ai-nk-cce 0.1.0__tar.gz → 0.1.1__tar.gz

{ai_nk_cce-0.1.0 → ai_nk_cce-0.1.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: ai-nk-cce
-Version: 0.1.0
+Version: 0.1.1
 Summary: This repository is used to train AI agents to predict good strategies in a social learning game based on a NK landscape.
 Author: Luis Mienhardt
 Author-email: mienhardt@mpib-berlin.mpg.de

{ai_nk_cce-0.1.0 → ai_nk_cce-0.1.1}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "ai-nk-cce"
-version = "0.1.0"
+version = "0.1.1"
 description = "This repository is used to train AI agents to predict good strategies in a social learning game based on a NK landscape."
 authors = [
 "Luis Mienhardt <mienhardt@mpib-berlin.mpg.de>",

{ai_nk_cce-0.1.0 → ai_nk_cce-0.1.1}/src/nk_model/biased_prediction_agent.py

@@ -32,7 +32,8 @@ class BiasedPredictionAgent:
             bias_power: Power applied to bias landscape G. Higher values
                 make the bias more pronounced for regions with higher G(x)
         """
-        self.landscape = landscape
+        self.landscape = self._normalize_landscape(landscape)
+        
         self.bias_seed = bias_seed
         self.bias_power = bias_power
 
@@ -49,6 +50,53 @@ class BiasedPredictionAgent:
             List[Tuple[np.ndarray, float]],
         ] = {}
 
+    def _normalize_landscape(self, landscape: NKLandscape) -> NKLandscape:
+        """
+        Normalize landscape payoffs to [0, 1] range.
+
+        Args:
+            landscape: NKLandscape to normalize
+
+        Returns:
+            New NKLandscape with normalized payoffs
+        """
+        # Extract payoffs
+        payoffs = np.array([item.payoff for item in landscape.items])
+        min_payoff = np.min(payoffs)
+        max_payoff = np.max(payoffs)
+
+        # Normalize to [0, 1]
+        if max_payoff > min_payoff:
+            normalized_payoffs = (payoffs - min_payoff) / (
+                max_payoff - min_payoff
+            )
+        else:
+            normalized_payoffs = np.zeros_like(payoffs)
+
+        # Create new landscape with normalized payoffs
+        normalized_items = [
+            Item(
+                coordinates=item.coordinates.copy(),
+                payoff=normalized_payoffs[i],
+            )
+            for i, item in enumerate(landscape.items)
+        ]
+
+        # Create a new NKLandscape object
+        normalized_landscape = NKLandscape.__new__(NKLandscape)
+        normalized_landscape.params = landscape.params
+        normalized_landscape.N = landscape.N
+        normalized_landscape.K = landscape.K
+        normalized_landscape.M = landscape.M
+        normalized_landscape.items = normalized_items
+        normalized_landscape._payoff_lookup = {
+            tuple(item.coordinates): item.payoff
+            for item in normalized_items
+        }
+        normalized_landscape.uuid = f"{landscape.uuid}_normalized"
+
+        return normalized_landscape
+
     def _create_bias(self) -> NKLandscape:
         """
         Create bias landscape G with N'=N, K'=0.