[Numpy-discussion] Errors with "**" for numpy.float16

[Edward C. Jones]

#! /usr/bin/env python3.2

import numpy

for t in (numpy.float16, numpy.float32, numpy.float64, numpy.float128):
 two = t(2)
 print(t, two, two ** two, numpy.power(two, two))

"""
I use up-to-date debian testing (wheezy), amd64 architecture.  The python
package is python3, version 3.2.3~rc1-2.  The numpy package is 
python3-numpy,
version 1:1.6.2~rc1-1.

"**" has problems for float16:

 2.0 1.1921e-07 4.0
 2.0 4.0 4.0
 2.0 4.0 4.0
 2.0 4.0 4.0
"""

___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion

[Numpy-discussion] Incorrect overflow warning message for float128

[Edward C. Jones]

I use up-to-date Debian testing (wheezy), amd64 architecture.

 From the docs for numpy.MachAr:

maxexp int Smallest (positive) power of ibeta that causes overflow.

On my machine, ibeta = 2 and maxexp = 16384.

For float64, float32, and float16 things behave as expected.  For float128,
I get the message about overflow for exponent 8192 (and greater) but the
correct answer is printed.  What is the problem?

#! /usr/bin/env python3.2

import numpy

print('float128')
fi = numpy.finfo(numpy.float128)
print('ibeta:', fi.machar.ibeta)  # 2
print('maxexp:', fi.machar.maxexp)  # 16384
print('xmax:', fi.machar.xmax)  # 1.18973149536e+4932
two = numpy.float128(2)
big = numpy.float128(8191)
x = numpy.power(two, big)
print('2**8191:', x)  # 5.4537406781e+2465
big = numpy.float128(8192)
# RuntimeWarning: overflow encountered in power
x = numpy.power(two, big)
print('2**8192:', x)  # 1.09074813562e+2466

___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion

Re: [Numpy-discussion] Are "min", "max" documented for scalars?

[Edward C. Jones]

Silly mistakes.

If a and b are Python ints, Python floats, or non-complex 
numpy.number's,  "max" returns, unchanged, the largrt of the two 
objects.  There is no coercion to a common type.  This useful behavior 
needs to be documented.

___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion

[Numpy-discussion] Are "min", "max" documented for scalars?

[Edward C. Jones]

Python "max" and "min" have an interesting and _useful_ behavior when
applied to numpy scalars and Python numbers.  Here is a piece of
pseudo-code:

def max(a, b):
 if int(b) > int(a):
 return b
 else:
 return a

The larger object is returned unchanged.  If the two objects are equal,
return the first unchanged.

Is the behavior of "max", "min", "<", "<=", etc. for numpy scalar objects
documented somewhere?

keywords: greater than less than

___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion

[Numpy-discussion] numpy.clip behavior at max and min of dtypes

[Edward C. Jones]

Can the following function be written using numpy.clip?  In some other way?
Does numpy.clip satisfy condition 4 below?  Does numpy.clip satisfy some
closely related condition?

Define a function clipcast:
 output = clipcast(arr, dtype=None, out=None)

1. All arrays have int or float dtypes.

2. Exactly one of the keyword arguments "dtype" and "out" must be used.  If
"dtype" is given, then output has that dtype.

3. "output" has the same shape as "arr".

4. Let ER be the set of all the real numbers that can be exactly represented
by the output dtype.  ER is finite and bounded.  Let themin = min(ER) and
themax = max(ER).  For any real number x, define a function f(x) by

If x is in ER, define f(x) = x.

If x is between two consecutive numbers, u and v, in ER, then define
   f(x) = u or f(x) = v.  Probably the choice would be made using a C
   cast.

If x < themin, define f(x) = themin.

If x > themax, define f(x) = themax.

If x is an element of arr, say Arr[I], then output[I] == f(x) where I is any
index that defines a single element of arr.

___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion

[Numpy-discussion] Permuting sparse arrays

[Edward C. Jones]

I have a vector of bits where there are many more zeros than one.  I 
store the array as a sorted list of the indexes where the bit is one.  
If the bit array is (0, 1, 0, 0, 0, 1, 1), it is stored as (1, 5, 6).  
If the bit array, b, has length n, and p is a random permutation of 
arange(n), then I can permute the bit array using fancy indexing: b[p].  
Is there some neat trick I can use to permute an array while leaving it 
in the list-of-indexes form?  Currently I am doing it with a Python loop 
but I am looking for a faster way.
___
NumPy-Discussion mailing list
NumPy-Discussion@scipy.org
http://mail.scipy.org/mailman/listinfo/numpy-discussion