classification
Title: Generalize math.hypot function
Type: enhancement Stage:
Components: Library (Lib) Versions: Python 3.0
process
Status: closed Resolution: rejected
Dependencies: Superseder:
Assigned To: Nosy List: LambertDW, georg.brandl, mark.dickinson
Priority: low Keywords:

Created on 2008-01-21 02:58 by LambertDW, last changed 2008-01-21 10:10 by georg.brandl. This issue is now closed.

Messages (3)
msg61374 - (view) Author: David W. Lambert (LambertDW) Date: 2008-01-21 02:58
Please generalize math.hypot.  While I don't have a survey of python 
codes, it seems to me unlikely for this change to break existing 
programs.

import math

def hypot(*args):
    '''
        Return the Euclidean vector length.
        >>> from math import hypot, sqrt
        >>> hypot(5,12)    # traditional definition
        13.0
        >>> hypot()
        0.0
        >>> hypot(-6.25)
        6.25
        >>> hypot(1,1,1) == sqrt(3) # diagonal of unit box
        True
    '''
    return math.sqrt(sum(arg*arg for arg in args))


I propose this version as closest to:
>>> print sys.version
2.5.1 (r251:54863, Jan  4 2008, 17:15:14) 
[GCC 3.4.4 (cygming special, gdc 0.12, using dmd 0.125)]
>>> print math.hypot.__doc__
hypot(x,y)

Return the Euclidean distance, sqrt(x*x + y*y).

Thanks,
Dave.

PS.  I don't understand why python is so restrictive.  Although hypot 
is in the math library, it could be written in EAFP style as

def hypot(*args):
    return sum(arg*arg for arg in args)**0.5

Rather than review the entire python library for items to generalize, 
I'll accept that the resulting errors would confuse "the penguin on my 
tele".  "hypot" crosses me most often.  I've not yet needed a version 
in the complex domain, such as my second version.

I typically fill my need for length with scipy.sqrt(scipy.dot(v,v)), 
only to realize that for the short vectors I use, standard python 
constructs always perform faster than scipy
msg61377 - (view) Author: Mark Dickinson (mark.dickinson) * (Python committer) Date: 2008-01-21 04:33
I'm not sure either that such a generalization would belong in math (which right now 
does little more than expose some basic functions from the C library) or that it 
should be called hypot.

It seems to me that this would belong with other vector-type math stuff, but there 
isn't any of that in core Python or the libraries at the moment.

On one hand, this falls foul of the 'not every one-liner should be a builtin' rule.  
On the other hand, it's not a one-liner if done properly:  compare the output of 
math.hypot(1e160, 1e160) with the output of your version.  Similarly for 
math.hypot(1e-160, 1e-160).
msg61380 - (view) Author: Georg Brandl (georg.brandl) * (Python committer) Date: 2008-01-21 10:10
IOW, this is rejected.
History
Date User Action Args
2008-01-21 10:10:22georg.brandlsetstatus: open -> closed
resolution: rejected
messages: + msg61380
nosy: + georg.brandl
2008-01-21 04:33:23mark.dickinsonsetpriority: low
nosy: + mark.dickinson
messages: + msg61377
2008-01-21 02:58:51LambertDWcreate