150

How should I compute log to the base two in python. Eg. I have this equation where I am using log base 2

import math
e = -(t/T)* math.log((t/T)[, 2])
Nicu Surdu
  • 8,172
  • 9
  • 68
  • 108
Soumya
  • 5,262
  • 8
  • 31
  • 30

10 Answers10

291

It's good to know that

log_b(a) = log(a)/log(b)

but also know that math.log takes an optional second argument which allows you to specify the base:

In [22]: import math

In [23]: math.log?
Type:       builtin_function_or_method
Base Class: <type 'builtin_function_or_method'>
String Form:    <built-in function log>
Namespace:  Interactive
Docstring:
    log(x[, base]) -> the logarithm of x to the given base.
    If the base not specified, returns the natural logarithm (base e) of x.


In [25]: math.log(8,2)
Out[25]: 3.0
alter_igel
  • 6,899
  • 3
  • 21
  • 40
unutbu
  • 842,883
  • 184
  • 1,785
  • 1,677
  • 7
    `base` argument added in version 2.3, btw. – Joe Koberg Sep 15 '10 at 18:09
  • 11
    What is this '?' syntax ? I can't find reference for it. – wap26 Apr 30 '13 at 13:59
  • 20
    @wap26: Above, I'm using the [IPython](http://ipython.org/) interactive interpreter. One of its features (accessed with the `?`) is [dynamic object introspection](http://ipython.org/ipython-doc/rel-0.13.1/overview.html#main-features-of-the-interactive-shell). – unutbu Apr 30 '13 at 17:51
103

Depends on whether the input or output is int or float.

assert 5.392317422778761 ==   math.log2(42.0)
assert 5.392317422778761 ==    math.log(42.0, 2.0)
assert 5                 ==  math.frexp(42.0)[1] - 1
assert 5                 ==            (42).bit_length() - 1

float → float math.log2(x)

import math

log2 = math.log(x, 2.0)
log2 = math.log2(x)   # python 3.3 or later

float → int math.frexp(x)

If all you need is the integer part of log base 2 of a floating point number, extracting the exponent is pretty efficient:

log2int_slow = int(math.floor(math.log(x, 2.0)))    # these give the
log2int_fast = math.frexp(x)[1] - 1                 # same result

  • Python frexp() calls the C function frexp() which just grabs and tweaks the exponent.

  • Python frexp() returns a tuple (mantissa, exponent). So [1] gets the exponent part.

  • For integral powers of 2 the exponent is one more than you might expect. For example 32 is stored as 0.5x2⁶. This explains the - 1 above. Also works for 1/32 which is stored as 0.5x2⁻⁴.

  • Floors toward negative infinity, so log₂31 computed this way is 4 not 5. log₂(1/17) is -5 not -4.

int → int x.bit_length()

If both input and output are integers, this native integer method could be very efficient:

log2int_faster = x.bit_length() - 1

  • - 1 because 2ⁿ requires n+1 bits. Works for very large integers, e.g. 2**10000.

  • Floors toward negative infinity, so log₂31 computed this way is 4 not 5.

Bob Stein
  • 16,271
  • 10
  • 88
  • 101
  • 1
    Interesting. So you're subtracting 1 there because the mantissa is in the range [0.5, 1.0)? I would give this one a few more upvotes if I could. – LarsH Feb 23 '15 at 11:49
  • 1
    Exactly right @LarsH. 32 is stored as 0.5x2⁶ so if you want log₂32=5 you need to **subtract 1**. Also true for 1/32 which is stored as 0.5x2⁻⁴. – Bob Stein Feb 23 '15 at 14:10
20

If you are on python 3.3 or above then it already has a built-in function for computing log2(x)

import math
'finds log base2 of x'
answer = math.log2(x)

If you are on older version of python then you can do like this

import math
'finds log base2 of x'
answer = math.log(x)/math.log(2)
akashchandrakar
  • 2,009
  • 3
  • 23
  • 47
11

Using numpy:

In [1]: import numpy as np

In [2]: np.log2?
Type:           function
Base Class:     <type 'function'>
String Form:    <function log2 at 0x03049030>
Namespace:      Interactive
File:           c:\python26\lib\site-packages\numpy\lib\ufunclike.py
Definition:     np.log2(x, y=None)
Docstring:
    Return the base 2 logarithm of the input array, element-wise.

Parameters
----------
x : array_like
  Input array.
y : array_like
  Optional output array with the same shape as `x`.

Returns
-------
y : ndarray
  The logarithm to the base 2 of `x` element-wise.
  NaNs are returned where `x` is negative.

See Also
--------
log, log1p, log10

Examples
--------
>>> np.log2([-1, 2, 4])
array([ NaN,   1.,   2.])

In [3]: np.log2(8)
Out[3]: 3.0
riza
  • 16,274
  • 7
  • 29
  • 29
7

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

def lg(x, tol=1e-13):
  res = 0.0

  # Integer part
  while x<1:
    res -= 1
    x *= 2
  while x>=2:
    res += 1
    x /= 2

  # Fractional part
  fp = 1.0
  while fp>=tol:
    fp /= 2
    x *= x
    if x >= 2:
        x /= 2
        res += fp

  return res
log0
  • 10,489
  • 4
  • 28
  • 62
  • Extra points for an algorithm that can be adapted to always give the correct integer part, unlike int(math.log(x, 2)) – user12861 Jan 10 '12 at 13:43
6
>>> def log2( x ):
...     return math.log( x ) / math.log( 2 )
... 
>>> log2( 2 )
1.0
>>> log2( 4 )
2.0
>>> log2( 8 )
3.0
>>> log2( 2.4 )
1.2630344058337937
>>>
puzz
  • 746
  • 4
  • 7
2

Try this ,

import math
print(math.log(8,2))  # math.log(number,base)
Akash Kandpal
  • 3,126
  • 28
  • 25
2

In python 3 or above, math class has the following functions

import math

math.log2(x)
math.log10(x)
math.log1p(x)

or you can generally use math.log(x, base) for any base you want.

Wolf
  • 9,679
  • 7
  • 62
  • 108
Masoud Mustamandi
  • 498
  • 9
  • 15
0

Don't forget that log[base A] x = log[base B] x / log[base B] A.

So if you only have log (for natural log) and log10 (for base-10 log), you can use

myLog2Answer = log10(myInput) / log10(2)
Platinum Azure
  • 45,269
  • 12
  • 110
  • 134
0

Use help method

>>> import math
>>> help(math.log)

Help on built-in function log in module math:

log(...)
    log(x, [base=math.e])
    Return the logarithm of x to the given base.
    
    If the base not specified, returns the natural logarithm (base e) of x.
(END)

log(x, [base=math.e])

Return the logarithm of x to the given base.

Udesh
  • 2,415
  • 2
  • 22
  • 32