Update app.py

app.py

@@ -29,6 +29,8 @@ import pandas as pd
 import plotly.express as px
 from plotly.graph_objects import Figure
 
+TOLERANCE, LIMIT = 1e-16, 1000
+
 
 def visualize(df_pairwise: pd.DataFrame) -> Figure:
     fig = px.imshow(df_pairwise, color_continuous_scale='RdBu', text_auto='.2f')
@@ -38,8 +40,7 @@ def visualize(df_pairwise: pd.DataFrame) -> Figure:
 
 
 # https://gist.github.com/dustalov/41678b70c40ba5a55430fa5e77b121d9#file-bradley_terry-py
-def bradley_terry(wins: npt.NDArray[np.int64], ties: npt.NDArray[np.int64],
-                  seed: int = 0, tolerance: float = 10e-6, limit: int = 20) -> npt.NDArray[np.float64]:
+def bradley_terry(wins: npt.NDArray[np.int64], ties: npt.NDArray[np.int64]) -> npt.NDArray[np.float64]:
     M = wins + .5 * ties
 
     T = M.T + M
@@ -65,7 +66,7 @@ def bradley_terry(wins: npt.NDArray[np.int64], ties: npt.NDArray[np.int64],
         p_new /= Z.sum(axis=0)
         p_new /= p_new.sum()
 
-        converged = bool(np.linalg.norm(p_new - p) < tolerance) or (iterations >= limit)
+        converged = bool(np.linalg.norm(p_new - p) < TOLERANCE) or (iterations >= LIMIT)
 
         p[:] = p_new
 
@@ -85,30 +86,26 @@ def centrality(algorithm: Callable[[nx.DiGraph], dict[int, float]],
     return p
 
 
-def counting(wins: npt.NDArray[np.int64], ties: npt.NDArray[np.int64],
-             seed: int = 0, tolerance: float = 10e-6, limit: int = 100) -> npt.NDArray[np.float64]:
+def counting(wins: npt.NDArray[np.int64], ties: npt.NDArray[np.int64]) -> npt.NDArray[np.float64]:
     M = wins + .5 * ties
 
     return cast(npt.NDArray[np.float64], M.sum(axis=1))
 
 
-def eigen(wins: npt.NDArray[np.int64], ties: npt.NDArray[np.int64],
-          seed: int = 0, tolerance: float = 10e-6, limit: int = 100) -> npt.NDArray[np.float64]:
-    algorithm = partial(nx.algorithms.eigenvector_centrality_numpy, max_iter=limit, tol=tolerance, weight='weight')
+def eigen(wins: npt.NDArray[np.int64], ties: npt.NDArray[np.int64]) -> npt.NDArray[np.float64]:
+    algorithm = partial(nx.algorithms.eigenvector_centrality_numpy, max_iter=LIMIT, tol=TOLERANCE, weight='weight')
 
     return centrality(algorithm, wins, ties)
 
 
-def pagerank(wins: npt.NDArray[np.int64], ties: npt.NDArray[np.int64],
-             seed: int = 0, tolerance: float = 10e-6, limit: int = 100) -> npt.NDArray[np.float64]:
-    algorithm = partial(nx.algorithms.pagerank, max_iter=limit, tol=tolerance, weight='weight')
+def pagerank(wins: npt.NDArray[np.int64], ties: npt.NDArray[np.int64]) -> npt.NDArray[np.float64]:
+    algorithm = partial(nx.algorithms.pagerank, max_iter=LIMIT, tol=TOLERANCE, weight='weight')
 
     return centrality(algorithm, wins, ties)
 
 
 # https://gist.github.com/dustalov/41678b70c40ba5a55430fa5e77b121d9#file-newman-py
-def newman(wins: npt.NDArray[np.int64], ties: npt.NDArray[np.int64],
-           seed: int = 0, tolerance: float = 10e-6, limit: int = 20) -> npt.NDArray[np.float64]:
+def newman(wins: npt.NDArray[np.int64], ties: npt.NDArray[np.int64]) -> npt.NDArray[np.float64]:
     pi, v = np.ones(wins.shape[0]), .5
 
     converged, iterations = False, 0
@@ -127,7 +124,7 @@ def newman(wins: npt.NDArray[np.int64], ties: npt.NDArray[np.int64],
         )
 
         v = v_numerator / v_denominator
-        v = np.nan_to_num(v, nan=tolerance)
+        v = np.nan_to_num(v, nan=TOLERANCE)
 
         pi_old = pi.copy()
 
@@ -144,10 +141,10 @@ def newman(wins: npt.NDArray[np.int64], ties: npt.NDArray[np.int64],
         )
 
         pi = pi_numerator / pi_denominator
-        pi = np.nan_to_num(pi, nan=tolerance)
+        pi = np.nan_to_num(pi, nan=TOLERANCE)
 
         converged = np.allclose(pi / (pi + 1), pi_old / (pi_old + 1),
-                                rtol=tolerance, atol=tolerance) or (iterations >= limit)
+                                rtol=TOLERANCE, atol=TOLERANCE) or (iterations >= LIMIT)
 
     return pi
 
@@ -219,7 +216,7 @@ def handler(file: BinaryIO, algorithm: str, filtered: bool, truncated: bool, see
 
     assert wins.shape == ties.shape, 'wins and ties shapes are different'
 
-    scores = ALGORITHMS[algorithm](wins, ties, seed=seed)
+    scores = ALGORITHMS[algorithm](wins, ties)
 
     df_result = pd.DataFrame(data={'score': scores}, index=index)