gmane.comp.python.numeric.general

Re: Equvalent function for Ceil() and Floor()

Robert Kern — 2013-05-20T19:51:51

On Mon, May 20, 2013 at 6:58 PM, Bakhtiyor Zokhidov
<bakhtiyor_zokhidov< at >mail.ru> wrote:

It doesn't affect those computations, no. It does have different
results for division (1/-0.0 -> -inf), and it will often determine
which branch gets evaluated for a complex function that has a branch
point at the origin. These are usually considered good things and are
the primary reason that floating point includes signed zeros.

http://en.wikipedia.org/wiki/Signed_zero

--
Robert Kern

Re: Equvalent function for Ceil() and Floor()

Bakhtiyor Zokhidov — 2013-05-20T17:58:13

 ok... I think -0.0 is mathematically wrong but in a program it is true.

What I suspect is that if -0.0 doesn't affect on result (e.g., 2*(-0.0 + 2) or (-0.0-2)*2 ) . If it does not affect results it would be good for me
Thanks


Понедельник, 20 мая 2013, 9:03 -07:00 от Chris Barker - NOAA Federal <chris.barker< at >noaa.gov>:

_______________________________________________
NumPy-Discussion mailing list
NumPy-Discussion< at >scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion

Re: Equvalent function for Ceil() and Floor()

Chris Barker - NOAA Federal — 2013-05-20T16:03:53

On Mon, May 20, 2013 at 8:54 AM, Bakhtiyor Zokhidov
<bakhtiyor_zokhidov< at >mail.ru> wrote:


ceil rounds toward +inf (and floor towards -inf) -- this is exactly
what you want if you're doing what I think you are...(note that
round() rounds towards and away from zero -- so different for negative
numbers...)

the minus sign means "minus 0.0", which is a seemingly odd floating
point thing -- but FP numbers have a sign bit, which, in this case, is
set to negative, even though, mathematically speaking, 0.0 doesn't
have a sign. But it will compare to 0.0 as you'd expect:

In [8]: -0.0 == 0.0
Out[8]: True

In [9]: -0.0 <  0.0
Out[9]: False

-Chris

Re: another indexing question

Robert Kern — 2013-05-20T16:03:55

All you need is a single loop over the alphabet, which is usually not
problematic.


means = np.empty([M])
for i in range(M):
    means[i] = y[x == i].mean()

--
Robert Kern

another indexing question

Neal Becker — 2013-05-20T16:00:31

I have a system that transmits signals for an alphabet of M symbols
over and additive Gaussian noise channel.  The receiver has a
1-d array of complex received values.  I'd like to find the means
of the received values according to the symbol that was transmitted.

So transmit symbol indexes might be:

x = [0, 1, 2, 1, 3, ...]

and receive output might be:

y = [(1+1j), (1-1j), ...]

Suppose the alphabet was M=4.  Then I'd like to get an array of means

m[0...3] that correspond to the values of y for each of the corresponding
values of x.

I can't think of a better way than manually using loops.  Any tricks here?

Re: Equvalent function for Ceil() and Floor()

Robert Kern — 2013-05-20T15:55:47

On Mon, May 20, 2013 at 4:54 PM, Bakhtiyor Zokhidov
<bakhtiyor_zokhidov< at >mail.ru> wrote:

What about it? What were you expecting to get? What do you think is
wrong about the answer you did get?

--
Robert Kern

Re: Equvalent function for Ceil() and Floor()

Bakhtiyor Zokhidov — 2013-05-20T15:54:26

 thanks a lot

what about the following example:
-0.0
???
thanks in advance for the reply


Понедельник, 20 мая 2013, 16:37 +01:00 от Robert Kern <robert.kern< at >gmail.com>:

_______________________________________________
NumPy-Discussion mailing list
NumPy-Discussion< at >scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion

Re: Equvalent function for Ceil() and Floor()

Robert Kern — 2013-05-20T15:37:16

On Mon, May 20, 2013 at 4:21 PM, Bakhtiyor Zokhidov
<bakhtiyor_zokhidov< at >mail.ru> wrote:

For most purposes, the following functions suffice:

def new_ceil(x, step):
    return math.ceil(x / step) * step

def new_floor(x, step):
    return math.floor(x / step) * step


Alternately:

def new_ceil(x, step):
    quotient = x // step
    remainder = x % step
    return (quotient + (remainder > 0)) * step

def new_floor(x, step):
    quotient = x // step
    return quotient * step


Floating point representation errors and accumulated floating point
arithmetic inaccuracies may give you unexpected results in many cases,
so be careful.

--
Robert Kern

Re: Equvalent function for Ceil() and Floor()

Nathaniel Smith — 2013-05-20T15:36:20

On Mon, May 20, 2013 at 4:21 PM, Bakhtiyor Zokhidov
<bakhtiyor_zokhidov< at >mail.ru> wrote:

def new_ceil(x, step):
  return np.ceil(x / step) * step

?

-n

Equvalent function for Ceil() and Floor()

Bakhtiyor Zokhidov — 2013-05-20T15:21:50

 Hello,

I am using ceil() and floor() function to get upper and lower value of some numbers. Let's say:

import math
x1 = 0.35
y1 = 4.46
1.0
4.0

The problem is that If I want to get upper and lower values for the certain step, for example, step = 0.25, ceil() function should give:
new_ceil(x1, step) => 0.5
new_floor(y1, step) => 4.25
Because, the step is 0.25

Question: How I can I achieve those results by using ceil() and floor() function, or Is there any equvalent function for that?
--  Bakhti_______________________________________________
NumPy-Discussion mailing list
NumPy-Discussion< at >scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion

Re: Newbie trying to install NumPy

Chris Barker - NOAA Federal — 2013-05-19T22:14:22


Make sure you match 32/64 bit. The message is a bit out of date,
you'll get the same error if you try to install a 32 bit numpy and
have a 64 bit python.

-Chris


_______________________________________________
NumPy-Discussion mailing list
NumPy-Discussion< at >scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion

Re: Newbie trying to install NumPy

2013-05-18T11:34:11


Did you try to do this (what it says here)?

Josef



_______________________________________________
NumPy-Discussion mailing list
NumPy-Discussion< at >scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion

Re: Newbie trying to install NumPy

Daπid — 2013-05-18T08:52:41

You probably have a different version installed. Grab Python 2.7 from
python.org and install it; then, run the numpy installer. It should
just work out of the box.

Re: Newbie trying to install NumPy

Johann Cohen-Tanugi — 2013-05-18T07:28:01

Hello Joe, I am not familiar with windows, but I believe that the 
Python(x,y) is here to help you : http://code.google.com/p/pythonxy/
HTH,
Johann

On 05/18/2013 07:11 AM, Joe Piccoli wrote:

faster (selection based) median, 2013 edition

Julian Taylor — 2013-05-18T06:12:45

hi,

once again I want to bring up the median algorithm which is implemented
in terms of sorting in numpy.
median (and percentile and a couple more functions) can be more
efficiently implemented in terms of a selection algorithm. The
complexity can them be linear instead of linearithmic.

I found numerous discussions of this in the list archives [1, 2, 3] but
I did not find why those attempts failed, the threads all just seemed to
stop.
Did the previous attempts fail due to lack of time or was there a
fundamental reason blocking this change?

In the hope of the former, I went ahead and implemented a prototype of a
partition function (similar to [3] but only one argument) and
implemented median in terms of it.
partition not like C++ partition, its equivalent to nth_element in C++,
maybe its better to name it nth_element?

The code is available here:
https://github.com/juliantaylor/numpy/tree/select-median

the partition interface is:
ndarray.partition(kth, axis=-1)
kth is an integer
The array is transformed so the k-th element of the array is in its
final sorted order, all below are smaller all above are greater, but the
ordering is undefined

Example:
In [1]: d = np.arange(10); np.random.shuffle(d)
In [2]: d
Out[2]: array([1, 7, 0, 2, 5, 6, 8, 9, 3, 4])
In [3]: np.partition(d, 3)
Out[3]: array([0, 1, 2, 3, 4, 6, 8, 9, 7, 5])
In [4]:  _[3] == 3
Out[5]: True

the performance of median improves as expected:
old vs new, 5000, uniform shuffled, out of place:
100us vs 40us
old vs new, 50000, uniform shuffled, out of place:
1.12ms vs 0.265ms
old vs new, 500000, uniform shuffled, out of place:
14ms vs 2.81ms

The implementation is very much still a prototype, apartition is not
exposed (and only implemented as a quicksort) and there is only one
algorithm (quickselect). One could still add median of medians for
better worst case performance.

If no blockers appear I want to fix this up and file a pull request to
have this in numpy 1.8.
Guidance on details of implementation in numpys C api is highly
appreciated, its the first time I'm dealing with it.

Cheers,
Julian Taylor

[1]
http://thread.gmane.org/gmane.comp.python.numeric.general/50931/focus=50941
[2]
http://thread.gmane.org/gmane.comp.python.numeric.general/32507/focus=41716
[3]
http://thread.gmane.org/gmane.comp.python.numeric.general/32341/focus=32348

Newbie trying to install NumPy

Joe Piccoli — 2013-05-18T05:11:16

Hello,

 

I've been trying to install NumPy to run with Eclipse on Windows Vista.
After installing (I thought) NumPy I was seeing:

 

ImportError: Error importing numpy: you should not try to import numpy from

        its source directory; please exit the numpy source tree, and
relaunch

        your python intepreter from there.

 

I next tried to follow the instructions from the scipy.org website and
downloaded and ran:

 

numpy-1.7.1-win32-superpack-python27.exe

 

This started up but I immediately saw the following dialog:

 

---------------------------

Cannot install

---------------------------

Python version 2.7 required, which was not found in the registry.

---------------------------

OK   

---------------------------

 

The next dialog prompted for a Python installation to use but the list box
was empty and it would not allow me to enter a path.

 

Is it absolutely necessary to build NumPy myself or is there a working
installation out there? I know I'm doing something wrong but I don't know
what it is. Any assistance would be greatly appreciated :).

 

Thanks,

Joseph A. Piccoli

joe13676< at >comcast.net

 

_______________________________________________
NumPy-Discussion mailing list
NumPy-Discussion< at >scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion

ANN: python-blosc 1.1 RC1 available for testing

Francesc Alted — 2013-05-17T18:36:51

================================
Announcing python-blosc 1.1 RC1
================================

What is it?
===========

python-blosc (http://blosc.pydata.org) is a Python wrapper for the
Blosc compression library.

Blosc (http://blosc.org) is a high performance compressor optimized for
binary data.  It has been designed to transmit data to the processor
cache faster than the traditional, non-compressed, direct memory fetch
approach via a memcpy() OS call.  Whether this is achieved or not
depends of the data compressibility, the number of cores in the system,
and other factors.  See a series of benchmarks conducted for many
different systems: http://blosc.org/trac/wiki/SyntheticBenchmarks.

Blosc works well for compressing numerical arrays that contains data
with relatively low entropy, like sparse data, time series, grids with
regular-spaced values, etc.

There is also a handy command line for Blosc called Bloscpack
(https://github.com/esc/bloscpack) that allows you to compress large
binary datafiles on-disk.  Although the format for Bloscpack has not
stabilized yet, it allows you to effectively use Blosc from your
favorite shell.


What is new?
============

- Added new `compress_ptr` and `decompress_ptr` functions that allows to
   compress and decompress from/to a data pointer.  These are low level
   calls and user must make sure that the pointer data area is safe.

- Since Blosc (the C library) already supports to be installed as an
   standalone library (via cmake), it is also possible to link
   python-blosc against a system Blosc library.

- The Python calls to Blosc are now thread-safe (another consequence of
   recent Blosc library supporting this at C level).

- Many checks on types and ranges of values have been added.  Most of
   the calls will now complain when passed the wrong values.

- Docstrings are much improved. Also, Sphinx-based docs are available
   now.

Many thanks to Valentin Hänel for his impressive work for this release.

For more info, you can see the release notes in:

https://github.com/FrancescAlted/python-blosc/wiki/Release-notes

More docs and examples are available in the documentation site:

http://blosc.pydata.org


Installing
==========

python-blosc is in PyPI repository, so installing it is easy:

$ pip install -U blosc  # yes, you should omit the blosc- prefix


Download sources
================

The sources are managed through github services at:

http://github.com/FrancescAlted/python-blosc


Documentation
=============

There is Sphinx-based documentation site at:

http://blosc.pydata.org/


Mailing list
============

There is an official mailing list for Blosc at:

blosc< at >googlegroups.com
http://groups.google.es/group/blosc


Licenses
========

Both Blosc and its Python wrapper are distributed using the MIT license.
See:

https://github.com/FrancescAlted/python-blosc/blob/master/LICENSES

for more details.

--
Francesc Alted

[ANN] Multidimensional Array - MDArray (0.5.0)

Rodrigo Botafogo — 2013-05-17T13:20:47

Although this is not directly connected to NumPy, I believe that it could
be of interest to the NymPy community.  If, by any reason it is inproper to
post this type of announcement on this list, please let me know.

I´m happy to announce a new version of MDArray...


MDArray
=======

MDArray is a multi dimensional array implemented for JRuby inspired by
NumPy (www.numpy.org)
and Narray (narray.rubyforge.org) by Masahiro Tanaka.  MDArray stands on
the shoulders of
Java-NetCDF and Parallel Colt.

NetCDF-Java Library is a Java interface to NetCDF files, as well as to many
other types of
scientific data formats.  It is developed and distributed by Unidata (
http://www.unidata.ucar.edu).

Parallel Colt (sites.google.com/site/piotrwendykier/software/parallelcolt)
is a multithreaded
version of Colt (http://acs.lbl.gov/software/colt/).  Colt provides a set
of Open Source
Libraries for High Performance Scientific and Technical Computing in Java.
Scientific
and technical computing is characterized by demanding problem sizes and a
need for high
performance at reasonably small memory footprint.

MDArray and SciRuby
===================

MDArray subscribes fully to the SciRuby Manifesto (http://sciruby.com/).

"Ruby has for some time had no equivalent to the beautifully constructed
NumPy, SciPy,
and matplotlib libraries for Python.

We believe that the time for a Ruby science and visualization package has
come. Sometimes
when a solution of sugar and water becomes super-saturated, from it
precipitates a pure,
delicious, and diabetes-inducing crystal of sweetness, induced by no more
than the tap
of a finger. So is occurring now, we believe, with numeric and
visualization libraries for Ruby."

Main properties
===============

  + Homogeneous multidimensional array, a table of elements (usually
numbers), all of the
      same type, indexed by a tuple of positive integers;
  + Easy calculation for large numerical multi dimensional arrays;
  + Basic types are: boolean, byte, short, int, long, float, double,
string, structure;
  + Based on JRuby, which allows importing Java libraries;
  + Operator: +,-,*,/,%,**, >, >=, etc.
  + Functions: abs, ceil, floor, truncate, is_zero, square, cube, fourth;
  + Binary Operators: &, |, ^, ~ (binary_ones_complement), <<, >>;
  + Ruby Math functions: acos, acosh, asin, asinh, atan, atan2, atanh,
cbrt, cos, erf, exp,
      gamma, hypot, ldexp, log, log10, log2, sin, sinh, sqrt, tan, tanh,
neg;
  + Boolean operations on boolean arrays: and, or, not;
  + Fast descriptive statistics from Parallel Colt (complete list found
bellow);
  + Easy manipulation of arrays: reshape, reduce dimension, permute,
section, slice, etc.
  + Reading of two dimensional arrays from CSV files (mainly for debugging
and simple
      testing purposes);
  + StatList: a list that can grow/shrink and that can compute Parallel
Colt descriptive
      statistics.

Descriptive statistics methods
==============================

auto_correlation, correlation, covariance, durbin_watson, frequencies,
geometric_mean,
harmonic_mean, kurtosis, lag1, max, mean, mean_deviation, median, min,
moment, moment3,
moment4, pooled_mean, pooled_variance, product, quantile, quantile_inverse,
rank_interpolated, rms, sample_covariance, sample_kurtosis,
sample_kurtosis_standard_error, sample_skew, sample_skew_standard_error,
sample_standard_deviation, sample_variance, sample_weighted_variance, skew,
split,
standard_deviation, standard_error, sum, sum_of_inversions,
sum_of_logarithms,
sum_of_powers, sum_of_power_deviations, sum_of_squares,
sum_of_squared_deviations,
trimmed_mean, variance, weighted_mean, weighted_rms, weighted_sums,
winsorized_mean.

Installation and download
=========================

  + Install Jruby
  + jruby -S gem install mdarray

Contributors
============

  + Contributors are welcome.

Homepages
=========

  + http://rubygems.org/gems/mdarray
  + https://github.com/rbotafogo/mdarray/wiki


HISTORY
=======

  + 16/05/2013: Version 0.5.0: All loops transfered to Java with over 50%
performance
      improvement.  Descriptive statistics from Parallel Colt.
  + 19/04/2013: Version 0.4.3: Fixes a simple (but fatal bug).  No new
features
  + 17/04/2013: Version 0.4.2: Adds simple statistics and boolean operators
  + 05/05/2013: Version 0.4.0: Initial release

Re: RuntimeWarning: divide by zero encounteredin log

Nathaniel Smith — 2013-05-17T13:20:04

how
you

Try adding some parens?

np.seterrcall (lambda a,b: traceback.print_stack())

-n
_______________________________________________
NumPy-Discussion mailing list
NumPy-Discussion< at >scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion

Re: RuntimeWarning: divide by zero encounteredin log

Neal Becker — 2013-05-17T13:14:35


Here's a trivial test:

import numpy as np

import traceback

np.seterrcall (lambda a,b: traceback.print_stack)
np.seterr (all='call')
np.seterrcall (lambda a,b: traceback.print_stack)

print 1./0

Re: RuntimeWarning: divide by zero encounteredin log

Neal Becker — 2013-05-17T13:12:53


I tried this:

import traceback

np.seterrcall (lambda a,b: traceback.print_stack)
np.seterr (all='call')
np.seterrcall (lambda a,b: traceback.print_stack)

but it doesn't seem to do anything, I still see numpy warning as before.