Small change to derive_predictors_and_scores for speed + normalization

.gitignore

@@ -155,3 +155,11 @@ data/communicators/communicators_data/*
 data/central_park_birds_2022/*
 data/foraging/central_park_birds_cleaned_2022/*
 .vscode/launch.json
+docs/foraging/random-hungry-followers/random_hungry_followers.ipynb
+docs/foraging/random-hungry-followers/random_hungry_followers.ipynb
+data/foraging/locust/ds/length_s0_e10_i500_15EQ20191202.pkl
+data/foraging/locust/ds/priors_sam30_15EQ20191202_s0_e10.pkl
+docs/foraging/communicators/communicators_derivedDF_0.pkl
+docs/foraging/communicators/communicators_derivedDF_6.pkl
+docs/foraging/communicators/communicators0.pkl
+docs/foraging/communicators/communicators6.pkl

collab/foraging/locust/ds_locust_functions.py

@@ -218,9 +218,9 @@ def forward(self, X: State[torch.Tensor]):
 
 def bayesian_locust(base_model=LocustDynamics) -> Dynamics[torch.Tensor]:
     with pyro.plate("attr", size=5):
-        attraction = pyro.sample("attraction", dist.Uniform(0.00001, 0.1))
+        attraction = pyro.sample("attraction", dist.Uniform(0.00001, 0.1))  # type: ignore
     with pyro.plate("wond", size=4):
-        wander = pyro.sample("wander", dist.Uniform(0.00001, 0.3))
+        wander = pyro.sample("wander", dist.Uniform(0.00001, 0.3))  # type: ignore
     locust_model = base_model(attraction, wander)
     return locust_model
 
@@ -237,7 +237,7 @@ def locust_noisy_model(X: State[torch.Tensor]) -> None:
     with pyro.plate("data", len(X["edge_l"])):
         pyro.sample(
             "counts_obs",
-            dist.Multinomial(total_count, probs=probs),  # .to_event(event_dim)
+            dist.Multinomial(total_count, probs=probs),  # type: ignore
         )
 
 

collab2/foraging/locust/ds_locust_functions.py

@@ -218,9 +218,9 @@ def forward(self, X: State[torch.Tensor]):
 
 def bayesian_locust(base_model=LocustDynamics) -> Dynamics[torch.Tensor]:
     with pyro.plate("attr", size=5):
-        attraction = pyro.sample("attraction", dist.Uniform(0.00001, 0.1))
+        attraction = pyro.sample("attraction", dist.Uniform(0.00001, 0.1))  # type: ignore
     with pyro.plate("wond", size=4):
-        wander = pyro.sample("wander", dist.Uniform(0.00001, 0.3))
+        wander = pyro.sample("wander", dist.Uniform(0.00001, 0.3))  # type: ignore
     locust_model = base_model(attraction, wander)
     return locust_model
 
@@ -237,7 +237,7 @@ def locust_noisy_model(X: State[torch.Tensor]) -> None:
     with pyro.plate("data", len(X["edge_l"])):
         pyro.sample(
             "counts_obs",
-            dist.Multinomial(total_count, probs=probs),  # .to_event(event_dim)
+            dist.Multinomial(total_count, probs=probs),  # type: ignore
         )
 
 

collab2/foraging/random_hungry_followers/follower_birds.py

@@ -3,7 +3,7 @@
 import numpy as np
 import pandas as pd
 
-from collab.foraging.toolkit import (
+from collab2.foraging.random_hungry_followers.rhf_helpers import (
     construct_visibility,
     generate_proximity_score,
     update_rewards,
@@ -19,6 +19,7 @@ def add_follower_foragers(
     getting_worse=1.5,
     optimal=4,
     proximity_decay=1,
+    initial_positions=None,
 ):
     """
         A function to add follower foragers to a simulation.
@@ -35,9 +36,9 @@ def add_follower_foragers(
     # TODO Check if different forager types mix well
     how_many_foragers_already = len(old_foragers)
 
-    new_foragers = sim.generate_random_foragers(num_follower_foragers, size=1)[
-        "random_foragers"
-    ]
+    new_foragers = sim.generate_random_foragers(
+        num_follower_foragers, size=1, initial_positions=initial_positions
+    )["random_foragers"]
 
     for new_forager in new_foragers:
         new_forager["forager"] = new_forager["forager"] + how_many_foragers_already
@@ -81,8 +82,8 @@ def add_follower_foragers(
                 new_row = {
                     "x": new_x,
                     "y": new_y,
-                    "time": t + 2,
-                    "forager": b + 1,
+                    "time": t + 1,
+                    "forager": b,
                     "type": "follower",
                 }
 

collab2/foraging/random_hungry_followers/hungry_birds.py

@@ -3,14 +3,12 @@
 import numpy as np
 import pandas as pd
 
-from collab.foraging.toolkit import (
+from collab2.foraging.random_hungry_followers.rhf_helpers import (
     construct_visibility,
     rewards_to_trace,
     update_rewards,
 )
 
-logging.basicConfig(level=logging.INFO, format="%(asctime)s:  %(message)s")
-
 
 def add_hungry_foragers(
     sim,
@@ -53,9 +51,10 @@ def add_hungry_foragers(
             end=t + 1,
         )["visibility"]
 
-        sim.rewards = update_rewards(
-            sim, sim.rewards, new_foragers, start=t, end=t + 1
-        )["rewards"]
+        if t > 0:  # DB: no need to update rewards for the first frame
+            sim.rewards = update_rewards(
+                sim, sim.rewards, new_foragers, start=t, end=t + 1
+            )["rewards"]
 
         sim.traces = rewards_to_trace(
             sim.rewards,
@@ -79,8 +78,8 @@ def add_hungry_foragers(
                 new_row = {
                     "x": new_x,
                     "y": new_y,
-                    "time": t + 2,
-                    "forager": b + 1,
+                    "time": t + 1,
+                    "forager": b,
                     "type": "hungry",
                 }
 

collab2/foraging/random_hungry_followers/random_foragers.py

@@ -3,7 +3,10 @@
 import numpy as np
 import pandas as pd
 
-from collab.foraging.toolkit import generate_grid, update_rewards
+from collab2.foraging.random_hungry_followers.rhf_helpers import (
+    generate_grid,
+    update_rewards,
+)
 
 
 class Foragers:
@@ -45,7 +48,7 @@ def __init__(
 
             include_random_foragers (bool): Whether to include random foragers in
                             the final output (if you only use the sim as
-                            a starting point for another foragint strategy,
+                            a starting point for another foraging strategy,
                             you might want to set this to False).
 
         Other attributes:
@@ -122,9 +125,16 @@ def __call__(self):
             self.rewards = rew["rewards"]
             self.rewardsDF = rew["rewardsDF"]
 
-    def generate_random_foragers(self, num_foragers, size=None):
+    def generate_random_foragers(self, num_foragers, size=None, initial_positions=None):
         if size is None:
             size = self.num_frames
+
+        # set to (grid_size/2,grid_size/2) for each forager if not provided
+        if initial_positions is None:
+            initial_positions = np.array(
+                [[self.grid_size / 2, self.grid_size / 2]] * num_foragers
+            )
+
         random_foragers = []
 
         size_warning_flag = False
@@ -137,14 +147,12 @@ def generate_random_foragers(self, num_foragers, size=None):
                     p=self.probabilities,
                     replace=True,
                 )
-            ) + (
-                self.grid_size / 2
-            )  # make centered
+            ) + (initial_positions[forager, 0])
 
-            if any(forager_x < 0) or any(forager_x > self.grid_size):
+            if any(forager_x < 0) or any(forager_x >= self.grid_size):
                 size_warning_flag = True
                 forager_x[forager_x < 0] = 0
-                forager_x[forager_x > self.grid_size] = self.grid_size
+                forager_x[forager_x >= self.grid_size] = self.grid_size - 1
 
             forager_y = np.cumsum(
                 np.random.choice(
@@ -153,11 +161,11 @@ def generate_random_foragers(self, num_foragers, size=None):
                     p=self.probabilities,
                     replace=True,
                 )
-            ) + (self.grid_size / 2)
+            ) + (initial_positions[forager, 1])
 
-            if any(forager_y < 0) or any(forager_y > self.grid_size):
+            if any(forager_y < 0) or any(forager_y >= self.grid_size):
                 forager_y[forager_y < 0] = 0
-                forager_y[forager_y > self.grid_size] = self.grid_size
+                forager_y[forager_y >= self.grid_size] = self.grid_size - 1
 
             if size_warning_flag:
                 warnings.warn(
@@ -170,8 +178,8 @@ def generate_random_foragers(self, num_foragers, size=None):
                 {
                     "x": forager_x,
                     "y": forager_y,
-                    "time": range(1, size + 1),
-                    "forager": forager + 1,
+                    "time": range(0, size),
+                    "forager": forager,
                     "type": "random",
                 }
             )
@@ -187,11 +195,11 @@ def generate_random_foragers(self, num_foragers, size=None):
     def generate_random_rewards(self, size=None):
         if size is None:
             size = self.num_rewards
-        rewardsX = np.random.choice(range(1, self.grid_size + 1), size=size)
-        rewardsY = np.random.choice(range(1, self.grid_size + 1), size=size)
+        rewardsX = np.random.choice(range(0, self.grid_size), size=size)
+        rewardsY = np.random.choice(range(0, self.grid_size), size=size)
 
         rewards = []
-        for t in range(1, self.num_frames + 1):
+        for t in range(0, self.num_frames):
             rewards.append(pd.DataFrame({"x": rewardsX, "y": rewardsY, "time": t}))
 
         return {"rewards": rewards, "rewardsDF": pd.concat(rewards)}