Merge branch 'release/1.3.7'

.github/ISSUE_TEMPLATE/bug_report.md

@@ -24,7 +24,7 @@ If applicable, add screenshots to help explain your problem.
 
 **Desktop (please complete the following information):**
  - OS: [e.g. mac/linux/windows]
- - Version [e.g. 1.3.6]
+ - Version [e.g. 1.3.7]
 
 **Additional context**
 Add any other context about the problem here.

CMakeLists.txt

@@ -1,6 +1,6 @@
 cmake_minimum_required(VERSION 3.10.0)
 set(CMAKE_OSX_DEPLOYMENT_TARGET "10.9" CACHE STRING "Minimum OS X deployment version")
-project(Griddly VERSION 1.3.6)
+project(Griddly VERSION 1.3.7)
 
 set(BINARY ${CMAKE_PROJECT_NAME})
 

bindings/python.cpp

@@ -12,7 +12,7 @@ namespace griddly {
 
 PYBIND11_MODULE(python_griddly, m) {
   m.doc() = "Griddly python bindings";
-  m.attr("version") = "1.3.6";
+  m.attr("version") = "1.3.7";
 
 #ifndef NDEBUG
   spdlog::set_level(spdlog::level::debug);

bindings/wrapper/GameWrapper.cpp

@@ -213,12 +213,21 @@ class Py_GameWrapper {
     bool terminated = false;
     py::dict info{};
 
-    for (int p = 0; p < playerSize; p++) {
+    std::vector<uint32_t> playerIdx;
+
+    for (uint32_t p = 0; p < playerSize; p++) {
+        playerIdx.push_back(p);
+    }
+
+    std::shuffle(playerIdx.begin(), playerIdx.end(),  gameProcess_->getGrid()->getRandomGenerator()->getEngine());
+
+    for (int i = 0; i < playerSize; i++) {
+      auto p = playerIdx[i];
       std::string actionName;
       std::vector<int32_t> actionArray;
       auto pStr = (int32_t*)stepArrayInfo.ptr + p * playerStride;
 
-      bool lastPlayer = p == (playerSize - 1);
+      bool lastPlayer = i == (playerSize - 1);
 
       switch (actionSize) {
         case 1:

docs/conf.py

@@ -22,7 +22,7 @@
 author = 'Chris Bamford'
 
 # The full version, including alpha/beta/rc tags
-release = '1.3.6'
+release = '1.3.7'
 
 
 # -- General configuration ---------------------------------------------------

js/jiddly-app/package.json

@@ -1,6 +1,6 @@
 {
   "name": "jiddly-app",
-  "version": "1.3.6",
+  "version": "1.3.7",
   "private": true,
   "dependencies": {
     "@fortawesome/fontawesome-svg-core": "^6.1.1",

python/setup.py

@@ -71,7 +71,7 @@ def griddly_package_data(config='Debug'):
 
 setup(
     name='griddly',
-    version="1.3.6",
+    version="1.3.7",
     author_email="[email protected]",
     description="Griddly Python Libraries",
     long_description=long_description,

python/tests/gdy/simultaneous_player_actions.yaml

@@ -0,0 +1,37 @@
+Version: "0.1"
+Environment:
+  Name: test
+  Observers:
+    Vector:
+      IncludePlayerId: true
+    Block2D:
+      TileSize: [10,10]
+  Player:
+    AvatarObject: avatar
+    Count: 2
+  Levels:
+    - |
+      .  .  .  .  .
+      .  .  .  .  .
+      .  .  .  .  .
+      .  a1 .  a2 .
+      .  .  .  .  .
+      .  .  .  .  .
+
+Actions:
+  - Name: move
+    Behaviours:
+    - Src:
+        Object: avatar
+        Commands:
+          - mov: _dest
+      Dst:
+        Object: _empty
+
+Objects:
+  - Name: avatar
+    MapCharacter: a
+    Observers:
+      Block2D:
+        - Shape: square
+

python/tests/player_simultaneous_actions.py

@@ -0,0 +1,46 @@
+import pytest
+import gym
+from griddly import GymWrapperFactory, gd
+
+from collections import Counter
+
+
+@pytest.fixture
+def test_name(request):
+    return request.node.name
+
+
+def build_test_env(test_name, yaml_file, **kwargs):
+    wrapper_factory = GymWrapperFactory()
+
+    wrapper_factory.build_gym_from_yaml(
+        test_name,
+        yaml_file,
+        **kwargs,
+    )
+
+    env = gym.make(f"GDY-{test_name}-v0")
+    env.reset()
+    return env
+
+
+def test_simultaneous_actions_stochasticity(test_name):
+    """
+    Using observers in 0.1 version of GDY
+    """
+    env = build_test_env(test_name, "tests/gdy/simultaneous_player_actions.yaml", global_observer_type=gd.ObserverType.VECTOR,
+                         player_observer_type=gd.ObserverType.VECTOR)
+
+    possible_obs_hashes = Counter()
+
+    for i in range(100):
+        obs, reward, done, info = env.step([3, 1])
+        obs_hash_player_1 = hash(tuple(obs[0].flatten()))
+        obs_hash_player_2 = hash(tuple(obs[1].flatten()))
+        possible_obs_hashes[obs_hash_player_1] += 1
+        possible_obs_hashes[obs_hash_player_2] += 1
+        env.reset()
+
+    assert len(possible_obs_hashes) == 4
+
+

src/Griddly/Core/Grid.cpp

@@ -297,6 +297,9 @@ std::unordered_map<uint32_t, int32_t> Grid::performActions(uint32_t playerId, st
 
   spdlog::trace("Tick {0}", *gameTicks_);
 
+  // Have to add some non-determinism here for multi-agent games so player 1 doesn't always benefit
+  std::shuffle(std::begin(actions), std::end(actions), randomGenerator_->getEngine());
+
   for (const auto& action : actions) {
     // Check if action is delayed or durative
     if (action->getDelay() > 0) {

src/Griddly/Core/Util/RandomGenerator.cpp

@@ -17,4 +17,9 @@ const float RandomGenerator::sampleFloat(float min, float max) {
   return dist(randomGenerator_);
 }
 
+std::mt19937& RandomGenerator::getEngine() {
+  return randomGenerator_;
+}
+
+
 }  // namespace griddly

src/Griddly/Core/Util/RandomGenerator.hpp

@@ -10,6 +10,8 @@ class RandomGenerator {
 
   virtual const float sampleFloat(float min, float max);
 
+  virtual std::mt19937& getEngine();
+
  private:
   // Random number generator for the grid and associated objects
   std::mt19937 randomGenerator_ = std::mt19937();