Class: WholeHistoryRating::Player
- Inherits:
-
Object
- Object
- WholeHistoryRating::Player
- Defined in:
- lib/whole_history_rating/player.rb
Instance Attribute Summary collapse
-
#anchor_gamma ⇒ Object
Returns the value of attribute anchor_gamma.
-
#days ⇒ Object
Returns the value of attribute days.
-
#debug ⇒ Object
Returns the value of attribute debug.
-
#id ⇒ Object
Returns the value of attribute id.
-
#name ⇒ Object
Returns the value of attribute name.
-
#w2 ⇒ Object
Returns the value of attribute w2.
Instance Method Summary collapse
- #add_game(game) ⇒ Object
- #compute_sigma2 ⇒ Object
- #covariance ⇒ Object
- #gradient(r, days, sigma2) ⇒ Object
- #hessian(days, sigma2) ⇒ Object
-
#initialize(name, config) ⇒ Player
constructor
A new instance of Player.
- #inspect ⇒ Object
- #log_likelihood ⇒ Object
- #run_one_newton_iteration ⇒ Object
- #update_by_ndim_newton ⇒ Object
- #update_uncertainty ⇒ Object
Constructor Details
#initialize(name, config) ⇒ Player
Returns a new instance of Player.
7 8 9 10 11 12 |
# File 'lib/whole_history_rating/player.rb', line 7 def initialize(name, config) @name = name @debug = config[:debug] @w2 = (Math.sqrt(config[:w2])*Math.log(10)/400)**2 # Convert from elo^2 to r^2 @days = [] end |
Instance Attribute Details
#anchor_gamma ⇒ Object
Returns the value of attribute anchor_gamma.
5 6 7 |
# File 'lib/whole_history_rating/player.rb', line 5 def anchor_gamma @anchor_gamma end |
#days ⇒ Object
Returns the value of attribute days.
5 6 7 |
# File 'lib/whole_history_rating/player.rb', line 5 def days @days end |
#debug ⇒ Object
Returns the value of attribute debug.
5 6 7 |
# File 'lib/whole_history_rating/player.rb', line 5 def debug @debug end |
#id ⇒ Object
Returns the value of attribute id.
5 6 7 |
# File 'lib/whole_history_rating/player.rb', line 5 def id @id end |
#name ⇒ Object
Returns the value of attribute name.
5 6 7 |
# File 'lib/whole_history_rating/player.rb', line 5 def name @name end |
#w2 ⇒ Object
Returns the value of attribute w2.
5 6 7 |
# File 'lib/whole_history_rating/player.rb', line 5 def w2 @w2 end |
Instance Method Details
#add_game(game) ⇒ Object
234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 |
# File 'lib/whole_history_rating/player.rb', line 234 def add_game(game) if days.last.nil? || days.last.day != game.day new_pday = PlayerDay.new(self, game.day) if days.empty? new_pday.is_first_day = true new_pday.gamma = 1 else new_pday.gamma = days.last.gamma end days << new_pday end if (game.white_player == self) game.wpd = days.last else game.bpd = days.last end days.last.add_game(game) end |
#compute_sigma2 ⇒ Object
90 91 92 93 94 95 96 |
# File 'lib/whole_history_rating/player.rb', line 90 def compute_sigma2 sigma2 = [] days.each_cons(2) do |d1,d2| sigma2 << (d2.day - d1.day).abs * @w2 end sigma2 end |
#covariance ⇒ Object
169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 |
# File 'lib/whole_history_rating/player.rb', line 169 def covariance r = days.map(&:r) sigma2 = compute_sigma2 h = hessian(days, sigma2) g = gradient(r, days, sigma2) n = days.count a = [] d = [h[0,0]] b = [h[0,1]] n = r.size (1..(n-1)).each do |i| a[i] = h[i,i-1] / d[i-1] d[i] = h[i,i] - a[i] * b[i-1] b[i] = h[i,i+1] end dp = [] dp[n-1] = h[n-1,n-1] bp = [] bp[n-1] = h[n-1,n-2] ap = [] (n-2).downto(0) do |i| ap[i] = h[i,i+1] / dp[i+1] dp[i] = h[i,i] - ap[i]*bp[i+1] bp[i] = h[i,i-1] end v = [] 0.upto(n-2) do |i| v[i] = dp[i+1]/(b[i]*bp[i+1] - d[i]*dp[i+1]) end v[n-1] = -1/d[n-1] #puts "a = #{a}" #puts "b = #{b}" #puts "bp = #{bp}" #puts "d = #{d}" #puts "dp = #{dp}" #puts "v = #{v}" Matrix.build(n) do |row,col| if row == col v[row] elsif row == col-1 -1*a[col]*v[col] else 0 end end end |
#gradient(r, days, sigma2) ⇒ Object
63 64 65 66 67 68 69 70 71 72 73 74 75 76 |
# File 'lib/whole_history_rating/player.rb', line 63 def gradient(r, days, sigma2) g = [] n = days.count days.each_with_index do |day,idx| prior = 0 prior += -(r[idx]-r[idx+1])/sigma2[idx] if idx < (n-1) prior += -(r[idx]-r[idx-1])/sigma2[idx-1] if idx > 0 if @debug puts "g[#{idx}] = #{day.log_likelihood_derivative} + #{prior}" end g << day.log_likelihood_derivative + prior end g end |
#hessian(days, sigma2) ⇒ Object
45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 |
# File 'lib/whole_history_rating/player.rb', line 45 def hessian(days, sigma2) n = days.count Matrix.build(n) do |row,col| if row == col prior = 0 prior += -1.0/sigma2[row] if row < (n-1) prior += -1.0/sigma2[row-1] if row > 0 days[row].log_likelihood_second_derivative + prior - 0.001 elsif row == col-1 1.0/sigma2[row] elsif row == col+1 1.0/sigma2[col] else 0 end end end |
#inspect ⇒ Object
14 15 16 |
# File 'lib/whole_history_rating/player.rb', line 14 def inspect "#{self}:(#{name})" end |
#log_likelihood ⇒ Object
18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 |
# File 'lib/whole_history_rating/player.rb', line 18 def log_likelihood sum = 0.0 sigma2 = compute_sigma2 n = days.count 0.upto(n-1) do |i| prior = 0 if i < (n-1) rd = days[i].r - days[i+1].r prior += (1/(Math.sqrt(2*Math::PI*sigma2[i]))) * Math.exp(-(rd**2)/2*sigma2[i]) end if i > 0 rd = days[i].r - days[i-1].r prior += (1/(Math.sqrt(2*Math::PI*sigma2[i-1]))) * Math.exp(-(rd**2)/2*sigma2[i-1]) end if prior == 0 sum += days[i].log_likelihood else if (days[i].log_likelihood.infinite? || Math.log(prior).infinite?) puts "Infinity at #{inspect}: #{days[i].log_likelihood} + #{Math.log(prior)}: prior = #{prior}, days = #{days.inspect}" exit end sum += days[i].log_likelihood + Math.log(prior) end end sum end |
#run_one_newton_iteration ⇒ Object
78 79 80 81 82 83 84 85 86 87 88 |
# File 'lib/whole_history_rating/player.rb', line 78 def run_one_newton_iteration days.each do |day| day.clear_game_terms_cache end if days.count == 1 days[0].update_by_1d_newtons_method elsif days.count > 1 update_by_ndim_newton end end |
#update_by_ndim_newton ⇒ Object
98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 |
# File 'lib/whole_history_rating/player.rb', line 98 def update_by_ndim_newton # r r = days.map(&:r) if @debug puts "Updating #{inspect}" days.each do |day| puts "day[#{day.day}] r = #{day.r}" puts "day[#{day.day}] win terms = #{day.won_game_terms}" puts "day[#{day.day}] win games = #{day.won_games}" puts "day[#{day.day}] lose terms = #{day.lost_game_terms}" puts "day[#{day.day}] lost games = #{day.lost_games}" puts "day[#{day.day}] log(p) = #{day.log_likelihood}" puts "day[#{day.day}] dlp = #{day.log_likelihood_derivative}" puts "day[#{day.day}] dlp2 = #{day.log_likelihood_second_derivative}" end end # sigma squared (used in the prior) sigma2 = compute_sigma2 h = hessian(days, sigma2) g = gradient(r, days, sigma2) a = [] d = [h[0,0]] b = [h[0,1]] n = r.size (1..(n-1)).each do |i| a[i] = h[i,i-1] / d[i-1] d[i] = h[i,i] - a[i] * b[i-1] b[i] = h[i,i+1] end y = [g[0]] (1..(n-1)).each do |i| y[i] = g[i] - a[i] * y[i-1] end x = [] x[n-1] = y[n-1] / d[n-1] (n-2).downto(0) do |i| x[i] = (y[i] - b[i] * x[i+1]) / d[i] end new_r = r.zip(x).map {|ri,xi| ri-xi} new_r.each do |r| if r > 650 raise WHR::UnstableRatingException, "Unstable r (#{new_r}) on player #{inspect}" end end if @debug puts "Hessian = #{h}" puts "gradient = #{g}" puts "a = #{a}" puts "d = #{d}" puts "b = #{b}" puts "y = #{y}" puts "x = #{x}" puts "#{inspect} (#{r}) => (#{new_r})" end days.each_with_index do |day,idx| day.r = day.r - x[idx] end end |
#update_uncertainty ⇒ Object
224 225 226 227 228 229 230 231 232 |
# File 'lib/whole_history_rating/player.rb', line 224 def update_uncertainty if days.count > 0 c = covariance u = (0..(days.count-1)).collect{|i| c[i,i]} days.zip(u) {|d,u| d.uncertainty = u} else 5 end end |