add option 'random_length'

automata.py

@@ -117,7 +117,12 @@ class AutomatonSampler:
             data_config['length'] = 20
         self.T = self.data_config['length']
 
-        self.__info__ = "  - T (int): sequence length"
+        if 'random_length' not in data_config:
+            data_config['random_length'] = 0
+        self.random_length = data_config['random_length']
+
+        self.__info__ = "  - T (int): sequence length.\n" \
+            + "  - random_length (int in {0, 1}): whether to randomly sample a length per sample.\n"
 
     def f(self, x):
         """
@@ -128,6 +133,11 @@ class AutomatonSampler:
     def sample(self):
         raise NotImplementedError()
 
+    def sample_length(self):
+        if self.random_length:
+          return self.np_rng.choice(range(1, self.T+1))
+        return self.T
+
     def help(self):
         print(self.__info__)
 
@@ -146,7 +156,8 @@ class BinaryInputSampler(AutomatonSampler):
         raise NotImplementedError()
 
     def sample(self):
-        x = self.np_rng.binomial(1, self.prob1, size=self.T)
+        T = self.sample_length()
+        x = self.np_rng.binomial(1, self.prob1, size=T)
         return x, self.f(x)
 
 class ParitySampler(BinaryInputSampler):
@@ -258,8 +269,9 @@ class ABABSampler(BinaryInputSampler):
   def sample(self):
     pos_sample = np.random.random() < self.prob_abab_pos_sample
     if pos_sample:
-      x = [0,1,0,1] * (self.T//4)
-      x += [0,1,0,1][:(self.T%4)]
+      T = self.sample_length()
+      x = [0,1,0,1] * (T//4)
+      x += [0,1,0,1][:(T%4)]
       x = np.array(x)
       return x, self.f(x)
     else:
@@ -295,9 +307,10 @@ class FlipFlopSampler(AutomatonSampler):
     return np.array(states)
 
   def sample(self):
-    rand = np.random.uniform(size=self.T)
+    T = self.sample_length()
+    rand = np.random.uniform(size=T)
     nonzero_pos = (rand < 0.5).astype(np.int64)
-    writes = np.random.choice(range(1, self.n_states+1), size=self.T)
+    writes = np.random.choice(range(1, self.n_states+1), size=T)
     x = writes * nonzero_pos
     return x, self.f(x)
 
@@ -389,7 +402,8 @@ class SymmetricSampler(AutomatonSampler):
     return np.array(labels)
 
   def sample(self):
-    x = np.random.choice(range(self.n_actions), replace=True, size=self.T)
+    T = self.sample_length()
+    x = np.random.choice(range(self.n_actions), replace=True, size=T)
 
     return x, self.f(x)