i am trying to add some float values in python 3 (never tested in 2) and i get some odd results, the only varying factor being the order of the elements in the summation.
a = [-1e30, 1e30, 1, 3]
print(sum(a)) # return 4.0
a = [-1e30, 1, 3, 1e30]
print(sum(a)) # return 0.0
Can anyone please tell me what did i miss here?
Thanks in advance!
When you're doing sums of sequences of floating point numbers, you want to use math.fsum:
>>> a = [-1e30, 1e30, 1, 3]
>>> math.fsum(a)
4.0
>>> a = [-1e30, 1, 3, 1e30]
>>> math.fsum(a)
4.0
Using the sum builtin is not going to get you good answers for very large (or small) floating point numbers because of the inherent precision problems. You can get a pretty good view of the gory details at What Every Computer Scientist Should Know About Floating-Point Arithmetic.
Floating point numbers have a 53-bit coefficient ("the number before the e").
10**30 is way bigger than 2**53, so adding 4 to that number is lost in the limits of precision for floating point numbers.
>>> 2**53
9007199254740992
>>> 10**30
1000000000000000000000000000000
>>> float(2**53)
9007199254740992.0
>>> float(2**53) + 1
9007199254740992.0
Related
I try get ration of variable and get unexpected result. Can somebody explain this?
>>> value = 3.2
>>> ratios = value.as_integer_ratio()
>>> ratios
(3602879701896397, 1125899906842624)
>>> ratios[0] / ratios[1]
3.2
I using python 3.3
But I think that (16, 5) is much better solution
And why it correct for 2.5
>>> value = 2.5
>>> value.as_integer_ratio()
(5, 2)
Use the fractions module to simplify fractions:
>>> from fractions import Fraction
>>> Fraction(3.2)
Fraction(3602879701896397, 1125899906842624)
>>> Fraction(3.2).limit_denominator()
Fraction(16, 5)
From the Fraction.limit_denominator() function:
Finds and returns the closest Fraction to self that has denominator at most max_denominator. This method is useful for finding rational approximations to a given floating-point number
Floating point numbers are limited in precision and cannot represent many numbers exactly; what you see is a rounded representation, but the real number is:
>>> format(3.2, '.50f')
'3.20000000000000017763568394002504646778106689453125'
because a floating point number is represented as a sum of binary fractions; 1/5 can only be represented by adding up 1/8 + 1/16 + 1/128 + more binary fractions for increasing exponents of two.
It's not 16/5 because 3.2 isn't 3.2 exactly... it's a floating point rough approximation of it... eg: 3.20000000000000017764
While using the fractions module, it is better to provide a string instead of a float to avoid floating point representation issues.
For example, if you pass '3.2' instead of 3.2 you get your desired result:
In : fractions.Fraction('3.2')
Out: Fraction(16, 5)
If you already have the value stored in a variable, you can use string formatting as well.
In : value = 3.2
In : fractions.Fraction(f'{value:.2f}')
Out: Fraction(16, 5)
The following code snippet is giving 6 as a result:
import math
number = (1 - 0.99) * 500
math.ceil(number)
while the (mathematically) correct answer would be 5. Presumably this is a rounding problem - what is the best way to enforce the correct solution?
Presumably this is a rounding problem
Yes:
>>> 1 - 0.99
0.010000000000000009
>>> (1 - 0.99) * 500
5.000000000000004
what is the best way to enforce the correct solution?
You could use a decimal.Decimal instead of a float:
>>> from decimal import Decimal
>>> import math
>>> (1 - Decimal("0.99")) * 500
Decimal('5.00')
>>> math.ceil((1 - Decimal("0.99")) * 500)
5.0
It's a floating-point error since some numbers can't be represented exactly (infinitely many numbers have to be represented using a finite number of bits -- there has to be some trade-offs). This is why you lose some precision with floating point operations:
>>> 1-0.99
0.010000000000000009
Try Decimal:
>>> from decimal import Decimal as d
>>> result = (1 - d("0.99")) * 500
>>> result
Decimal('5.00')
>>> math.ceil(result)
5.0
Edit
It may look like all the numbers have exact representations:
>>> a = 1.0; b = 0.99; c = 0.01
>>> a, b, c
(1.0, 0.99, 0.01)
So this result might seem surprising:
>>> a - b
0.010000000000000009
>>> a - b == c
False
But it's just the precision and rounding errors that accumulate. Here are the same numbers and calculation, but showing more digits:
>>> def o(f): return "%.30f" % f
>>> o(a)
'1.000000000000000000000000000000'
>>> o(b)
'0.989999999999999991118215802999'
>>> o(c)
'0.010000000000000000208166817117'
>>> o(a-b)
'0.010000000000000008881784197001'
Python 2.7 rounds to 17 significant digits. It is a different model from real math.
The given answers are correct, this is a case of rounding error. However, I think it would be useful to include why this happens.
In hardware, floating point numbers are base 2 (AKA binary). The problem is that most decimal fractions cannot be represented exactly as binary fractions. The translation of that is (in general) floating point numbers are only approximated by the binary floating point numbers actually stored in the machine.
I try get ration of variable and get unexpected result. Can somebody explain this?
>>> value = 3.2
>>> ratios = value.as_integer_ratio()
>>> ratios
(3602879701896397, 1125899906842624)
>>> ratios[0] / ratios[1]
3.2
I using python 3.3
But I think that (16, 5) is much better solution
And why it correct for 2.5
>>> value = 2.5
>>> value.as_integer_ratio()
(5, 2)
Use the fractions module to simplify fractions:
>>> from fractions import Fraction
>>> Fraction(3.2)
Fraction(3602879701896397, 1125899906842624)
>>> Fraction(3.2).limit_denominator()
Fraction(16, 5)
From the Fraction.limit_denominator() function:
Finds and returns the closest Fraction to self that has denominator at most max_denominator. This method is useful for finding rational approximations to a given floating-point number
Floating point numbers are limited in precision and cannot represent many numbers exactly; what you see is a rounded representation, but the real number is:
>>> format(3.2, '.50f')
'3.20000000000000017763568394002504646778106689453125'
because a floating point number is represented as a sum of binary fractions; 1/5 can only be represented by adding up 1/8 + 1/16 + 1/128 + more binary fractions for increasing exponents of two.
It's not 16/5 because 3.2 isn't 3.2 exactly... it's a floating point rough approximation of it... eg: 3.20000000000000017764
While using the fractions module, it is better to provide a string instead of a float to avoid floating point representation issues.
For example, if you pass '3.2' instead of 3.2 you get your desired result:
In : fractions.Fraction('3.2')
Out: Fraction(16, 5)
If you already have the value stored in a variable, you can use string formatting as well.
In : value = 3.2
In : fractions.Fraction(f'{value:.2f}')
Out: Fraction(16, 5)
I want to be able to compare Decimals in Python. For the sake of making calculations with money, clever people told me to use Decimals instead of floats, so I did. However, if I want to verify that a calculation produces the expected result, how would I go about it?
>>> a = Decimal(1./3.)
>>> a
Decimal('0.333333333333333314829616256247390992939472198486328125')
>>> b = Decimal(2./3.)
>>> b
Decimal('0.66666666666666662965923251249478198587894439697265625')
>>> a == b
False
>>> a == b - a
False
>>> a == b - Decimal(1./3.)
False
so in this example a = 1/3 and b = 2/3, so obviously b-a = 1/3 = a, however, that cannot be done with Decimals.
I guess a way to do it is to say that I expect the result to be 1/3, and in python i write this as
Decimal(1./3.).quantize(...)
and then I can compare it like this:
(b-a).quantize(...) == Decimal(1./3.).quantize(...)
So, my question is: Is there a cleaner way of doing this? How would you write tests for Decimals?
You are not using Decimal the right way.
>>> from decimal import *
>>> Decimal(1./3.) # Your code
Decimal('0.333333333333333314829616256247390992939472198486328125')
>>> Decimal("1")/Decimal("3") # My code
Decimal('0.3333333333333333333333333333')
In "your code", you actually perform "classic" floating point division -- then convert the result to a decimal. The error introduced by floats is propagated to your Decimal.
In "my code", I do the Decimal division. Producing a correct (but truncated) result up to the last digit.
Concerning the rounding. If you work with monetary data, you must know the rules to be used for rounding in your business. If not so, using Decimal will not automagically solve all your problems. Here is an example: $100 to be share between 3 shareholders.
>>> TWOPLACES = Decimal(10) ** -2
>>> dividende = Decimal("100.00")
>>> john = (dividende / Decimal("3")).quantize(TWOPLACES)
>>> john
Decimal('33.33')
>>> paul = (dividende / Decimal("3")).quantize(TWOPLACES)
>>> georges = (dividende / Decimal("3")).quantize(TWOPLACES)
>>> john+paul+georges
Decimal('99.99')
Oups: missing $.01 (free gift for the bank ?)
Your verbiage states you want to to monetary calculations, minding your round off error. Decimals are a good choice, as they yield EXACT results under addition, subtraction, and multiplication with other Decimals.
Oddly, your example shows working with the fraction "1/3". I've never deposited exactly "one-third of a dollar" in my bank... it isn't possible, as there is no such monetary unit!
My point is if you are doing any DIVISION, then you need to understand what you are TRYING to do, what the organization's policies are on this sort of thing... in which case it should be possible to implement what you want with Decimal quantizing.
Now -- if you DO really want to do division of Decimals, and you want to carry arbitrary "exactness" around, you really don't want to use the Decimal object... You want to use the Fraction object.
With that, your example would work like this:
>>> from fractions import Fraction
>>> a = Fraction(1,3)
>>> a
Fraction(1, 3)
>>> b = Fraction(2,3)
>>> b
Fraction(2, 3)
>>> a == b
False
>>> a == b - a
True
>>> a + b == Fraction(1, 1)
True
>>> 2 * a == b
True
OK, well, even a caveat there: Fraction objects are the ratio of two integers, so you'd need to multiply by the right power of 10 and carry that around ad-hoc.
Sound like too much work? Yes... it probably is!
So, head back to the Decimal object; implement quantization/rounding upon Decimal division and Decimal multiplication.
Floating-point arithmetics is not accurate :
Decimal numbers can be represented exactly. In contrast, numbers like
1.1 and 2.2 do not have exact representations in binary floating point. End users typically would not expect 1.1 + 2.2 to display as
3.3000000000000003 as it does with binary floating point
You have to choose a resolution and truncate everything past it :
>>> from decimal import *
>>> getcontext().prec = 6
>>> Decimal(1) / Decimal(7)
Decimal('0.142857')
>>> getcontext().prec = 28
>>> Decimal(1) / Decimal(7)
Decimal('0.1428571428571428571428571429')
You will obviously get some rounding error which will grow with the number of operations so you have to choose your resolution carefully.
There is another approach that may work for you:
Continue to do all your calculations in floating point values
When you need to compare for equality, use round(val, places)
For example:
>>> a = 1./3
>>> a
0.33333333333333331
>>> b = 2./3
>>> b
0.66666666666666663
>>> b-a
0.33333333333333331
>>> round(a,2) == round(b-a, 2)
True
If you'd like, create a function equals_to_the_cent():
>>> def equals_to_the_cent(a, b):
... return round(a, 2) == round(b, 2)
...
>>> equals_to_the_cent(a, b)
False
>>> equals_to_the_cent(a, b-a)
True
>>> equals_to_the_cent(1-a, b)
True
This question already has answers here:
Why does floating-point arithmetic not give exact results when adding decimal fractions?
(31 answers)
Closed 3 years ago.
i am kind of confused why python add some additional decimal number in this case, please help to explain
>>> mylist = ["list item 1", 2, 3.14]
>>> print mylist ['list item 1', 2, 3.1400000000000001]
Floating point numbers are an approximation, they cannot store decimal numbers exactly. Because they try to represent a very large range of numbers in only 64 bits, they must approximate to some extent.
It is very important to be aware of this, because it results in some weird side-effects. For example, you might very reasonably think that the sum of ten lots of 0.1 would be 1.0. While this seems logical, it is also wrong when it comes to floating point:
>>> f = 0.0
>>> for _ in range (10):
... f += 0.1
...
>>> print f == 1.0
False
>>> f
0.99999999999999989
>>> str(f)
1.0
You might think that n / m * m == n. Once again, floating-point world disagrees:
>>> (1.0 / 103.0) * 103.0
0.99999999999999989
Or perhaps just as strangely, one might think that for all n, n + 1 != n. In floating point land, numbers just don't work like this:
>>> 10.0**200
9.9999999999999997e+199
>>> 10.0**200 == 10.0**200 + 1
True
# How much do we have to add to 10.0**200 before its
# floating point representation changes?
>>> 10.0**200 == 10.0**200 + 10.0**183
True
>>> 10.0**200 == 10.0**200 + 10.0**184
False
See What every computer scientist should know about floating point numbers for an excellent summary of the issues.
If you need exact decimal representation, check out the decimal module, part of the python standard library since 2.4. It allows you to specify the number of significant figures. The downside is, it is much slower than floating point, because floating point operations are implemented in hardware whereas decimal operations happen purely in software. It also has its own imprecision issues, but if you need exact representation of decimal numbers (e.g. for a financial application) it's ideal.
For example:
>>> 3.14
3.1400000000000001
>>> import decimal
>>> decimal.Decimal('3.14')
>>> print decimal.Decimal('3.14')
3.14
# change the precision:
>>> decimal.getcontext().prec = 6
>>> decimal.Decimal(1) / decimal.Decimal(7)
Decimal('0.142857')
>>> decimal.getcontext().prec = 28
>>> decimal.Decimal(1) / decimal.Decimal(7)
Decimal('0.1428571428571428571428571429')
It is worthwhile to note that Python 3.1 has a new floating point output routine that rounds this in the expected manner (it has also been backported to Python 2.7):
Python 3.1 (r31:73572, Aug 15 2009, 17:12:41)
[GCC 4.0.1 (Apple Computer, Inc. build 5367)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> a = [3.14]
>>> print(a)
[3.14]
From the What's New in Python 3.1 document:
Python now uses David Gay’s algorithm for finding the shortest floating point representation that doesn’t change its value. This should help mitigate some of the confusion surrounding binary floating point numbers.
The significance is easily seen with a number like 1.1 which does not have an exact equivalent in binary floating point. Since there is no exact equivalent, an expression like float('1.1') evaluates to the nearest representable value which is 0x1.199999999999ap+0 in hex or 1.100000000000000088817841970012523233890533447265625 in decimal. That nearest value was and still is used in subsequent floating point calculations.
As mentioned before, it's all about floating points being an approximation.
If you want exactness you can use a decimal (which is a precise representation):
http://docs.python.org/library/decimal.html
a = [1.5, 1.49999]
a
[1.5, 1.4999899999999999]
from decimal import Decimal
b = [1.5, Decimal('1.4999')]
b
[1.5, Decimal('1.4999')]
We can fix it by this command:
>>> x = 1.2 - 1.0
>>> x
0.19999999999999996
>>> y = float(str(x))
>>> y
0.2
I add an answer from #mark