYXN0ZXIuPEJSPlRoZSBuZF9pbWFnZSB2ZXJzaW9uIGRvZXMgbm90IGRvIHRoYXQgYmVjYXVz ZSBmb3IgbGFyZ2UgbXVsdGktZGltZW5zaW9uYWw8QlI+YXJyYXlzIHRoYXQgbWF5IGNvc3QgYSBs b3Qgb2YgbWVtb3J5LiBJIHRoaW5rIGlmIHlvdSBkbyB0aGUgY29udmVyc2lvbiB0bzxCUj5GbG9h dDY0IGJlZm9yZSB0aGUgc3BlZWQgc2hvdWxkIGJlIGNsb3Nlci4gRm9yIG9uZS1kaW1lbnNpb25h bCBjYXNlcywgeW91PEJSPnNob3VsZCBwcm9iYWJseSBqdXN0IHN0aWNrIHRvIHRoZSBjb252b2x2 ZSB2ZXJzaW9uLjxCUj48L0ZPTlQ+PEJSPjxGT05UIEZBQ0U9Ik1vbm9zcGFjZSxDb3VyaWVyIj5J IGhvcGUgdGhpcyBzb2x2ZWQgeW91ciBwcm9ibGVtLjxCUj48L0ZPTlQ+PEJSPjxGT05UIEZBQ0U9 Ik1vbm9zcGFjZSxDb3VyaWVyIj5DaGVlcnMsIFBldGVyPEJSPjwvRk9OVD48QlI+PEZPTlQgRkFD RT0iTW9ub3NwYWNlLENvdXJpZXIiPi0tPEJSPkRyLiBQZXRlciBKLiBWZXJ2ZWVyPEJSPkNlbGwg QmlvbG9neSBhbmQgQ2VsbCBCaW9waHlzaWNzIFByb2dyYW1tZTxCUj5FdXJvcGVhbiBNb2xlY3Vs YXIgQmlvbG9neSBMYWJvcmF0b3J5PEJSPk1leWVyaG9mc3RyYXNzZSAxPEJSPkQtNjkxMTcgSGVp ZGVsYmVyZzxCUj5HZXJtYW55PEJSPlRlbC4gOiArNDkgNjIyMSAzODcyNDU8QlI+RmF4ICZuYnNw OzogKzQ5IDYyMjEgMzg3MzA2PC9GT05UPjwvUD4= |
|
From: Francesc A. <fa...@op...> - 2003-12-16 10:01:52
|
Hi,
Conversion between Numeric arrays and numarray fails for zero-sized Numeric
arrays:
>>> numarray.array(Numeric.zeros((0,), 'd'))
Traceback (most recent call last):
File "<stdin>", line 1, in ?
File "/usr/local/lib/python2.3/site-packages/numarray/numarraycore.py", line
301, in array
raise ValueError("Unknown input type")
ValueError: Unknown input type
but this works:
>>> numarray.array(Numeric.zeros((1,), 'd'))
array([ 0.])
I'm wondering if it would be better to log a bug of the SF project site for
such a no too grave bugs rather than report them to the list, although it
seems to me that people only logs Numeric reports on the SF project. Would
it be correct using the same place for numarray reports as well?
Cheers,
--
Francesc Alted
|
|
From: Peter V. <ve...@em...> - 2003-12-16 09:54:59
|
Hi Remco, > OK, here is the example: [code deleted] > When I run this code, the first convolution gives an array of zeros, the > second convolution is good. The problem is that the output of the convolve functions (both the convolve and the nd_image versions) return the result with the same type as the input. In your first case the input was of Bool type and thus the result could only be 0 or 1. Conversion to Float64 before calculation seems to fix the problem. (I inserted 'data1 = inputarray(data1, Float64)' before the function call.) Alternatively in nd_image.convolve1d you can use the output_type flag to save you the conversion. > I also did a quick try with convolve from nd_image, but this function was > slower than the convolve from numarray with a large Gaussian distribution. Yes, nd_image convolve is designed for the general multi-dimensional case and may therefore be slower in some cases. Also I think the convolve version uses an internal conversion of the input to double type which appears to faster. The nd_image version does not do that because for large multi-dimensional arrays that may cost a lot of memory. I think if you do the conversion to Float64 before the speed should be closer. For one-dimensional cases, you should probably just stick to the convolve version. I hope this solved your problem. Cheers, Peter -- Dr. Peter J. Verveer Cell Biology and Cell Biophysics Programme European Molecular Biology Laboratory Meyerhofstrasse 1 D-69117 Heidelberg Germany Tel. : +49 6221 387245 Fax : +49 6221 387306 |
|
From: <R....@ma...> - 2003-12-16 09:33:43
|
OK, here is the example: [code] from numarray import * import numarray.convolve as conv lx=20 center=10 index=arange(lx,type='Float64') data1=(abs(center-index))<3 print 'The binary pattern:' print data1 PSF=5 xwidth=18 xmid=xwidth/2 sigma=PSF/(sqrt(8*log(2))) index=arange(xwidth,type='Float64') ShortGauss = (1/(sqrt(2*pi)*sigma)) * exp(-0.5*((index-xmid)/sigma)*((index-xmid)/sigma)) print '\nThe short gauss:' print ShortGauss image=conv.convolve(data1,ShortGauss,mode=conv.SAME) print '\nThe convolution of the binary pattern with the short gauss:' print image PSF=5 xwidth=22 xmid=xwidth/2 sigma=PSF/(sqrt(8*log(2))) index=arange(xwidth,type='Float64') LongGauss = (1/(sqrt(2*pi)*sigma)) * exp(-0.5*((index-xmid)/sigma)*((index-xmid)/sigma)) print '\nThe long gauss:' print LongGauss image=conv.convolve(data1,LongGauss,mode=conv.SAME) print '\nThe convolution of the binary pattern with the long gauss:' print image [\code] When I run this code, the first convolution gives an array of zeros, the second convolution is good. I also did a quick try with convolve from nd_image, but this function was slower than the convolve from numarray with a large Gaussian distribution. Remco Jager MAPPER Lithography Lorentzweg 1 2628 CJ Delft, The Netherlands tel.: +31 (0)15 2789439 fax: +31 (0)15-2789473 http://www.mapperlithography.com This e-mail, attachments and (any part of) its content are (i) intended for the named addressee(s) only and (ii) strictly confidential and proprietary. All rights are reserved by MAPPER Lithography. Any unauthorized use, disclosure and/or copying are strictly prohibited, except with prior and express written permission by MAPPER Lithography. Should you have received this e-mail, attachments and its content by mistake, please bring this to our attention and destroy this e-mail and attachments in full. Thank you. |---------+----------------------------> | | Nadav Horesh | | | <nadavh@visionsen| | | se.com> | | | | | | 15/12/2003 14:13 | | | | |---------+----------------------------> >------------------------------------------------------------------------------------------------------------------------------| | | | To: R....@ma... | | cc: Peter Verveer <ve...@em...>, num...@li... | | Subject: Re: [Numpy-discussion] Convolve returning zero array | >------------------------------------------------------------------------------------------------------------------------------| 1. Can you post an example? 2. Whenever convolution runs slow because the arrays are large, use FFT (with a proper padding) --- it can be an order of magnitude (or more) faster. Nadav. R....@ma... wrote: > Hi Peter, > > At first I used numarray 0.6. This morning I have installed 0.8 but the > results are the same. I will try the convolve functions in the nd_image > package. > > Thanks, > > Remco Jager > > MAPPER Lithography > Lorentzweg 1 > 2628 CJ Delft, The Netherlands > tel.: +31 (0)15 2789439 > fax: +31 (0)15-2789473 > http://www.mapperlithography.com > > This e-mail, attachments and (any part of) its content are (i) intended for > the named addressee(s) only and (ii) strictly confidential and proprietary. > All rights are reserved by MAPPER Lithography. Any unauthorized use, > disclosure and/or copying are strictly prohibited, except with prior and > express written permission by MAPPER Lithography. Should you have received > this e-mail, attachments and its content by mistake, please bring this to > our attention and destroy this e-mail and attachments in full. Thank you. > > > |---------+----------------------------> > | | Peter Verveer | > | | <verveer@embl-hei| > | | delberg.de> | > | | | > | | 15/12/2003 12:04 | > | | | > |---------+----------------------------> > >------------------------------------------------------------------------------------------------------------------------------| > | | > | To: R....@ma..., num...@li... | > | cc: | > | Subject: Re: [Numpy-discussion] Convolve returning zero array | > >------------------------------------------------------------------------------------------------------------------------------| > > > > > Hi Remco, > > Sounds like a bug. Which version of numarray do you use? Version 0.8 of > numarray should have appeared on sourceforge now. If the convolve in that > version does still not work, you could try out the convolution function in > the new nd_image package that is part of numarray 0.8. If that does not > work, > let me know since I am the author of that package, and will fix problems > with > it. > > Cheers, Peter > > On Monday 15 December 2003 11:24, R....@ma... wrote: > >>Hi list, >> >>I already posted this on the numarray forum on freshmeat, but Jay T > > Miller > >>advised me to post my problem to this list. OK, now for the problem: I > > try > >>to convolve a Gaussian distribution with a binary pattern. For small > > values > >>of the sigma of the Gaussian distribution the convolution returns an > > array > >>of zeros. For a large value the results are OK. >>I did some more research and found out that the zero array is returned if >>the length of the Gaussian is smaller than the length of the binary >>pattern. In the function call the Gaussian is the kernel and the binary >>pattern is the data. The convolution mode is 'SAME'. I have swapped the >>data and kernel in the convolve function call, but this has no influence > > on > >>the result, as this is swapped again in convolve.py. A quick and dirty >>workaround is to always make the Gaussian distribution longer than the >>binary pattern, but for very large binary patterns this increases the >>calculation time significantly. Does anyone have an idea how to solve > > this > >>properly? >> >>Met vriendelijke groeten, >> >>Remco Jager >> >>MAPPER Lithography >>Lorentzweg 1 >>2628 CJ Delft, The Netherlands >>tel.: +31 (0)15 2789439 >>fax: +31 (0)15-2789473 >>http://www.mapperlithography.com >> >>This e-mail, attachments and (any part of) its content are (i) intended > > for > >>the named addressee(s) only and (ii) strictly confidential and > > proprietary. > >>All rights are reserved by MAPPER Lithography. Any unauthorized use, >>disclosure and/or copying are strictly prohibited, except with prior and >>express written permission by MAPPER Lithography. Should you have > > received > >>this e-mail, attachments and its content by mistake, please bring this to >>our attention and destroy this e-mail and attachments in full. Thank you. >> >> >> >> >> >> >>------------------------------------------------------- >>This SF.net email is sponsored by: SF.net Giveback Program. >>Does SourceForge.net help you be more productive? Does it >>help you create better code? SHARE THE LOVE, and help us help >>YOU! Click Here: http://sourceforge.net/donate/ >>_______________________________________________ >>Numpy-discussion mailing list >>Num...@li... >>https://lists.sourceforge.net/lists/listinfo/numpy-discussion > > > -- > Dr. Peter J. Verveer > Cell Biology and Cell Biophysics Programme > European Molecular Biology Laboratory > Meyerhofstrasse 1 > D-69117 Heidelberg > Germany > Tel. : +49 6221 387245 > Fax : +49 6221 387306 > > > > > > > > > > > ------------------------------------------------------- > This SF.net email is sponsored by: SF.net Giveback Program. > Does SourceForge.net help you be more productive? Does it > help you create better code? SHARE THE LOVE, and help us help > YOU! Click Here: http://sourceforge.net/donate/ > _______________________________________________ > Numpy-discussion mailing list > Num...@li... > https://lists.sourceforge.net/lists/listinfo/numpy-discussion > |
|
From: Nadav H. <na...@vi...> - 2003-12-16 09:19:12
|
It seems that section 13.1.1 (Installation using LAPACK) is not valid any more --- is there another way? Nadav. |
|
From: Francesc A. <fa...@op...> - 2003-12-15 16:45:28
|
Hi,
NumArray class seems to detect invalid sequences, but not in CharArray:
>>> array([3, [4, 5.2]])
Traceback (most recent call last):
File "<stdin>", line 1, in ?
File "/usr/local/lib/python2.3/site-packages/numarray/numarraycore.py", line
286, in array
return fromlist(sequence, type, shape)
File "/usr/local/lib/python2.3/site-packages/numarray/numarraycore.py", line
195, in fromlist
arr.fromlist(seq)
libnumarray.error: setArrayFromSequence: invalid sequence.
>>> strings.array(["aaa", ["bbb", "ccc"]])
CharArray(['aaa', 'bbb'])
--
Francesc Alted
|
|
From: Nadav H. <na...@vi...> - 2003-12-15 13:15:22
|
1. Can you post an example? 2. Whenever convolution runs slow because the arrays are large, use FFT (with a proper padding) --- it can be an order of magnitude (or more) faster. Nadav. R....@ma... wrote: > Hi Peter, > > At first I used numarray 0.6. This morning I have installed 0.8 but the > results are the same. I will try the convolve functions in the nd_image > package. > > Thanks, > > Remco Jager > > MAPPER Lithography > Lorentzweg 1 > 2628 CJ Delft, The Netherlands > tel.: +31 (0)15 2789439 > fax: +31 (0)15-2789473 > http://www.mapperlithography.com > > This e-mail, attachments and (any part of) its content are (i) intended for > the named addressee(s) only and (ii) strictly confidential and proprietary. > All rights are reserved by MAPPER Lithography. Any unauthorized use, > disclosure and/or copying are strictly prohibited, except with prior and > express written permission by MAPPER Lithography. Should you have received > this e-mail, attachments and its content by mistake, please bring this to > our attention and destroy this e-mail and attachments in full. Thank you. > > > |---------+----------------------------> > | | Peter Verveer | > | | <verveer@embl-hei| > | | delberg.de> | > | | | > | | 15/12/2003 12:04 | > | | | > |---------+----------------------------> > >------------------------------------------------------------------------------------------------------------------------------| > | | > | To: R....@ma..., num...@li... | > | cc: | > | Subject: Re: [Numpy-discussion] Convolve returning zero array | > >------------------------------------------------------------------------------------------------------------------------------| > > > > > Hi Remco, > > Sounds like a bug. Which version of numarray do you use? Version 0.8 of > numarray should have appeared on sourceforge now. If the convolve in that > version does still not work, you could try out the convolution function in > the new nd_image package that is part of numarray 0.8. If that does not > work, > let me know since I am the author of that package, and will fix problems > with > it. > > Cheers, Peter > > On Monday 15 December 2003 11:24, R....@ma... wrote: > >>Hi list, >> >>I already posted this on the numarray forum on freshmeat, but Jay T > > Miller > >>advised me to post my problem to this list. OK, now for the problem: I > > try > >>to convolve a Gaussian distribution with a binary pattern. For small > > values > >>of the sigma of the Gaussian distribution the convolution returns an > > array > >>of zeros. For a large value the results are OK. >>I did some more research and found out that the zero array is returned if >>the length of the Gaussian is smaller than the length of the binary >>pattern. In the function call the Gaussian is the kernel and the binary >>pattern is the data. The convolution mode is 'SAME'. I have swapped the >>data and kernel in the convolve function call, but this has no influence > > on > >>the result, as this is swapped again in convolve.py. A quick and dirty >>workaround is to always make the Gaussian distribution longer than the >>binary pattern, but for very large binary patterns this increases the >>calculation time significantly. Does anyone have an idea how to solve > > this > >>properly? >> >>Met vriendelijke groeten, >> >>Remco Jager >> >>MAPPER Lithography >>Lorentzweg 1 >>2628 CJ Delft, The Netherlands >>tel.: +31 (0)15 2789439 >>fax: +31 (0)15-2789473 >>http://www.mapperlithography.com >> >>This e-mail, attachments and (any part of) its content are (i) intended > > for > >>the named addressee(s) only and (ii) strictly confidential and > > proprietary. > >>All rights are reserved by MAPPER Lithography. Any unauthorized use, >>disclosure and/or copying are strictly prohibited, except with prior and >>express written permission by MAPPER Lithography. Should you have > > received > >>this e-mail, attachments and its content by mistake, please bring this to >>our attention and destroy this e-mail and attachments in full. Thank you. >> >> >> >> >> >> >>------------------------------------------------------- >>This SF.net email is sponsored by: SF.net Giveback Program. >>Does SourceForge.net help you be more productive? Does it >>help you create better code? SHARE THE LOVE, and help us help >>YOU! Click Here: http://sourceforge.net/donate/ >>_______________________________________________ >>Numpy-discussion mailing list >>Num...@li... >>https://lists.sourceforge.net/lists/listinfo/numpy-discussion > > > -- > Dr. Peter J. Verveer > Cell Biology and Cell Biophysics Programme > European Molecular Biology Laboratory > Meyerhofstrasse 1 > D-69117 Heidelberg > Germany > Tel. : +49 6221 387245 > Fax : +49 6221 387306 > > > > > > > > > > > ------------------------------------------------------- > This SF.net email is sponsored by: SF.net Giveback Program. > Does SourceForge.net help you be more productive? Does it > help you create better code? SHARE THE LOVE, and help us help > YOU! Click Here: http://sourceforge.net/donate/ > _______________________________________________ > Numpy-discussion mailing list > Num...@li... > https://lists.sourceforge.net/lists/listinfo/numpy-discussion > |
|
From: <R....@ma...> - 2003-12-15 11:50:24
|
Hi Peter, At first I used numarray 0.6. This morning I have installed 0.8 but the results are the same. I will try the convolve functions in the nd_image package. Thanks, Remco Jager MAPPER Lithography Lorentzweg 1 2628 CJ Delft, The Netherlands tel.: +31 (0)15 2789439 fax: +31 (0)15-2789473 http://www.mapperlithography.com This e-mail, attachments and (any part of) its content are (i) intended for the named addressee(s) only and (ii) strictly confidential and proprietary. All rights are reserved by MAPPER Lithography. Any unauthorized use, disclosure and/or copying are strictly prohibited, except with prior and express written permission by MAPPER Lithography. Should you have received this e-mail, attachments and its content by mistake, please bring this to our attention and destroy this e-mail and attachments in full. Thank you. |---------+----------------------------> | | Peter Verveer | | | <verveer@embl-hei| | | delberg.de> | | | | | | 15/12/2003 12:04 | | | | |---------+----------------------------> >------------------------------------------------------------------------------------------------------------------------------| | | | To: R....@ma..., num...@li... | | cc: | | Subject: Re: [Numpy-discussion] Convolve returning zero array | >------------------------------------------------------------------------------------------------------------------------------| Hi Remco, Sounds like a bug. Which version of numarray do you use? Version 0.8 of numarray should have appeared on sourceforge now. If the convolve in that version does still not work, you could try out the convolution function in the new nd_image package that is part of numarray 0.8. If that does not work, let me know since I am the author of that package, and will fix problems with it. Cheers, Peter On Monday 15 December 2003 11:24, R....@ma... wrote: > Hi list, > > I already posted this on the numarray forum on freshmeat, but Jay T Miller > advised me to post my problem to this list. OK, now for the problem: I try > to convolve a Gaussian distribution with a binary pattern. For small values > of the sigma of the Gaussian distribution the convolution returns an array > of zeros. For a large value the results are OK. > I did some more research and found out that the zero array is returned if > the length of the Gaussian is smaller than the length of the binary > pattern. In the function call the Gaussian is the kernel and the binary > pattern is the data. The convolution mode is 'SAME'. I have swapped the > data and kernel in the convolve function call, but this has no influence on > the result, as this is swapped again in convolve.py. A quick and dirty > workaround is to always make the Gaussian distribution longer than the > binary pattern, but for very large binary patterns this increases the > calculation time significantly. Does anyone have an idea how to solve this > properly? > > Met vriendelijke groeten, > > Remco Jager > > MAPPER Lithography > Lorentzweg 1 > 2628 CJ Delft, The Netherlands > tel.: +31 (0)15 2789439 > fax: +31 (0)15-2789473 > http://www.mapperlithography.com > > This e-mail, attachments and (any part of) its content are (i) intended for > the named addressee(s) only and (ii) strictly confidential and proprietary. > All rights are reserved by MAPPER Lithography. Any unauthorized use, > disclosure and/or copying are strictly prohibited, except with prior and > express written permission by MAPPER Lithography. Should you have received > this e-mail, attachments and its content by mistake, please bring this to > our attention and destroy this e-mail and attachments in full. Thank you. > > > > > > > ------------------------------------------------------- > This SF.net email is sponsored by: SF.net Giveback Program. > Does SourceForge.net help you be more productive? Does it > help you create better code? SHARE THE LOVE, and help us help > YOU! Click Here: http://sourceforge.net/donate/ > _______________________________________________ > Numpy-discussion mailing list > Num...@li... > https://lists.sourceforge.net/lists/listinfo/numpy-discussion -- Dr. Peter J. Verveer Cell Biology and Cell Biophysics Programme European Molecular Biology Laboratory Meyerhofstrasse 1 D-69117 Heidelberg Germany Tel. : +49 6221 387245 Fax : +49 6221 387306 |
|
From: Peter V. <ve...@em...> - 2003-12-15 11:04:51
|
Hi Remco, Sounds like a bug. Which version of numarray do you use? Version 0.8 of numarray should have appeared on sourceforge now. If the convolve in that version does still not work, you could try out the convolution function in the new nd_image package that is part of numarray 0.8. If that does not work, let me know since I am the author of that package, and will fix problems with it. Cheers, Peter On Monday 15 December 2003 11:24, R....@ma... wrote: > Hi list, > > I already posted this on the numarray forum on freshmeat, but Jay T Miller > advised me to post my problem to this list. OK, now for the problem: I try > to convolve a Gaussian distribution with a binary pattern. For small values > of the sigma of the Gaussian distribution the convolution returns an array > of zeros. For a large value the results are OK. > I did some more research and found out that the zero array is returned if > the length of the Gaussian is smaller than the length of the binary > pattern. In the function call the Gaussian is the kernel and the binary > pattern is the data. The convolution mode is 'SAME'. I have swapped the > data and kernel in the convolve function call, but this has no influence on > the result, as this is swapped again in convolve.py. A quick and dirty > workaround is to always make the Gaussian distribution longer than the > binary pattern, but for very large binary patterns this increases the > calculation time significantly. Does anyone have an idea how to solve this > properly? > > Met vriendelijke groeten, > > Remco Jager > > MAPPER Lithography > Lorentzweg 1 > 2628 CJ Delft, The Netherlands > tel.: +31 (0)15 2789439 > fax: +31 (0)15-2789473 > http://www.mapperlithography.com > > This e-mail, attachments and (any part of) its content are (i) intended for > the named addressee(s) only and (ii) strictly confidential and proprietary. > All rights are reserved by MAPPER Lithography. Any unauthorized use, > disclosure and/or copying are strictly prohibited, except with prior and > express written permission by MAPPER Lithography. Should you have received > this e-mail, attachments and its content by mistake, please bring this to > our attention and destroy this e-mail and attachments in full. Thank you. > > > > > > > ------------------------------------------------------- > This SF.net email is sponsored by: SF.net Giveback Program. > Does SourceForge.net help you be more productive? Does it > help you create better code? SHARE THE LOVE, and help us help > YOU! Click Here: http://sourceforge.net/donate/ > _______________________________________________ > Numpy-discussion mailing list > Num...@li... > https://lists.sourceforge.net/lists/listinfo/numpy-discussion -- Dr. Peter J. Verveer Cell Biology and Cell Biophysics Programme European Molecular Biology Laboratory Meyerhofstrasse 1 D-69117 Heidelberg Germany Tel. : +49 6221 387245 Fax : +49 6221 387306 |
|
From: <R....@ma...> - 2003-12-15 10:24:50
|
Hi list, I already posted this on the numarray forum on freshmeat, but Jay T Miller advised me to post my problem to this list. OK, now for the problem: I try to convolve a Gaussian distribution with a binary pattern. For small values of the sigma of the Gaussian distribution the convolution returns an array of zeros. For a large value the results are OK. I did some more research and found out that the zero array is returned if the length of the Gaussian is smaller than the length of the binary pattern. In the function call the Gaussian is the kernel and the binary pattern is the data. The convolution mode is 'SAME'. I have swapped the data and kernel in the convolve function call, but this has no influence on the result, as this is swapped again in convolve.py. A quick and dirty workaround is to always make the Gaussian distribution longer than the binary pattern, but for very large binary patterns this increases the calculation time significantly. Does anyone have an idea how to solve this properly? Met vriendelijke groeten, Remco Jager MAPPER Lithography Lorentzweg 1 2628 CJ Delft, The Netherlands tel.: +31 (0)15 2789439 fax: +31 (0)15-2789473 http://www.mapperlithography.com This e-mail, attachments and (any part of) its content are (i) intended for the named addressee(s) only and (ii) strictly confidential and proprietary. All rights are reserved by MAPPER Lithography. Any unauthorized use, disclosure and/or copying are strictly prohibited, except with prior and express written permission by MAPPER Lithography. Should you have received this e-mail, attachments and its content by mistake, please bring this to our attention and destroy this e-mail and attachments in full. Thank you. |
|
From: J. <bos...@mf...> - 2003-12-15 09:53:07
|
Hello ! I want to create a vector using Python C/API with connection to numpy. My idea is to create vector on the fly without previously knowing what the size will be. Is this possible? B. |
|
From: Todd M. <jm...@st...> - 2003-12-12 15:03:10
|
This is fixed in CVS for numarray-0.8. Thanks for the report! Todd On Fri, 2003-12-12 at 09:39, Edward C. Jones wrote: > I am using a CVS copy of numarray downloaded on Dec 9, 2003. The > following program raises "TypeError: Can't reshape non-contiguous numarray": > > #! /usr/bin/env python > > import numarray > > z = numarray.zeros((2,2,2)) > z.max() > > The error happens because "numarray.maximum.areduce" returns a > non-contiguous array. > > > > ------------------------------------------------------- > This SF.net email is sponsored by: SF.net Giveback Program. > Does SourceForge.net help you be more productive? Does it > help you create better code? SHARE THE LOVE, and help us help > YOU! Click Here: http://sourceforge.net/donate/ > _______________________________________________ > Numpy-discussion mailing list > Num...@li... > https://lists.sourceforge.net/lists/listinfo/numpy-discussion -- Todd Miller Space Telescope Science Institute 3700 San Martin Drive Baltimore MD, 21030 (410) 338 - 4576 |
|
From: Todd M. <jm...@st...> - 2003-12-12 14:58:18
|
I agree that this is a necessary feature... but making take() work
consistently across the range of array types in numarray requires more
thought (and changes I think) than I can do on the eve of a release.
I logged this as the first bug against numarray-0.8.
However, I got array based indexing to work with minimal changes, so
all is not lost:
>>> a=strings.fromlist(["this"]*25); a.shape=(5,5); a[ range(2,4) ]
CharArray([['this', 'this', 'this', 'this', 'this'],
['this', 'this', 'this', 'this', 'this']])
>>> a[ range(2,4) ] = strings.fromlist(["that"]); a
CharArray([['this', 'this', 'this', 'this', 'this'],
['this', 'this', 'this', 'this', 'this'],
['that', 'that', 'that', 'that', 'that'],
['that', 'that', 'that', 'that', 'that'],
['this', 'this', 'this', 'this', 'this']])
works now in CVS and will be part of numarray-0.8.
Todd
On Fri, 2003-12-12 at 07:34, Francesc Alted wrote:
> Hi,
>
> take() doesn't seems to work when applied to numarray CharArrays:
>
> >>> a=strings.array("a"*400, shape=(2,4), itemsize=3)
> >>> take(a, (1,2))
> Traceback (most recent call last):
> File "<stdin>", line 1, in ?
> File "/usr/local/lib/python2.3/site-packages/numarray/ufunc.py", line 1766,
> in take
> return array._take((indices,), clipmode=clipmode)
> File "/usr/local/lib/python2.3/site-packages/numarray/generic.py", line 799,
> in _take
> impliedShape, N = _takeShape(self, indices)
> File "/usr/local/lib/python2.3/site-packages/numarray/generic.py", line 133,
> in _takeShape
> if scattered.isbyteswapped() or not scattered.iscontiguous():
> AttributeError: 'CharArray' object has no attribute 'isbyteswapped'
>
> and, IMO, that functionality would be nice to have.
>
> Cheers,
>
> --
> Francesc Alted
>
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: SF.net Giveback Program.
> Does SourceForge.net help you be more productive? Does it
> help you create better code? SHARE THE LOVE, and help us help
> YOU! Click Here: http://sourceforge.net/donate/
> _______________________________________________
> Numpy-discussion mailing list
> Num...@li...
> https://lists.sourceforge.net/lists/listinfo/numpy-discussion
--
Todd Miller
Space Telescope Science Institute
3700 San Martin Drive
Baltimore MD, 21030
(410) 338 - 4576
|
|
From: Edward C. J. <edc...@er...> - 2003-12-12 14:43:25
|
I am using a CVS copy of numarray downloaded on Dec 9, 2003. The following program raises "TypeError: Can't reshape non-contiguous numarray": #! /usr/bin/env python import numarray z = numarray.zeros((2,2,2)) z.max() The error happens because "numarray.maximum.areduce" returns a non-contiguous array. |
|
From: Francesc A. <fa...@op...> - 2003-12-12 12:35:06
|
Hi,
take() doesn't seems to work when applied to numarray CharArrays:
>>> a=strings.array("a"*400, shape=(2,4), itemsize=3)
>>> take(a, (1,2))
Traceback (most recent call last):
File "<stdin>", line 1, in ?
File "/usr/local/lib/python2.3/site-packages/numarray/ufunc.py", line 1766,
in take
return array._take((indices,), clipmode=clipmode)
File "/usr/local/lib/python2.3/site-packages/numarray/generic.py", line 799,
in _take
impliedShape, N = _takeShape(self, indices)
File "/usr/local/lib/python2.3/site-packages/numarray/generic.py", line 133,
in _takeShape
if scattered.isbyteswapped() or not scattered.iscontiguous():
AttributeError: 'CharArray' object has no attribute 'isbyteswapped'
and, IMO, that functionality would be nice to have.
Cheers,
--
Francesc Alted
|
258 messages has been excluded from this view by a project administrator.
