Mailman 3 May 2007 - NumPy-Discussion

numpy and freeze.py
by Hanno Klemm 23 May '07

23 May '07

Hi, I want to use freeze.py on code that heavily relies on numpy. If I just try python2.5 /scratch/src/Python-2.5/Tools/freeze/freeze.py pylay.py the make works but then I get the error: Traceback (most recent call last): File "pylay.py", line 1, in <module> import kuvBeta4 as kuv File "kuvBeta4.py", line 6, in <module> import mfunBeta4 as mfun File "mfunBeta4.py", line 2, in <module> import numpy File "/glb/eu/siep_bv/proj/yot04/apps/python2.5/lib/python2.5/site-packages/numpy/__init__.py", line 39, in <module> import core File "/glb/eu/siep_bv/proj/yot04/apps/python2.5/lib/python2.5/site-packages/numpy/core/__init__.py", line 5, in <module> import multiarray ImportError: No module named multiarray Am I doing something wrong? Or does freeze.py not work with numpy? Hanno -- Hanno Klemm klemm(a)phys.ethz.ch

3 3

norm in scipy.linalg but not numpy?
by dmitrey 23 May '07

23 May '07

hi all, I was very surprised that norm() is present in scipy.linalg but absent in numpy. Don't you think it's better to add the one to numpy? As for me, I use the func very intensively, and I don't want to write my own (slow) func in Python or use sqrt(dot(v,v)) or scipy.linalg.norm(v) (i.e. make dependence on scipy because of single function linalg.norm()) (as it is proposed in http://www.scipy.org/NumPy_for_Matlab_Users) and what about second argument to norm, as MATLAB provides? norm(x) = norm(x,2) by default, norm(x, inf) = max(abs(x)), norm(x, -inf) = min(abs(x)), norm(x,1) = sum(abs(x)), norm(x,p) = (x1^p+...)^(1/p) etc... Of course, I could write those 10-15 lines of code in Python, but I think C or Pyrex would be better and faster. D.

4 8

More scalar math issues
by Erin Sheldon 23 May '07

23 May '07

Hi all- The types of objects are not preserved through mathematical operations when numpy scalars are involved. This has been discussed on this list before. I think the following error is problematic, however. >>> x=N.array(0.1) >>> type(x) <type 'numpy.ndarray'> >>> y = N.cos(x) # the type has changed >>> type(y) <type 'numpy.float64'> # Now some functions cannot be used on the data, for example. >>> N.arccos(y,y) <type 'exceptions.TypeError'>: return arrays must be of ArrayType So if I write a function which is supposed to work with any numpy object it will fail with this type of code. One solution would be to do this at the top of every file, for each input: var = N.array(var_input, ndmin=1, copy=False) or I could do this in the code: y=x.copy() N.cos(x,y) but wouldn't it be better to just preserve the type through functions like cos? Or does this cause repercussions in other places? Erin

2 1

Question about flags of fancy indexed array
by Albert Strasheim 23 May '07

23 May '07

Hello all Consider the following example: In [43]: x = N.zeros((3,2)) In [44]: x.flags Out[44]: C_CONTIGUOUS : True F_CONTIGUOUS : False OWNDATA : True WRITEABLE : True ALIGNED : True UPDATEIFCOPY : False In [45]: x[:,[1,0]].flags Out[45]: C_CONTIGUOUS : False F_CONTIGUOUS : True OWNDATA : False WRITEABLE : True ALIGNED : True UPDATEIFCOPY : False Is it correct that the F_CONTIGUOUS flag is set in the case of the fancy indexed x? I'm running NumPy 1.0.3.dev3792 here. Cheers, Albert

6 10

NumPy 1.0.3 release tomorrow, SciPy 0.5.3 next week
by Travis Oliphant 22 May '07

22 May '07

I'd like to tag the tree and make a NumPy 1.0.4 release tomorrow. Is there anything that needs to be done before that can happen? I'd also like to get another release of SciPy out the door as soon as possible. At the end of this week or early next week. I've added a bunch of 1-d classic spline interpolation algorithms (not the fast B-spline versions but the basic flexible algorithms of order 2, 3, 4, and 5 which have different boundary conditions). There are several ways to get at them in scipy.interpolate. I've also added my own "smoothest" algorithm which specifies the extra degrees of freedom available in spline formulations by making the resulting Kth-order spline have the smallest Kth-order derivative discontinuity. I've never seen this described by anyone but I'm sure somebody must have done discussed it before. If anybody knows something about splines who recalls seeing something like this, I would appreciate a reference. This specification actually allows very decent looking quadratic (2nd-order) splines. Best regards, -Travis

1 0

More help with numpy and SWIG
by vallis.35530053＠bloglines.com 22 May '07

22 May '07

Hello, I am trying to use numpy in conjuction with a custom SWIGged C struct (real_vec_t) that contains array data. I'd like to use the array interface to share the array data between real_vec_t and numpy. I have two questions for a combined numpy/SWIG magician: - This is how I implement the array interface in real_vec_t, so that if rvt is a real_vec_t instance, I can do na = numpy.array(rvt) and get a numpy array with the same content. I define the python attribute, %extend real_vec_t { %pythoncode %{ __array_struct__ = property(get_array_struct,doc='Array protocol') %} [...] %} which calls the C function (thanks to Lisandro Dalcin for suggesting this) %extend real_vec_t { PyObject *get_array_struct() { [...create the PyArrayInterface pointed to by "inter"...] [...do Py_INCREF(self) (see below)...] [...return PyCObject pointing to inter (see below)...] } %} I have problems with the last two steps. I'm supposed to do a Py_INCREF on the SWIG object so that the array data is preserved until all numpy arrays that may come to use it are destroyed. But I don't think it's correct to say "Py_INCREF(self)", because in this context self is a pointer to a real_vec_t structure, not the SWIG PyObject that wraps it... how should I call Py_INCREF? I have a related problem for the last step. I'm supposed to return PyCObject_FromVoidPtrAndDesc(inter,self,inter_free), where inter is the pointer to the PyArrayInterface, inter_free points to a function that will free any memory allocated with the PyArrayInterface, and self, again, should be a pointer to the SWIG wrapper, not just the real_vec_t C struct. - The second question is about the conditions under which the data pointed to by PyArrayInterface is shared with the numpy array I create with numpy.array(rvt), or it is copied. Is this a function of the flags I set in PyArrayInterface, of flags that must be passed to numpy.array(), or both? I couldn't quite dig this out of the numpy manual (but will happily take a pointer to the discussion therein...)

1 0

MAX_INT?
by Bill Baxter 22 May '07

22 May '07

Is there a way to obtain the equivalent of MAX_INT for the integral types numpy knows about? I know about numpy.finfo for the floating point types, but is there anything like that for integral types? Thanks, --Bill

2 2

array vs. matrix performance
by Robert Cimrman 22 May '07

22 May '07

I have come to a case where using a matrix would be easier than an array. The code uses lots of dot products, so I tested scipy.dot() performance with the code below and found that the array version is much faster (about 3 times for the given shape). What is the reason for this? Or is something wrong with my measurement? regards, r. --- import timeit setup = """ import numpy as nm import scipy as sc X = nm.random.rand( 100, 3 ) X = nm.asmatrix( X ) # (un)comment this line. print X.shape """ tt = timeit.Timer( 'sc.dot( X.T, X )', setup ) print tt.timeit()

4 5

removing references to Scipy Core
by Jarrod Millman 21 May '07

21 May '07

Hello, I noticed that there are a few places where Scipy Core is still referenced: 1). On http://numpy.scipy.org/new_features.html the title is still Scipy Core even though the heading has been updated: "<head> <meta content="text/html; charset=ISO-8859-1" http-equiv="content-type"> <title>New Features of SciPy Core</title> </head> <body> <h1>New Features of NumPy</h1>" 2). If you do a search for "scipy" on PyPi the highest score is scipy_core: http://cheeseshop.python.org/pypi/scipy_core/0.8.4 I am not that familiar with the cheeseshop, but if possible I think it would be a good idea to remove this page. Thanks, -- Jarrod Millman Computational Infrastructure for Research Labs 10 Giannini Hall, UC Berkeley phone: 510.643.4014 http://cirl.berkeley.edu/

1 0

Do we still need support for Python 2.3?
by Pearu Peterson 21 May '07

21 May '07

Hi, I noticed that numpy revision 3794 introduces `set` but that is available starting from Python 2.4. Hence the question. Best regards, Pearu

2 2