pick_me_too 1.1.1 → 1.1.2
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.gitignore +1 -1
- data/CHANGES.md +2 -0
- data/README.md +2 -2
- data/lib/pick_me_too.rb +60 -36
- data/test/basic_test.rb +24 -0
- data/test/documentation_test.rb +2 -2
- metadata +2 -2
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA256:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: 070a1c66bfe69fe534fc449c39a0f187a7b46e3bbcf3bc9827953993cfa30e28
|
4
|
+
data.tar.gz: 41ea453f24ccb5a1602aca774f488295e68280ece7b44c0d12f060295c645d58
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: 7f15740e07eaf7c91548df5cae303e598a13c0e03c9f2af514b6368a497ff60741dfb6b957ee9b4b72cc222575be7ece4faaa5582753e1486fd5f84d2256c734
|
7
|
+
data.tar.gz: 751c64c9aece6e27fd53ecd52af26eec09883542b78077259487804e056594a8f5ab95e4c614fbc8b6dfa206d91e2d4271bec570153487f1da8d4968f0c53030
|
data/.gitignore
CHANGED
data/CHANGES.md
CHANGED
@@ -1,5 +1,7 @@
|
|
1
1
|
# Change Log
|
2
2
|
|
3
|
+
## 1.1.2 *2022-11-6*
|
4
|
+
* optimized ternerization code still further
|
3
5
|
## 1.1.1 *2022-10-30*
|
4
6
|
* simplified, and perhaps improved, the code to compile the frequency list into a nested ternary expression; ***NOTE***: I am not considering this a breaking change, though it may change the sequence of items picked by a fixed random number sequence
|
5
7
|
## 1.1.0 *2022-8-21*
|
data/README.md
CHANGED
@@ -18,7 +18,7 @@ picker = PickMeToo.new([["prevention", 1], ["cure", 16]], -> { rng.rand })
|
|
18
18
|
counter = Hash.new 0
|
19
19
|
32.times { counter[picker.pick] += 1 }
|
20
20
|
counter
|
21
|
-
# => {"cure"=>
|
21
|
+
# => {"cure"=>31, "prevention"=>1}
|
22
22
|
|
23
23
|
# you can also use a hash to map items to frequencies
|
24
24
|
# frequencies don't need to be whole numbers
|
@@ -29,7 +29,7 @@ picker = PickMeToo.new({foo: 1, bar: 2, baz: 0.5}, -> { rng.rand })
|
|
29
29
|
counter = Hash.new 0
|
30
30
|
32.times { counter[picker.pick] += 1 }
|
31
31
|
counter
|
32
|
-
# => {:bar=>
|
32
|
+
# => {:bar=>22, :foo=>5, :baz=>5}
|
33
33
|
|
34
34
|
# you don't need to provide your own random number sequence
|
35
35
|
picker = PickMeToo({a: 1, b: 2, c: 3})
|
data/lib/pick_me_too.rb
CHANGED
@@ -10,14 +10,13 @@
|
|
10
10
|
# # => [:goblin, :orc, :bugbear, :orc, :goblin, :bugbear, :goblin, :goblin, :orc, :goblin]
|
11
11
|
#
|
12
12
|
# irrational = PickMeToo.new({e: Math::E, pi: Math::PI})
|
13
|
-
#
|
13
|
+
# 10.times.map { irrational.pick }
|
14
14
|
# # => [:e, :e, :e, :pi, :e, :e, :e, :pi, :pi, :e]
|
15
15
|
#
|
16
16
|
# Items once picked are "placed back in the urn", so if you pick a cat this doesn't reduce the
|
17
|
-
# probability the next thing you pick is also a cat, and the urn will never be picked empty.
|
18
|
-
# this is all a metaphor.)
|
17
|
+
# probability the next thing you pick is also a cat, and the urn will never be picked empty.
|
19
18
|
class PickMeToo
|
20
|
-
VERSION = '1.1.
|
19
|
+
VERSION = '1.1.2'
|
21
20
|
|
22
21
|
class Error < StandardError; end
|
23
22
|
|
@@ -42,9 +41,9 @@ class PickMeToo
|
|
42
41
|
if @objects.length == 1
|
43
42
|
@picker = ->(_p) { 0 }
|
44
43
|
else
|
45
|
-
root =
|
44
|
+
root = optimize(frequencies)
|
46
45
|
# compile everything into a nested ternary expression
|
47
|
-
@picker = eval "->(p) { #{ternarize(root
|
46
|
+
@picker = eval "->(p) { #{ternarize(root)} }"
|
48
47
|
end
|
49
48
|
end
|
50
49
|
|
@@ -80,49 +79,74 @@ class PickMeToo
|
|
80
79
|
raise Error, "the following have non-positive frequencies: #{bad.inspect}" if bad.any?
|
81
80
|
|
82
81
|
total = good.map(&:last).sum.to_f
|
83
|
-
# sort by size of probability interval
|
82
|
+
# sort by size of probability interval
|
83
|
+
# in general we will want to consider wide intervals before narrow ones
|
84
84
|
good.sort_by(&:last).reverse.map { |o, n| [o, n / total] }
|
85
85
|
end
|
86
86
|
|
87
|
-
#
|
88
|
-
#
|
89
|
-
def
|
87
|
+
# optimize the order of threshold comparisons to map a random number to an index in the array
|
88
|
+
# of choices
|
89
|
+
def optimize(frequencies)
|
90
90
|
frequencies = frequencies.each_with_index.map { |(*, i), idx| { interval: i, index: idx } }
|
91
|
-
|
92
|
-
|
93
|
-
|
91
|
+
root = build_branch(frequencies)
|
92
|
+
add_thresholds(root, 0)
|
93
|
+
root
|
94
|
+
end
|
94
95
|
|
95
|
-
|
96
|
-
|
97
|
-
|
98
|
-
|
99
|
-
|
96
|
+
def add_thresholds(node, acc)
|
97
|
+
# acc represents the accumulated probability mass known to be before anything in the tree
|
98
|
+
# currently under consideration
|
99
|
+
if (l = node[:left])
|
100
|
+
add_thresholds(l, acc)
|
101
|
+
node[:left_threshold] = acc += l[:sum]
|
102
|
+
end
|
103
|
+
if (r = node[:right])
|
104
|
+
acc = node[:right_threshold] = acc + node[:interval]
|
105
|
+
add_thresholds(r, acc)
|
100
106
|
end
|
101
|
-
frequencies[0]
|
102
107
|
end
|
103
108
|
|
104
|
-
|
105
|
-
|
106
|
-
|
109
|
+
def build_branch(frequencies)
|
110
|
+
sum = frequencies.sum { |o| o[:interval] }
|
111
|
+
node = frequencies.shift
|
112
|
+
if frequencies.any?
|
113
|
+
if node[:interval] * 3 >= sum
|
114
|
+
# a binary search would be wasteful because the frequencies are so skewed
|
115
|
+
node[:right] = build_branch(frequencies)
|
116
|
+
else
|
117
|
+
# build a binary-branching search tree
|
118
|
+
left, right = frequencies.each_with_index.partition { |*, i| left_branch?(i + 1) }
|
119
|
+
node[:left] = build_branch(left.map(&:first))
|
120
|
+
node[:right] = build_branch(right.map(&:first)) if right.any?
|
121
|
+
end
|
122
|
+
end
|
123
|
+
node[:sum] = sum
|
124
|
+
node
|
125
|
+
end
|
107
126
|
|
108
|
-
|
109
|
-
|
110
|
-
|
111
|
-
|
127
|
+
# this implements the heap rule for matching a node to its parent
|
128
|
+
# our binary-branching trees are heaps with wider intervals towards the root
|
129
|
+
def left_branch?(index)
|
130
|
+
if index == 1
|
131
|
+
true
|
132
|
+
elsif index < 1
|
133
|
+
false
|
134
|
+
else
|
135
|
+
left_branch?((index - 1) / 2)
|
112
136
|
end
|
113
137
|
end
|
114
138
|
|
115
139
|
# reduce the probability tree to nested ternary expressions
|
116
|
-
def ternarize(node
|
117
|
-
|
118
|
-
|
119
|
-
|
120
|
-
|
121
|
-
|
122
|
-
|
123
|
-
|
124
|
-
|
140
|
+
def ternarize(node)
|
141
|
+
l, r = node.values_at :left, :right
|
142
|
+
if l && r
|
143
|
+
"(p > #{node[:left_threshold]} ? (p > #{node[:right_threshold]} ? #{ternarize(r)} : #{node[:index]}) : #{ternarize(l)})"
|
144
|
+
elsif l
|
145
|
+
"(p > #{node[:left_threshold]} ? #{node[:index]} : #{ternarize(l)})"
|
146
|
+
elsif r
|
147
|
+
"(p > #{node[:right_threshold]} ? #{ternarize(r)} : #{node[:index]})"
|
148
|
+
else
|
149
|
+
node[:index]
|
125
150
|
end
|
126
|
-
"(p < #{left + acc} ? #{ternarize(node[:left], acc)} : #{right})"
|
127
151
|
end
|
128
152
|
end
|
data/test/basic_test.rb
CHANGED
@@ -71,6 +71,30 @@ class BasicTest < Minitest::Test
|
|
71
71
|
end
|
72
72
|
end
|
73
73
|
|
74
|
+
# force unary branching
|
75
|
+
def test_big
|
76
|
+
rnd = Random.new 1
|
77
|
+
frequencies = %w[a b c d e f g h].each_with_index.map { |k, i| [k, 2**i] }
|
78
|
+
picker = PickMeToo.new(frequencies, -> { rnd.rand })
|
79
|
+
counter = Hash.new(0)
|
80
|
+
(frequencies.map(&:last).sum * 6000).times { counter[picker.pick] += 1 }
|
81
|
+
frequencies.each do |key, n|
|
82
|
+
assert_equal n, (counter[key] / 6000.0).round, "right number of #{key}"
|
83
|
+
end
|
84
|
+
end
|
85
|
+
|
86
|
+
# force binary branching
|
87
|
+
def test_small
|
88
|
+
rnd = Random.new 1
|
89
|
+
frequencies = %w[a b c d e f g h].each_with_index.map { |k, _i| [k, 1] }
|
90
|
+
picker = PickMeToo.new(frequencies, -> { rnd.rand })
|
91
|
+
counter = Hash.new(0)
|
92
|
+
(frequencies.map(&:last).sum * 6000).times { counter[picker.pick] += 1 }
|
93
|
+
frequencies.each do |key, n|
|
94
|
+
assert_equal n, (counter[key] / 6000.0).round, "right number of #{key}"
|
95
|
+
end
|
96
|
+
end
|
97
|
+
|
74
98
|
def test_randomize
|
75
99
|
rnd1 = Random.new 1
|
76
100
|
rnd2 = Random.new 1
|
data/test/documentation_test.rb
CHANGED
@@ -13,7 +13,7 @@ class BasicTest < Minitest::Test
|
|
13
13
|
picker = PickMeToo.new([['prevention', 1], ['cure', 16]], -> { rng.rand })
|
14
14
|
counter = Hash.new 0
|
15
15
|
32.times { counter[picker.pick] += 1 }
|
16
|
-
assert_equal({ 'cure' =>
|
16
|
+
assert_equal({ 'cure' => 31, 'prevention' => 1 }, counter)
|
17
17
|
end
|
18
18
|
|
19
19
|
def test_synopsis_hash
|
@@ -21,6 +21,6 @@ class BasicTest < Minitest::Test
|
|
21
21
|
picker = PickMeToo.new({ foo: 1, bar: 2, baz: 0.5 }, -> { rng.rand })
|
22
22
|
counter = Hash.new 0
|
23
23
|
32.times { counter[picker.pick] += 1 }
|
24
|
-
assert_equal({:
|
24
|
+
assert_equal({ bar: 22, foo: 5, baz: 5 }, counter)
|
25
25
|
end
|
26
26
|
end
|
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: pick_me_too
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 1.1.
|
4
|
+
version: 1.1.2
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- David F. Houghton
|
8
8
|
autorequire:
|
9
9
|
bindir: bin
|
10
10
|
cert_chain: []
|
11
|
-
date: 2022-
|
11
|
+
date: 2022-11-06 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: bundler
|