noregret 0.0.0.dev0__tar.gz → 0.0.0.dev1__tar.gz

{noregret-0.0.0.dev0 → noregret-0.0.0.dev1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: noregret
-Version: 0.0.0.dev0
+Version: 0.0.0.dev1
 Summary: No-regret learning dynamics
 Home-page: https://github.com/uoftcprg/noregret
 Author: Universal, Open, Free, and Transparent Computer Poker Research Group

{noregret-0.0.0.dev0 → noregret-0.0.0.dev1}/noregret/games.py

@@ -1,10 +1,11 @@
 from abc import ABC, abstractmethod
-from dataclasses import dataclass, field
+from dataclasses import dataclass
 from functools import partial
-from itertools import count, permutations
+from itertools import permutations
 from math import factorial
 from typing import Any
 
+from ordered_set import OrderedSet
 from scipy.sparse import lil_array
 import numpy as np
 
@@ -255,13 +256,11 @@ class NormalFormGame(Serializable, Game):
 
     actions: Any
     utilities: Any
-    indices: Any = field(init=False, default_factory=list)
 
     def __post_init__(self):
         super().__post_init__()
 
-        for i, actions in enumerate(self.actions):
-            self.indices.append(dict(zip(actions, count())))
+        self.actions = tuple(map(OrderedSet, self.actions))
 
     def _verify(self, *, utilities_shape=None, **kwargs):
         super()._verify(**kwargs)
@@ -304,14 +303,6 @@ class TwoPlayerNormalFormGame(TwoPlayerGame, NormalFormGame):
     def column_actions(self):
         return self.actions[1]
 
-    @property
-    def row_indices(self):
-        return self.indices[0]
-
-    @property
-    def column_indices(self):
-        return self.indices[1]
-
     @property
     def row_utilities(self):
         return self.utilities[:, :, 0]
@@ -414,7 +405,7 @@ class TwoPlayerExtensiveFormGame(TwoPlayerGame, ExtensiveFormGame):
             for tfsdp, sequence in zip(tfsdps, raw_utility['sequences']):
                 sequence = tuple(sequence)
 
-                indices.append(tfsdp.indices[sequence])
+                indices.append(tfsdp.sequences.index(sequence))
 
             indices = tuple(indices)
             row_utilities[indices] = raw_utility['values'][0]
@@ -448,14 +439,6 @@ class TwoPlayerExtensiveFormGame(TwoPlayerGame, ExtensiveFormGame):
     def column_sequences(self):
         return self.column_tree_form_sequential_decision_process.sequences
 
-    @property
-    def row_indices(self):
-        return self.row_tree_form_sequential_decision_process.indices
-
-    @property
-    def column_indices(self):
-        return self.column_tree_form_sequential_decision_process.indices
-
     @property
     def row_utilities(self):
         return self.utilities[0]
@@ -530,7 +513,7 @@ class TwoPlayerZeroSumExtensiveFormGame(
             for tfsdp, sequence in zip(tfsdps, raw_utility['sequences']):
                 sequence = tuple(sequence)
 
-                indices.append(tfsdp.indices[sequence])
+                indices.append(tfsdp.sequences.index(sequence))
 
             indices = tuple(indices)
             utilities[indices] = raw_utility['value']

{noregret-0.0.0.dev0 → noregret-0.0.0.dev1}/noregret/utilities.py

@@ -153,7 +153,6 @@ class TreeFormSequentialDecisionProcess(Serializable):
     decision_points: Any = field(init=False, default_factory=OrderedSet)
     observation_points: Any = field(init=False, default_factory=OrderedSet)
     sequences: Any = field(init=False, default_factory=OrderedSet)
-    indices: Any = field(init=False, default_factory=dict)
     parent_sequences: Any = field(init=False, default_factory=dict)
     actions: Any = field(
         init=False,
@@ -203,8 +202,6 @@ class TreeFormSequentialDecisionProcess(Serializable):
             if is_sequence:
                 self.sequences.add(parent_edge)
 
-                self.indices[parent_edge] = len(self.indices)
-
             self.parent_sequences[p] = parent_sequence
 
     def behavioral_uniform_strategy(self):
@@ -227,7 +224,7 @@ class TreeFormSequentialDecisionProcess(Serializable):
 
                     for i, a in enumerate(self.actions[p]):
                         value = (
-                            utility[self.indices[p, a]]
+                            utility[self.sequences.index((p, a))]
                             + V[self.transitions[p, a]]
                         )
 
@@ -250,15 +247,15 @@ class TreeFormSequentialDecisionProcess(Serializable):
 
     def behavioral_to_sequence_form(self, behavioral_strategy):
         strategy = np.zeros(len(self.sequences))
-        strategy[self.indices[()]] = 1
+        strategy[self.sequences.index(())] = 1
 
         for j in self.decision_points:
             p_j = self.parent_sequences[j]
 
             for i, a in enumerate(self.actions[j]):
-                strategy[self.indices[j, a]] = (
+                strategy[self.sequences.index((j, a))] = (
                     behavioral_strategy[j][i]
-                    * strategy[self.indices[p_j]]
+                    * strategy[self.sequences.index(p_j)]
                 )
 
         return strategy
@@ -273,7 +270,7 @@ class TreeFormSequentialDecisionProcess(Serializable):
                         V[p] += (
                             behavioral_strategy[p][i]
                             * (
-                                utility[self.indices[p, a]]
+                                utility[self.sequences.index((p, a))]
                                 + V[self.transitions[p, a]]
                             )
                         )
@@ -288,7 +285,7 @@ class TreeFormSequentialDecisionProcess(Serializable):
 
             for i, a in enumerate(self.actions[j]):
                 utilities[j][i] = (
-                    utility[self.indices[j, a]]
+                    utility[self.sequences.index((j, a))]
                     + V[self.transitions[j, a]]
                 )
 

{noregret-0.0.0.dev0 → noregret-0.0.0.dev1}/noregret.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: noregret
-Version: 0.0.0.dev0
+Version: 0.0.0.dev1
 Summary: No-regret learning dynamics
 Home-page: https://github.com/uoftcprg/noregret
 Author: Universal, Open, Free, and Transparent Computer Poker Research Group

{noregret-0.0.0.dev0 → noregret-0.0.0.dev1}/setup.py

@@ -4,7 +4,7 @@ from setuptools import find_packages, setup
 
 setup(
     name='noregret',
-    version='0.0.0.dev0',
+    version='0.0.0.dev1',
     description='No-regret learning dynamics',
     long_description=open('README.md').read(),
     long_description_content_type='text/markdown',