continual-foragax 0.36.1__py3-none-any.whl → 0.38.0__py3-none-any.whl

{continual_foragax-0.36.1.dist-info → continual_foragax-0.38.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: continual-foragax
-Version: 0.36.1
+Version: 0.38.0
 Summary: A continual reinforcement learning benchmark
 Author-email: Steven Tang <stang5@ualberta.ca>
 Requires-Python: >=3.8

{continual_foragax-0.36.1.dist-info → continual_foragax-0.38.0.dist-info}/RECORD

@@ -1,8 +1,8 @@
 foragax/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 foragax/colors.py,sha256=rqNPiywP4Nvr0POhsGpasRk-nMMTS3DOwFRUgperlUk,2065
-foragax/env.py,sha256=7IhLCEosM_IH19sMzRnGMvTNq9tlhVg4PU52hGz9XLE,66531
-foragax/objects.py,sha256=PPuLYjD7em7GL404eSpP6q8TxF8p7JtQ1kIwh7uD_tU,26860
-foragax/registry.py,sha256=y_MEvM0K_E4fzdp4XokzLh1D94UAuCu16wGwJV4wa_A,19289
+foragax/env.py,sha256=9EB02GgSc80KRjj35z_UyM4VXTGuyHBAHX3d9M5AvY0,66966
+foragax/objects.py,sha256=aVc7lD3CTyRP9wm_Vs93qo4l_B1kbiYGKPtkd_SVXjs,27061
+foragax/registry.py,sha256=u8mv5mPTHKRsIE7FTi36J5tXp3FuVJ-AmJRb7HWW9hI,20636
 foragax/rendering.py,sha256=bms7wvBZTofoR-K-2QD2Ggeed7Viw8uwAEiEpEM3eSo,2768
 foragax/weather.py,sha256=KNAiwuFz8V__6G75vZIWQKPocLzXqxXn-Vt4TbHIpcA,1258
 foragax/data/ECA_non-blended_custom/TG_SOUID100897.txt,sha256=N7URbX6VlCZvCboUogYjMzy1I-0cfNPOn0QTLSHHfQ0,1776751
@@ -128,8 +128,8 @@ foragax/data/ECA_non-blended_custom/TG_SOUID156887.txt,sha256=juzTPgJoJxfqmZkorL
 foragax/data/ECA_non-blended_custom/elements.txt,sha256=OtcUBoDAHxuln79BPKGu0tsQxG_5G2BfAX3Ck130kEA,4507
 foragax/data/ECA_non-blended_custom/metadata.txt,sha256=nudnmOCy5cPJfSXt_IjyX0S5-T7NkCZREICZSimqeqc,48260
 foragax/data/ECA_non-blended_custom/sources.txt,sha256=1j3lSmINAoCMqPqFrHfZJriOz6sTYZNOhXzUwvTLas0,20857
-continual_foragax-0.36.1.dist-info/METADATA,sha256=SgkncBLbu9_MerET0r6Zto2YSLqfvmF4VE9vYXXh87g,4713
-continual_foragax-0.36.1.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-continual_foragax-0.36.1.dist-info/entry_points.txt,sha256=Qiu6iE_XudrDO_bVAMeA435h4PO9ourt8huvSHiuMPc,41
-continual_foragax-0.36.1.dist-info/top_level.txt,sha256=-z3SDK6RfLIcLI24n8rdbeFzlVY3hunChzlu-v1Fncs,8
-continual_foragax-0.36.1.dist-info/RECORD,,
+continual_foragax-0.38.0.dist-info/METADATA,sha256=ODZtbDZHv2GRj0yRC4rR34Hu0PaN_8e4Oo8XPmhWGxU,4713
+continual_foragax-0.38.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+continual_foragax-0.38.0.dist-info/entry_points.txt,sha256=Qiu6iE_XudrDO_bVAMeA435h4PO9ourt8huvSHiuMPc,41
+continual_foragax-0.38.0.dist-info/top_level.txt,sha256=-z3SDK6RfLIcLI24n8rdbeFzlVY3hunChzlu-v1Fncs,8
+continual_foragax-0.38.0.dist-info/RECORD,,

foragax/env.py

@@ -132,6 +132,7 @@ class ForagaxEnv(environment.Environment):
         observation_type: str = "object",
         dynamic_biomes: bool = False,
         biome_consumption_threshold: float = 0.9,
+        dynamic_biome_spawn_empty: float = 0.0,
         max_expiries_per_step: int = 1,
     ):
         super().__init__()
@@ -156,6 +157,7 @@ class ForagaxEnv(environment.Environment):
         self.teleport_interval = teleport_interval
         self.dynamic_biomes = dynamic_biomes
         self.biome_consumption_threshold = biome_consumption_threshold
+        self.dynamic_biome_spawn_empty = dynamic_biome_spawn_empty
         if max_expiries_per_step < 1:
             raise ValueError("max_expiries_per_step must be at least 1")
         self.max_expiries_per_step = max_expiries_per_step
@@ -888,7 +890,7 @@ class ForagaxEnv(environment.Environment):
         biome_freqs = self.biome_object_frequencies[biome_idx]
         biome_mask = self.biome_masks_array[biome_idx]
 
-        key, spawn_key, color_key, params_key = jax.random.split(key, 4)
+        key, spawn_key, color_key, params_key, dropout_key = jax.random.split(key, 5)
 
         # Generate object IDs using deterministic or random spawn
         if deterministic:
@@ -927,6 +929,13 @@ class ForagaxEnv(environment.Environment):
                 jnp.searchsorted(cumulative_freqs, grid_rand, side="right") - 1
             )
 
+        # Apply random dropout if configured
+        if self.dynamic_biome_spawn_empty > 0:
+            dropout_mask = jax.random.bernoulli(
+                dropout_key, 1.0 - self.dynamic_biome_spawn_empty, object_grid.shape
+            )
+            object_grid = jnp.where(dropout_mask, object_grid, 0)
+
         # Initialize color grid
         color_grid = jnp.full((self.size[1], self.size[0], 3), 255, dtype=jnp.uint8)
 

foragax/objects.py

@@ -241,6 +241,7 @@ class FourierObject(BaseForagaxObject):
         reward_delay: int = 0,
         max_reward_delay: Optional[int] = None,
         regen_delay: Optional[Tuple[int, int]] = None,
+        reward_repeat: int = 1,
     ):
         if max_reward_delay is None:
             max_reward_delay = reward_delay
@@ -257,6 +258,7 @@ class FourierObject(BaseForagaxObject):
         self.base_magnitude = base_magnitude
         self.reward_delay_val = reward_delay
         self.regen_delay_range = regen_delay
+        self.reward_repeat = reward_repeat
 
     def get_state(self, key: jax.Array) -> jax.Array:
         """Generate random Fourier series parameters.
@@ -321,7 +323,7 @@ class FourierObject(BaseForagaxObject):
         y_max = params[2]
 
         # Normalize time to [0, 2π] using the object's period
-        t = 2.0 * jnp.pi * (clock % period) / period
+        t = 2.0 * jnp.pi * ((clock // self.reward_repeat) % period) / period
 
         # Extract interleaved coefficients: [a1, b1, a2, b2, ...]
         ab_coeffs = params[3:]
@@ -715,6 +717,7 @@ def create_fourier_objects(
     base_magnitude: float = 1.0,
     reward_delay: int = 0,
     regen_delay: Optional[Tuple[int, int]] = None,
+    reward_repeat: int = 1,
 ):
     """Create HOT and COLD FourierObject instances.
 
@@ -733,6 +736,7 @@ def create_fourier_objects(
         color=(0, 0, 0),
         reward_delay=reward_delay,
         regen_delay=regen_delay,
+        reward_repeat=reward_repeat,
     )
 
     cold = FourierObject(
@@ -742,6 +746,7 @@ def create_fourier_objects(
         color=(0, 0, 0),
         reward_delay=reward_delay,
         regen_delay=regen_delay,
+        reward_repeat=reward_repeat,
     )
 
     return hot, cold

foragax/registry.py

@@ -128,6 +128,32 @@ ENV_CONFIGS: Dict[str, Dict[str, Any]] = {
         "deterministic_spawn": True,
         "dynamic_biomes": True,
     },
+    "ForagaxDiwali-v4": {
+        "size": (15, 15),
+        "aperture_size": None,
+        "objects": None,
+        "biomes": (
+            # Hot biome
+            Biome(start=(0, 2), stop=(15, 6), object_frequencies=(0.5, 0.0)),
+            # Cold biome
+            Biome(start=(0, 9), stop=(15, 13), object_frequencies=(0.0, 0.5)),
+        ),
+        "nowrap": False,
+        "deterministic_spawn": True,
+        "dynamic_biomes": True,
+        "biome_consumption_threshold": 1000,
+    },
+    "ForagaxDiwali-v5": {
+        "size": (15, 15),
+        "aperture_size": None,
+        "objects": None,
+        "biomes": (Biome(start=(0, 0), stop=(15, 15), object_frequencies=(0.2,)),),
+        "nowrap": False,
+        "deterministic_spawn": True,
+        "dynamic_biomes": True,
+        "biome_consumption_threshold": 1000,
+        "dynamic_biome_spawn_empty": 0.1,
+    },
     "ForagaxTwoBiome-v1": {
         "size": (15, 15),
         "aperture_size": None,
@@ -575,6 +601,20 @@ def make(
             reward_delay=reward_delay,
             regen_delay=(9, 11),
         )[:1]
+    if env_id == "ForagaxDiwali-v4":
+        config["objects"] = create_fourier_objects(
+            num_fourier_terms=10,
+            reward_delay=reward_delay,
+            regen_delay=(9, 11),
+            reward_repeat=100,
+        )
+    if env_id == "ForagaxDiwali-v5":
+        config["objects"] = create_fourier_objects(
+            num_fourier_terms=10,
+            reward_delay=reward_delay,
+            regen_delay=(9, 11),
+            reward_repeat=100,
+        )[:1]
 
     if env_id == "ForagaxSineTwoBiome-v1":
         biome1_oyster, biome1_deathcap, biome2_oyster, biome2_deathcap = (