Mailman 3 October 2008 - NumPy-Discussion

collecting the bluest pixels
by paul taney 08 Oct '08

08 Oct '08

Hi, I have this silly color filter that Stefan gave me: def vanderwalt(image, f): """colorfilter, thanks to Stefan van der Walt""" RED, GRN, BLU = 0, 1, 2 bluemask = (image[...,BLU] > f*image[...,GRN]) & \ (image[...,BLU] > f*image[...,RED]) return bluemask To collect the right number of the bluest pixels I am calling it from this arduous successive approximation routine. It occured to me that someone on this list knows how to do this in a couple of lines... def successive_approximation(image, density, width, height, bpp): """keep calling vanderwalt till line length is within 10% of density target""" count = 0 failsafe = 20 # max iterations gimp.progress_init("this is the fun part...") init_high = 1.400001 init_low = 1.399991 high_guesses = [(init_high, width*height*bpp)] # we collect recent stats in this list low_guesses = [(init_low, 0)] crossed_high = crossed_low = False factor = 1.4 delta_high = .18 # adjust the factor by this much when it is too high delta_low = .14 # adjust the factor by this much when it is too low # if they were the same amount then they"d cancel xlength = 0 while not ((xlength < density*6.3) and (xlength > density*5.7)): bluemask = vanderwalt(image, factor) line = np.array(bluemask.nonzero()).swapaxes(0,1).tolist() xlength = len(line) # spread the search outward from initial condition (1.4) # until we cross target. we collect 6 times the desired # line length because we will take an average y per x # (unless they are too spread out and then we skip this column (despeckling)) # My dataset averages 6 blue pixels per column, so... if xlength > density*6: """too many points so raise the factor""" print "LOOP1.%i: factor too low =%.5f, target=%i, line length = %i" % (count, factor, density*6, xlength) crossed_low = True low_guesses.append((factor, xlength)) factor += delta_low if crossed_high: crossed_high = False delta_high *= .38 # cut the stepsizes delta_low *= .38 print "crossed target. factor now=%.5f" % (factor) else: """not enuf points so lower the factor""" print "LOOP1.%i: factor too high=%.5f, target=%i, line length = %i" % (count, factor, density*6, xlength) crossed_high = True high_guesses.append((factor, xlength)) factor -= delta_high if factor < 0: lineno = getframeinfo(currentframe())[1] # Beaz p 157 print "algorithm failed with factor=%.5f (line %i)" % (factor, lineno) break if crossed_low: crossed_low = False delta_high *= .38 # cut the stepsizes delta_low *= .38 print "crossed target. factor now=%.5f" % (factor) count += 1 gimp.progress_update(float(count) / failsafe) #print "count=%i, factor=%f\nhigh_guesses=%r\nlow_guesses=%r" % (count, factor, high_guesses, low_guesses) if count == failsafe: print "failsafe: level 1 failed to converge." break if (xlength < density*6.3) and (xlength > density*5.7): print "success at density*%1.3f" % (xlength/float(density)) else: pdb.gimp_message("maybe not enuf blue pixels for this algorithm.") return bluemask # end successive_approximation

4 5

ImportError: No module named ma
by John Harrold 07 Oct '08

07 Oct '08

Howdy, I'm trying to run some scripts associated with a paper I was reading. They use the matplotlib to generate figures which in turn calles numpy. I'm new to python, but I'm very familiar with perl. I'm getting the following error, and I'm not quite sure what is causing it. Can anyone here offer any suggestions? jmh@dhcp068-156:$ python2.5 Bb-plot.py Traceback (most recent call last): File "Bb-plot.py", line 7, in <module> from pylab import * File "/Library/Python/2.5/site-packages/matplotlib-0.98.3-py2.5.egg/pylab.py", line 1, in <module> from matplotlib.pylab import * File "/Library/Python/2.5/site-packages/matplotlib-0.98.3-py2.5.egg/matplotlib/__init__.py", line 128, in <module> from rcsetup import defaultParams, validate_backend, validate_toolbar File "/Library/Python/2.5/site-packages/matplotlib-0.98.3-py2.5.egg/matplotlib/rcsetup.py", line 19, in <module> from matplotlib.colors import is_color_like File "/Library/Python/2.5/site-packages/matplotlib-0.98.3-py2.5.egg/matplotlib/colors.py", line 39, in <module> import matplotlib.cbook as cbook File "/Library/Python/2.5/site-packages/matplotlib-0.98.3-py2.5.egg/matplotlib/cbook.py", line 9, in <module> import numpy.ma as ma ImportError: No module named ma -- John M Harrold

4 5

Merge of generalised ufuncs branch
by Stéfan van der Walt 07 Oct '08

07 Oct '08

Hi all, The generalised ufuncs branch was made available before SciPy'08. We solicited comments on its implementation and structuring, but received very little feedback. Unless there are any further comments from the community, I propose that we merge it. It is unfortunate that we have so many patches waiting for review (SciPy suffers worst, I'm afraid); clearly there are too few hours in a day. Nothing discourages contributions as much as a stale project, and I hope we can avoid that. Regards, Stéfan

4 4

Re: [Numpy-discussion] Scipy in Python 2.6
by Nadav Horesh 07 Oct '08

07 Oct '08

Scipy from svn compiles with python2.6 with no problem. Problems that came later: 1. Could not install packages of scikits: I got a strange error about unknown name: log 2. When trying to run from "idle" warnings become errors. I think that is a bug of idle. Nadav. -----הודעה מקורית----- מאת: jah [mailto:jah.mailinglist@gmail.com] נשלח: ג 07-אוקטובר-08 00:45 אל: Nadav Horesh נושא: Scipy in Python 2.6 Hi Nadav, I saw: http://projects.scipy.org/pipermail/numpy-discussion/2008-July/035968.html where you reported some compilation problems with scipy in Python 2.6. Did you resolve this issue? Do you have any knowledge if the current svn still has this problem? Thanks

3 3

multi-comparison expressions
by John 06 Oct '08

06 Oct '08

hi, why does the ValueError appear below, and how can i make that 2<a<5 expression work when a is an array? thanks. >>> from numpy import reshape,arange >>> a=reshape(arange(9),(3,3)) >>> a array([[0, 1, 2], [3, 4, 5], [6, 7, 8]]) >>> 2<a array([[False, False, False], [ True, True, True], [ True, True, True]], dtype=bool) >>> a<5 array([[ True, True, True], [ True, True, False], [False, False, False]], dtype=bool) >>> 2<a<5 Traceback (most recent call last): File "<stdin>", line 1, in <module> ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all() nor does it work with constant arrays: >>> from numpy import zeros >>> twos=zeros(a.shape)+2 >>> fives=zeros(a.shape)+5 >>> twos<a array([[False, False, False], [ True, True, True], [ True, True, True]], dtype=bool) >>> a<fives array([[ True, True, True], [ True, True, False], [False, False, False]], dtype=bool) >>> twos<a<fives Traceback (most recent call last): File "<stdin>", line 1, in <module> ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all() but it works with builtin numbers: >>> b=3 >>> 2<b<5 True

5 4

Merged clean_math_config branch
by David Cournapeau 06 Oct '08

06 Oct '08

Hi there, Just to mention that I merged back my changes from the clean_math_config branch into trunk. The main point of the branch is to clean our math configuration. If this causes problems, please report it. I built and tested on mac os x, linux 32 bits and windows (both mingw32 and VS 2003). It breaks windows 64 bits ATM, but this will be fixed soon. The numscons built is broken as well, but the missing features are already backported from numpy.distutils to numscons; a new working version of numscons is about to be released. Some details for the record: - more code is platform independent, with platform specifics math functions at one location - instead of platform-specific heuristics, every function is tested at the configuration stage. Some distutils.command.config functionalities were improved to make this process as fast as before for relatively standard-compliant platforms (mac os X and linux are as fast to build as before). - the distutils.command.config.check_func can now detect MS intrinsics (functions which are not visible without optimization flags). - should also be more robust for non standard platforms (MS compilers, mostly). In the process, a few hundred lines of C code have been removed. cheers, David

4 3

install without admin rights
by Roeland Huys 06 Oct '08

06 Oct '08

Hi, I am very interested in my work to switch from Matlab to Numpy. However, I want to test some algorithms first. I tried to install NumPy on one of our corporate servers. However, I do not have root access, so i did following: tar -xvzf numpy-1.2.0.tar.gz [OK] cd numpy-1.2.0 python setup.py build [OK] set PYTHONPATH=$HOME/pythonlib mv numpy ~/pythonlib python import numpy [ FAILS!] Could you tell me how I can install NumPy on my local account, (without admin rights) ? Thanks, Roeland.

4 5

cannot find numpy 1.2 for python 2.6 on window
by Daniel Wolff 05 Oct '08

05 Oct '08

Hello, I recently upgraded to python 2.6. I cannot seem to find the appropriate numpy install file. I tried to make use of the .exe built for python 2.5 but the installer hung. Any suggestions? Regards, Daniel

6 6

asscalar(number) - why yields error, why can't return the number?!
by dmitrey 05 Oct '08

05 Oct '08

hi all, I wonder why numpy.asscalar(1.5) yields error, why it can't just return 1.5? Is it intended to be ever changed? >>> numpy.__version__ '1.3.0.dev5864' D.

1 0

making numpy.dot faster
by Gael Varoquaux 04 Oct '08

04 Oct '08

I am doing a calculation where one call numpy.dot ends up taking 90% of the time (the array is huge: (61373, 500) ). Any chance I can make this faster? I would believe BLAS/ATLAS would be behind this, but from my quick analysis (ldd on numpy/core/multiarray.so) it doesn't seem so. Have I done something stupid when building numpy (disclaimer: I am on a system I don't know well --Mandriva--, so I could very well have done something stupid). Cheers, Gaël

4 5