use fitness shaping

main.lua

@@ -40,9 +40,9 @@ local eps_frames = 1000000
 local consider_past_rewards = false
 local learn_start_select = false
 --
-local epoch_trials = 40 -- 24
+local epoch_trials = 40
-local learning_rate = 1e-3
+local learning_rate = 0.3 -- bigger now that i'm shaping trials etc.
-local deviation = 1e-2 -- 4e-3
+local deviation = 0.05
 --
 local cap_time = 400
 local timer_loser = 1/3
@@ -127,6 +127,7 @@ local randomseed = math.randomseed
 local insert = table.insert
 local remove = table.remove
 local unpack = table.unpack or unpack
+local sort = table.sort
 local R = memory.readbyteunsigned
 local S = memory.readbyte --signed
 local W = memory.writebyte
@@ -143,12 +144,22 @@ local ror = bit.ror
 
 -- utilities.
 
+local function ifind(haystack, needle)
+    for i, v in ipairs(haystack) do
+        if v == needle then return i end
+    end
+    return nil
+end
+
 local function boolean_xor(a, b)
     if a and b then return false end
     if not a and not b then return false end
     return true
 end
 
+local _invlog2 = 1 / log(2)
+local function log2(x) return log(x) * _invlog2 end
+
 local function clamp(x, l, u) return min(max(x, l), u) end
 
 local function lerp(a, b, t) return a + (b - a) * clamp(t, 0, 1) end
@@ -517,6 +528,8 @@ local function prepare_epoch()
     base_params = network:collect()
     empty(trial_noise)
     empty(trial_rewards)
+    -- TODO: save memory. generate noise as needed by saving the seed
+    --       (the os.time() as of here) and calling nn.normal() each trial.
     for i = 1, epoch_trials do
         local noise = nn.zeros(#base_params)
         for j = 1, #base_params do noise[j] = nn.normal() end
@@ -537,6 +550,30 @@ local function load_next_trial()
     network:distribute(W)
 end
 
+local function fitness_shaping(rewards)
+    -- lifted from: https://github.com/atgambardella/pytorch-es/blob/master/train.py
+    local decreasing = nn.copy(rewards)
+    sort(decreasing, function(a, b) return a > b end)
+    local shaped_returns = {}
+    local lamb = #rewards
+
+    local denom = 0
+    for i, v in ipairs(rewards) do
+        local l = log2(lamb / 2 + 1)
+        local r = log2(ifind(decreasing, v))
+        denom = denom + max(0, l - r)
+    end
+
+    for i, v in ipairs(rewards) do
+        local l = log2(lamb / 2 + 1)
+        local r = log2(ifind(decreasing, v))
+        local numer = max(0, l - r)
+        insert(shaped_returns, numer / denom + 1 / lamb)
+    end
+
+    return shaped_returns
+end
+
 local function learn_from_epoch()
     print()
     print('rewards:', trial_rewards)
@@ -552,37 +589,19 @@ local function learn_from_epoch()
     end
     --print('normalized:', trial_rewards)
 
-    local reward_mean, reward_dev = calc_mean_dev(trial_rewards)
-
     local step = nn.zeros(#base_params)
+    local shaped_rewards = fitness_shaping(trial_rewards)
+
+    local altogether = learning_rate / (epoch_trials * deviation)
     for i = 1, epoch_trials do
-        local reward = trial_rewards[i]
+        local reward = shaped_rewards[i]
         local noise = trial_noise[i]
         for j, v in ipairs(noise) do
-            step[j] = step[j] + reward * v
+            step[j] = step[j] + altogether * (reward * v)
         end
     end
 
-    local magnitude = learning_rate / deviation
-    --print('stepping with magnitude', magnitude)
-    -- throw the division from the averaging in there too.
-    local altogether = magnitude / epoch_trials
-    for i, v in ipairs(step) do
-        step[i] = altogether * v
-    end
-
     local step_mean, step_dev = calc_mean_dev(step)
-    if step_dev < 1e-8 then
-        -- we didn't get anywhere. step in a random direction.
-        print("stepping randomly.")
-        local noise = trial_noise[1]
-        local devsqrt = sqrt(deviation)
-        for i, v in ipairs(step) do
-            step[i] = devsqrt * noise[i]
-        end
-
-        step_mean, step_dev = calc_mean_dev(step)
-    end
     if abs(step_mean) > 1e-3 then print("step mean:", step_mean) end
     print("step stddev:", step_dev)