Mailman 3 October 2013 - NumPy-Discussion

strange behavior of += with object array
by Neal Becker 01 Nov '13

01 Nov '13

import numpy as np #from accumulator import stat2nd_double ## Just to make this really clear, I'm making a dummy ## class here that overloads += class stat2nd_double (object): def __iadd__ (self, x): return self m = np.empty ((2,3), dtype=object) m[:,:] = stat2nd_double() m[0,0] += 1.0 <<<< no error here m += np.ones ((2,3)) <<< but this gives an error Traceback (most recent call last): File "test_stat.py", line 13, in <module> m += np.ones ((2,3)) TypeError: unsupported operand type(s) for +: 'stat2nd_double' and 'float'

2 4

memmory management question
by Georgios Exarchakis 31 Oct '13

31 Oct '13

Hi, I am using numpy with ipython from anaconda and I observe the following behavior: Python 2.7.5 |Anaconda 1.7.0 (64-bit)| (default, Jun 28 2013, 22:10:09) Type "copyright", "credits" or "license" for more information. IPython 1.0.0 -- An enhanced Interactive Python. ? -> Introduction and overview of IPython's features. %quickref -> Quick reference. help -> Python's own help system. object? -> Details about 'object', use 'object??' for extra details. Using matplotlib backend: Qt4Agg In [1]: a= np.random.rand(500000,1000) In [2]: a = a[:10000] In [3]: c= np.random.rand(500000,1000) In [4]: After In[1] I have an extra 3.7 GB of memory used, but this memory is not released at In[2]. I thought there might be some clever memory management trick so I executted In[3] but that just added an extra 3.7GB of memorry without releasing anything. Is that the right behavior in this case? If yes then how do you release memorry by slicing away parts of an array? Can you give me a description of the numpy internals in this case? Thank you very much for your time, Georgios

5 6

timedelta64 NaT doesn't print correctly when printing an ndarray
by Tanay Shah 30 Oct '13

30 Oct '13

When printing a timedelta64('NaT') scalar, it converts to string correctly but when you print an ndarray where an element is timedelta64('NaT'), it prints the element as its internal representation (-9223372036854775808) >>> numpy.timedelta64('NaT') numpy.timedelta64('NaT') >>> print numpy.timedelta64('NaT') NaT >>> numpy.array([None], dtype=numpy.timedelta64) array([-9223372036854775808], dtype=timedelta64) >>> print numpy.array([None], dtype=numpy.timedelta64) [-9223372036854775808] Could someone suggest a method by which I can get around this to display the NaT correctly? -- View this message in context: http://numpy-discussion.10968.n7.nabble.com/timedelta64-NaT-doesn-t-print-c… Sent from the Numpy-discussion mailing list archive at Nabble.com.

1 0

1.7.2
by Charles R Harris 30 Oct '13

30 Oct '13

Hi All, I'm going to tag 1.7.2 soon. That is, unless someone else would like the experience of making a release. Any volunteers? Chuck

3 4

getting the equivalent complex dtype from a real or int array
by Henry Gomersall 30 Oct '13

30 Oct '13

Is there a way to extract the size of array that would be created by doing 1j*array? The problem I'm having is in creating an empty array to fill with complex values without knowing a priori what the input data type is. For example, I have a real or int array `a`. I want to create an array `b` which can hold values from 1j*a in such a way that I don't need to compute those explicitly (because I only need parts of the array say), without upcasting (or indeed downcasting) the result. So if `a` was dtype 'float32`, `b` would be of dtype `complex64`. If `a` was `int64`, `b` would be of dtype `complex128` etc. Cheers, Henry

3 5

Re: [Numpy-discussion] getting the equivalent complex dtype from a real or int array
by David Goldsmith 30 Oct '13

30 Oct '13

We really ought to have a special page for all of Robert's little gems! DG On Tue, Oct 29, 2013 at 10:00 AM, <numpy-discussion-request(a)scipy.org>wrote: > > -----------------------------Message: 5 > Date: Tue, 29 Oct 2013 17:02:33 +0000 > From: Robert Kern <robert.kern(a)gmail.com> > Subject: Re: [Numpy-discussion] getting the equivalent complex dtype > from a real or int array > To: Discussion of Numerical Python <numpy-discussion(a)scipy.org> > Message-ID: > < > CAF6FJiuYnDbE1Uo9J6OnL1pq+oVZX-ecqKz0Qe9MigyQT69V_g(a)mail.gmail.com> > Content-Type: text/plain; charset="utf-8" > > On Tue, Oct 29, 2013 at 4:47 PM, Henry Gomersall <heng(a)cantab.net> wrote: > > > > Is there a way to extract the size of array that would be created by > > doing 1j*array? > > > > The problem I'm having is in creating an empty array to fill with > > complex values without knowing a priori what the input data type is. > > > > For example, I have a real or int array `a`. > > > > I want to create an array `b` which can hold values from 1j*a in such a > > way that I don't need to compute those explicitly (because I only need > > parts of the array say), without upcasting (or indeed downcasting) the > > result. > > > > So if `a` was dtype 'float32`, `b` would be of dtype `complex64`. If `a` > > was `int64`, `b` would be of dtype `complex128` etc. > > Quick and dirty: > > # Get a tiny array from `a` to test the dtype of its output when multiplied > # by a complex float. It must be an array rather than a scalar since the > # casting rules are different for array*scalar and scalar*scalar. > dt = (a.flat[:2] * 1j).dtype > b = np.empty(shape, dtype=dt) > > -- > Robert Kern >

2 1

numpy random uniform seed?
by Ao Liu 29 Oct '13

29 Oct '13

Hi, I've been using np.random.uniform and mpi4py. I found that the random number each processor (or rank) generated are the same, so I was wondering how random.uniform chose its seeds. Theoretically, those ranks shouldn't have anything to do with others. The only possibility that I can think of is that they all used the computer time as the seed, so the random numbers were the same. Anyone has more ideas and suggestions on how to correct this? Thanks, Best, Ao -- I am Ao.Liu, always hungry and still waiting for what I've always been believing. To know me more, please visit my little tiny personal website: www.frankliuao.com

2 1

Robust Sorting of Points
by Freddie Witherden 29 Oct '13

29 Oct '13

Hi all, This is a question which has been bugging me for a while. I have an (N, 3) array where N ~ 16 of points. These points are all unique and separated by a reasonable distance. I wish to sort these points into a canonical order in a fashion which is robust against small perturbations. In other words changing any component of any of the points by an epsilon ~ 1e-12 should not affect the resulting sorted order. Considering a direct application of np.lexsort: In [6]: my_array = np.array([[-0.5, 0, 2**0.5], [0.5, 0, 2**0.5 - 1e-15]]) In [7]: my_array[np.lexsort(my_array.T)] Out[7]: array([[ 0.5 , 0. , 1.41421356], [-0.5 , 0. , 1.41421356]]) however, if the small 1e-15 perturbation is removed the order changes to the 'natural' ordering. Hence, np.lexsort is out. Rounding the array before sorting is not suitable either; just because (a - b) < epsilon does not mean that np.around(a, decimals=x) == np.around(b, decimals=b). I am therefore looking for an efficient (= within a factor of 10 of np.lexsort) solution to the problem. I've looked at writing my own comparison function cmp(x, y) which looks at the next dimension if abs(x[i] - y[i]) < epsilon however using this with sorted is thousands of times slower. Given that I have well over 100,000 of these arrays this is nuisance. My other idea is to therefore find a means of quickly replacing all numbers within 10*epsilon of a given number in an array with that number. This should permit the application of np.lexsort in order to obtain the desired ordering (which is what I'm interesting in). However, I am yet to figure out how to do this efficiently. Before I throw in the towel and drop down to C are there any other neat tricks I am missing? Regards, Freddie.

9 18

Python parallel programming on Mac OS X Maverick
by Siegfried Gonzi 28 Oct '13

28 Oct '13

Hi all Quick question: What is the minimum RAM requirement for doing parallel programming with Python/Numpy on Mac OS X Maverick? I am about to buy a Macbook Pro 15" and I'd like to know if 8GB RAM (with SSD flash storage) for the Haswell quad core will be enough. I have never done any parallel programming with Python/Numpy but plan to get to grips with it on my new Macbook Pro where memory from now on is being soldered on and non-replaceable. Apple has a 14 days no quibbles refund policy but I am not sure if I can work out what I need within 14 days. Thanks, Siegfried -- The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.

3 2

is np vector a sequence?
by Neal Becker 28 Oct '13

28 Oct '13

isinstance (np.zeros (10), collections.Sequence) Out[36]: False That's unfortunate.

2 1