float is legacy
DESCRIPTION
My presentation in RubyConf 2011.You can see the description at http://rubyconf.org/presentations/33You can get the implementation proposed this presentation from https://github.com/mrkn/ruby/tree/decimal_rational_implementationTRANSCRIPT
Float is LegacyKenta Murata
RubyConf 2011
1Monday, October 10, 11
http://www.flickr.com/photos/recompile_net/5951998279/
Kenta MurataCRuby committer
bigdecimal maintainer
OS X platform maintainer
Interested in number system
@mrkn
2Monday, October 10, 11
https://twitter.com/#!/shyouhei/status/1198029834238074883Monday, October 10, 11
https://twitter.com/#!/shyouhei/status/1198029834238074883Monday, October 10, 11
http://www.flickr.com/photos/recompile_net/5951998279/
Kenta MurataCRuby committer
bigdecimal maintainer
OS X platform maintainer
Interested in number system
Ruby Sapporo
@mrkn
4Monday, October 10, 11
Sapporo, Japanhttp://www.flickr.com/photos/muraken/6174655831
5Monday, October 10, 11
Sapporo, Japanhttp://www.flickr.com/photos/irasally/4708650832/
6Monday, October 10, 11
The RubyKaigi is finished.
7Monday, October 10, 11
Regional RubyKaigi is continue.
8Monday, October 10, 11
Sapporo RubyKaigi 04
in the next summer.
9Monday, October 10, 11
Official informationwill be coming soon.
10Monday, October 10, 11
Acknowledgement
Tatsuhiro Ujihisa, @ujmHootSuite Media, Inc.
Yoshimasa Niwa, @niwTwitter, Inc.
11Monday, October 10, 11
Float is LegacyKenta Murata
RubyConf 2011
12Monday, October 10, 11
Summary
Float requires us the advanced knowledge
Most rubyists don’t need Float
Rational is enough for us
Literal of decimal fraction interpreted as Rational makes us more happy
13Monday, October 10, 11
Float class
14Monday, October 10, 11
What is Float class?
A wrapper for C double.
Boxing a value of double.
Need to allocate an object to generate a new Float.
15Monday, October 10, 11
Do you know C double?
Floating point number with double precision.
No concrete representation is specified.
Most current platforms employ IEEE754.
It is IEEE754 binary64 on these platforms.
There are platforms employing other spec.
16Monday, October 10, 11
CRuby and JIS Ruby
Not requiring IEEE754.
17Monday, October 10, 11
Floating point numbers
18Monday, October 10, 11
The origin
NA = +6.022 141 79⇥ 10
23(±0.000 000 0030⇥ 10
23) [1/mol]
h = +6.626 069 57⇥ 10
�34(±0.000 000 0029⇥ 10
�34) [J s]
19Monday, October 10, 11
The origin
NA = +6.022 141 79⇥ 10
23(±0.000 000 0030⇥ 10
23) [1/mol]
h = +6.626 069 57⇥ 10
�34(±0.000 000 0029⇥ 10
�34) [J s]
sign
19Monday, October 10, 11
The origin
NA = +6.022 141 79⇥ 10
23(±0.000 000 0030⇥ 10
23) [1/mol]
h = +6.626 069 57⇥ 10
�34(±0.000 000 0029⇥ 10
�34) [J s]
fraction part
sign
19Monday, October 10, 11
The origin
NA = +6.022 141 79⇥ 10
23(±0.000 000 0030⇥ 10
23) [1/mol]
h = +6.626 069 57⇥ 10
�34(±0.000 000 0029⇥ 10
�34) [J s]
exponent partfraction part
sign
19Monday, October 10, 11
The origin
NA = +6.022 141 79⇥ 10
23(±0.000 000 0030⇥ 10
23) [1/mol]
h = +6.626 069 57⇥ 10
�34(±0.000 000 0029⇥ 10
�34) [J s]
exponent part:fraction part:
sign: s 2 {0, 1}
0 f Bn � 1
emin
e� q emax
20Monday, October 10, 11
Floating point numbersNumbers can be identified by (s, e, f ).
Represent approximation of real numbers.
Float types can be described by B, N, q, emin, and emax.
B is the base number of the exponent part.
N is the number of digits in the fraction part.
q is the bias for the exponent part.
emax and emin specify the limit of the exponent part.
21Monday, October 10, 11
(s, e, f) = (�1)s ⇥ f
BN⇥Be�q
22Monday, October 10, 11
e.g. IEEE754 binary64
B = 2
N = 53
q = 1,023
emin = –1,022
emax = +1,023
The maximum positive:1.797 693 134 862 315 7 ×10+308
The minimum nonzero positive:2.225 073 858 507 201 4 ×10–308
23Monday, October 10, 11
(s, e, f) = (�1)s ⇥ f
BN⇥Be�q
24Monday, October 10, 11
e.g. IEEE754 decimal64
B = 10
N = 16
q = 398
emin = –383
emax = +384
The maximum positive:9.999 999 999 999 999 ×10+384
The minimum nonzero positive:0.000 000 000 000 001 ×10–383
25Monday, October 10, 11
e.g. IBM’s double precision
B = 16
N = 56
q = 64
emin = –64
emax = +63
The maximum positive:7.237 005 577 332 262 11 ×10+75
The minimum nonzero positive:5.397 605 346 934 027 89 ×10–79
26Monday, October 10, 11
Floating point numbersNumbers can be identified by (s, e, f ).
Represent approximation of real numbers.
Float types can be described by B, N, q, emin, and emax.
B is the base number of the exponent part.
N is the number of digits in the fraction part.
q is the bias for the exponent part.
emax and emin specify the limit of the exponent part.
27Monday, October 10, 11
Every float is approximation
28Monday, October 10, 11
Every float is approximation
0 3/2–1
28Monday, October 10, 11
Every float is approximation
0 3/2–1
0.0 1.5–1.0
{ { {28Monday, October 10, 11
Every float is approximation
0 3/2–1
0.0 1.5–1.0
{ { {28Monday, October 10, 11
Every float is approximation
0 3/2–1
0.0 1.5–1.0
{ { {28Monday, October 10, 11
Every float is approximation
We should think:
There are no numbers represented exactly.
Floating point numbers always include errors.
Magnitude of errors depend on B, N, and e.
29Monday, October 10, 11
Why including errors?
Unavoidable issue from place-value notationwith finite digits rounding.
Very few values can be specified exactly.
We shouldn’t expect that a given value is exact.
30Monday, October 10, 11
How many decimal fractions can be exactly represented in the form of binary fraction?
31Monday, October 10, 11
32Monday, October 10, 11
Decimal form:
(0.1234)10 =(1234)10
104
32Monday, October 10, 11
Decimal form:
Binary form:
(0.10111)2 =(10111)2
25
(0.1234)10 =(1234)10
104
32Monday, October 10, 11
Decimal form:
Binary form:
(0.10111)2 =(10111)2
25
(0.1234)10 =(1234)10
104
0.b1b2 · · · bn =(b1b2 · · · bn)2
2n
0.d1d2 · · · dm =(d1d2 · · · dm)10
10m
32Monday, October 10, 11
Decimal form:
Binary form:
(0.10111)2 =(10111)2
25
(0.1234)10 =(1234)10
104
0.b1b2 · · · bn =(b1b2 · · · bn)2
2n
0.d1d2 · · · dm =(d1d2 · · · dm)10
10m
32Monday, October 10, 11
(d1d2 · · · dm)1010m
=(d1d2 · · · dm)10
2m 5m=
C 5m
2m 5m=
C
2m
33Monday, October 10, 11
1.0
0.5
0.05 10 15 20 25 300
The ratio of inexact numbers
The ratio of exact numbers
The number of decimal digits
34Monday, October 10, 11
1.0
0.5
0.05 10 15 20 25 300
The ratio of inexact numbers
The ratio of exact numbers
17
The number of decimal digits
34Monday, October 10, 11
1.0
0.5
0.05 10 15 20 25 300
The ratio of inexact numbers
The ratio of exact numbersIEEE
754 bina
ry64
17
The number of decimal digits
34Monday, October 10, 11
Decimal in Binary
A N-digit decimal notation is exactly represented in binary notation only if its numerator divisible by 5N.
The ratio of N-digit decimal fractions exactly represented as binary fraction is 1 / 5N.
In IEEE754 binary64, almost all numbers are inexact.
35Monday, October 10, 11
Floating-point arithmetics
add, sub, mul, div, sqrt, ...
These operations work with errors.
Please read detail description:
“What Every Computer Scientist Should Know About Floating-Point Arithmetic”
36Monday, October 10, 11
Decimal fraction of Ruby
37Monday, October 10, 11
What’s the problem?
Ruby interprets literals of decimal fraction as Float
The following three numbers are Float, so they have errors.
1.0
1.2
0.42e+12
38Monday, October 10, 11
The issues from Float
There are many issues about Float reported to redmine.ruby-lang.org
They are caused by that Ruby interpretes the literals of decimal fraction as Float, I think.
Do you know these issues?
39Monday, October 10, 11
http://redmine.ruby-lang.org/issues/457640Monday, October 10, 11
Demonstration
41Monday, October 10, 11
$ ruby -vruby 1.9.4dev (2011-09-28 trunk 33354) [x86_64-darwin10.8.0]$ irb --simple-prompt>> (1.0 .. 12.7).step(1.3).to_a=> [1.0, 2.3, 3.6, 4.9, 6.2, 7.5, 8.8, 10.1, 11.4, 12.700000000000001]>> (1.0 ... 128.4).step(18.2).to_a=> [1.0, 19.2, 37.4, 55.599999999999994, 73.8, 92.0, 110.19999999999999, 128.39999999999998]>> (1.0 ... 128.4).step(18.2).to_a.size=> 8>> (1 ... 1284.quo(10)).step(182.quo(10)).to_a=> [1, (96/5), (187/5), (278/5), (369/5), (92/1), (551/5)]>> (1 ... 1284.quo(10)).step(182.quo(10)).to_a.size=> 7
42Monday, October 10, 11
$ ruby -vruby 1.9.4dev (2011-09-28 trunk 33354) [x86_64-darwin10.8.0]$ irb --simple-prompt>> (1.0 .. 12.7).step(1.3).to_a=> [1.0, 2.3, 3.6, 4.9, 6.2, 7.5, 8.8, 10.1, 11.4, 12.700000000000001]>> (1.0 ... 128.4).step(18.2).to_a=> [1.0, 19.2, 37.4, 55.599999999999994, 73.8, 92.0, 110.19999999999999, 128.39999999999998]>> (1.0 ... 128.4).step(18.2).to_a.size=> 8>> (1 ... 1284.quo(10)).step(182.quo(10)).to_a=> [1, (96/5), (187/5), (278/5), (369/5), (92/1), (551/5)]>> (1 ... 1284.quo(10)).step(182.quo(10)).to_a.size=> 7
The last value of the array should be equal to the end of the range
43Monday, October 10, 11
$ ruby -vruby 1.9.4dev (2011-09-28 trunk 33354) [x86_64-darwin10.8.0]$ irb --simple-prompt>> (1.0 .. 12.7).step(1.3).to_a=> [1.0, 2.3, 3.6, 4.9, 6.2, 7.5, 8.8, 10.1, 11.4, 12.700000000000001]>> (1.0 ... 128.4).step(18.2).to_a=> [1.0, 19.2, 37.4, 55.599999999999994, 73.8, 92.0, 110.19999999999999, 128.39999999999998]>> (1.0 ... 128.4).step(18.2).to_a.size=> 8>> (1 ... 1284.quo(10)).step(182.quo(10)).to_a=> [1, (96/5), (187/5), (278/5), (369/5), (92/1), (551/5)]>> (1 ... 1284.quo(10)).step(182.quo(10)).to_a.size=> 7
Some elements include errors
44Monday, October 10, 11
$ ruby -vruby 1.9.4dev (2011-09-28 trunk 33354) [x86_64-darwin10.8.0]$ irb --simple-prompt>> (1.0 .. 12.7).step(1.3).to_a=> [1.0, 2.3, 3.6, 4.9, 6.2, 7.5, 8.8, 10.1, 11.4, 12.700000000000001]>> (1.0 ... 128.4).step(18.2).to_a=> [1.0, 19.2, 37.4, 55.599999999999994, 73.8, 92.0, 110.19999999999999, 128.39999999999998]>> (1.0 ... 128.4).step(18.2).to_a.size=> 8>> (1 ... 1284.quo(10)).step(182.quo(10)).to_a=> [1, (96/5), (187/5), (278/5), (369/5), (92/1), (551/5)]>> (1 ... 1284.quo(10)).step(182.quo(10)).to_a.size=> 7
The array size is one larger than the correct size
45Monday, October 10, 11
Range#step with Float
The first case
The last value of the array is not equal to the end of the range.
The second case
Some elements include errors.
The array size is one larger than the right size.
46Monday, October 10, 11
Rational with decimal notation
Introducing one flag into a Rational object.
The flag represents a Rational seems which fraction or decimal.
If the flag is true, a Rational is converted decimal string by to_s.
47Monday, October 10, 11
Literal for Rational with decimal notation
Simple change for parser.
Interpreting literal of decimal fraction without exponent as Rational with decimal notation.
Literal of decimal fraction with exponent stays on Float.
48Monday, October 10, 11
Demonstrationusing the patched Rubyhttps://github.com/mrkn/ruby/tree/decimal_rational_implementation
49Monday, October 10, 11
$ ruby -vruby 1.9.4dev (2011-09-28 trunk 33354) [x86_64-darwin10.8.0]$ irb --simple-prompt>> (1.0 .. 12.7).step(1.3).to_a=> [1.0, 2.3, 3.6, 4.9, 6.2, 7.5, 8.8, 10.1, 11.4, 12.7]>> (1.0 .. 12.7).step(1.3).map(&:class)=> [Rational, Rational, Rational, Rational, Rational, Rational, Rational, Rational, Rational, Rational]>> (1.0 ... 128.4).step(18.2).to_a=> [1.0, 19.2, 37.4, 55.6, 73.8, 92.0, 110.2]>> (1.0 ... 128.4).step(18.2).to_a.size=> 7>> (1 ... 1284.quo(10)).step(182.quo(10)).to_a=> [1, (96/5), (187/5), (278/5), (369/5), (92/1), (551/5)]>> (1 ... 1284.quo(10)).step(182.quo(10)).to_a.size=> 7
50Monday, October 10, 11
$ ruby -vruby 1.9.4dev (2011-09-28 trunk 33354) [x86_64-darwin10.8.0]$ irb --simple-prompt>> (1.0 .. 12.7).step(1.3).to_a=> [1.0, 2.3, 3.6, 4.9, 6.2, 7.5, 8.8, 10.1, 11.4, 12.7]>> (1.0 .. 12.7).step(1.3).map(&:class)=> [Rational, Rational, Rational, Rational, Rational, Rational, Rational, Rational, Rational, Rational]>> (1.0 ... 128.4).step(18.2).to_a=> [1.0, 19.2, 37.4, 55.6, 73.8, 92.0, 110.2]>> (1.0 ... 128.4).step(18.2).to_a.size=> 7>> (1 ... 1284.quo(10)).step(182.quo(10)).to_a=> [1, (96/5), (187/5), (278/5), (369/5), (92/1), (551/5)]>> (1 ... 1284.quo(10)).step(182.quo(10)).to_a.size=> 7
The last value of the array is equal to the end of the range.
51Monday, October 10, 11
$ ruby -vruby 1.9.4dev (2011-09-28 trunk 33354) [x86_64-darwin10.8.0]$ irb --simple-prompt>> (1.0 .. 12.7).step(1.3).to_a=> [1.0, 2.3, 3.6, 4.9, 6.2, 7.5, 8.8, 10.1, 11.4, 12.7]>> (1.0 .. 12.7).step(1.3).map(&:class)=> [Rational, Rational, Rational, Rational, Rational, Rational, Rational, Rational, Rational, Rational]>> (1.0 ... 128.4).step(18.2).to_a=> [1.0, 19.2, 37.4, 55.6, 73.8, 92.0, 110.2]>> (1.0 ... 128.4).step(18.2).to_a.size=> 7>> (1 ... 1284.quo(10)).step(182.quo(10)).to_a=> [1, (96/5), (187/5), (278/5), (369/5), (92/1), (551/5)]>> (1 ... 1284.quo(10)).step(182.quo(10)).to_a.size=> 7
All elements in the array is Rational rather than Float.
52Monday, October 10, 11
$ ruby -vruby 1.9.4dev (2011-09-28 trunk 33354) [x86_64-darwin10.8.0]$ irb --simple-prompt>> (1.0 .. 12.7).step(1.3).to_a=> [1.0, 2.3, 3.6, 4.9, 6.2, 7.5, 8.8, 10.1, 11.4, 12.7]>> (1.0 .. 12.7).step(1.3).map(&:class)=> [Rational, Rational, Rational, Rational, Rational, Rational, Rational, Rational, Rational, Rational]>> (1.0 ... 128.4).step(18.2).to_a=> [1.0, 19.2, 37.4, 55.6, 73.8, 92.0, 110.2]>> (1.0 ... 128.4).step(18.2).to_a.size=> 7>> (1 ... 1284.quo(10)).step(182.quo(10)).to_a=> [1, (96/5), (187/5), (278/5), (369/5), (92/1), (551/5)]>> (1 ... 1284.quo(10)).step(182.quo(10)).to_a.size=> 7
The result array size is correct.
53Monday, October 10, 11
Benchmarking
Comparing Float, Rational, and C double.
Experimental environment:
MacBook Pro 15in (Mid 2010)
Core i7 2.66 GHz
Ruby 1.9.4dev (r33300) with gcc-4.2 -O3
C with llvm-gcc -O0
54Monday, October 10, 11
Benchmarking codes
Ruby code
https://gist.github.com/1253088
C code
https://gist.github.com/1253090
55Monday, October 10, 11
0 [s]
0.75 [s]
1.5 [s]
2.25 [s]
3 [s]
1M additions 1M subtractions 1M multiplications
Based on ruby-1.9.4dev (r33300)
Float Rational C double
0.37
2.16
0.73
2.17
0.70
1.78
0.00777 0.00670 0.00770
56Monday, October 10, 11
0 [s]
0.003 [s]
0.005 [s]
0.008 [s]
0.01 [s]
1M additions 1M subtractions 1M multiplications
Based on ruby-1.9.4dev (r33300)
Float Rational C double
0.37 2.16 0.73 2.17 0.70 1.78
0.00777
0.00670
0.00770
57Monday, October 10, 11
Benchmarking summary
Rational is 2-5 times slower than Float.
Float is 2-digit order slower than C double.
C is amazingly fast.
58Monday, October 10, 11
If you said Rational is slow,Float isn’t as fast as your expect.
59Monday, October 10, 11
Rational vs Float
60Monday, October 10, 11
Rational vs Float
61Monday, October 10, 11
Rational vs Float
Exact computation is required by domains such as finance.
Float is required by scientific computation.
61Monday, October 10, 11
Rational vs Float
Exact computation is required by domains such as finance.
Float is required by scientific computation.
Other aspects indepenend of whether Rational or Float.
61Monday, October 10, 11
Conclusion
Float is difficult, troublesome, and not human oriented.
Rational is easy to understand, and human oriented.
It makes us more happy that Ruby interprets literal of decimal fraction as Rational.
62Monday, October 10, 11
Float is Legacy
63Monday, October 10, 11