sinter 1.14.dev1718081493__tar.gz → 1.14.dev1719955661__tar.gz

{sinter-1.14.dev1718081493 → sinter-1.14.dev1719955661}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: sinter
-Version: 1.14.dev1718081493
+Version: 1.14.dev1719955661
 Summary: Samples stim circuits and decodes them using pymatching.
 Author: Craig Gidney
 Author-email: craig.gidney@gmail.com

{sinter-1.14.dev1718081493 → sinter-1.14.dev1719955661}/setup.py

@@ -19,7 +19,7 @@ with open('README.md', encoding='UTF-8') as f:
 with open('requirements.txt', encoding='UTF-8') as f:
     requirements = f.read().splitlines()
 
-__version__ = '1.14.dev1718081493'
+__version__ = '1.14.dev1719955661'
 
 setup(
     name='sinter',

{sinter-1.14.dev1718081493 → sinter-1.14.dev1719955661}/src/sinter/__init__.py

@@ -1,4 +1,4 @@
-__version__ = '1.14.dev1718081493'
+__version__ = '1.14.dev1719955661'
 
 from sinter._anon_task_stats import (
     AnonTaskStats,

{sinter-1.14.dev1718081493 → sinter-1.14.dev1719955661}/src/sinter/_probability_util.py

@@ -7,7 +7,16 @@ import numpy as np
 
 if TYPE_CHECKING:
     import sinter
-    from scipy.stats._stats_mstats_common import LinregressResult
+
+    # Go on a magical journey looking for scipy's linear regression type.
+    try:
+        from scipy.stats._stats_py import LinregressResult
+    except ImportError:
+        try:
+            from scipy.stats._stats_mstats_common import LinregressResult
+        except ImportError:
+            from scipy.stats import linregress
+            LinregressResult = type(linregress([0, 1], [0, 1]))
 
 
 def log_binomial(*, p: Union[float, np.ndarray], n: int, hits: int) -> np.ndarray:
@@ -64,8 +73,8 @@ def log_binomial(*, p: Union[float, np.ndarray], n: int, hits: int) -> np.ndarra
         result[p_clipped == 1] = -np.inf
 
     # Multiply p**hits and (1-p)**misses onto the total, in log space.
-    result[p_clipped != 0] += np.log(p_clipped[p_clipped != 0]) * hits
-    result[p_clipped != 1] += np.log1p(-p_clipped[p_clipped != 1]) * misses
+    result[p_clipped != 0] += np.log(p_clipped[p_clipped != 0]) * float(hits)
+    result[p_clipped != 1] += np.log1p(-p_clipped[p_clipped != 1]) * float(misses)
 
     # Multiply (n choose hits) onto the total, in log space.
     log_n_choose_hits = log_factorial(n) - log_factorial(misses) - log_factorial(hits)
@@ -150,7 +159,10 @@ def least_squares_cost(*, xs: np.ndarray, ys: np.ndarray, intercept: float, slop
 def least_squares_through_point(*, xs: np.ndarray, ys: np.ndarray, required_x: float, required_y: float) -> 'LinregressResult':
     # Local import to reduce initial cost of importing sinter.
     from scipy.optimize import leastsq
-    from scipy.stats._stats_mstats_common import LinregressResult
+    from scipy.stats import linregress
+
+    # HACK: get scipy's linear regression result type
+    LinregressResult = type(linregress([0, 1], [0, 1]))
 
     xs2 = xs - required_x
     ys2 = ys - required_y
@@ -169,7 +181,11 @@ def least_squares_with_slope(*, xs: np.ndarray, ys: np.ndarray, required_slope:
 
     # Local import to reduce initial cost of importing sinter.
     from scipy.optimize import leastsq
-    from scipy.stats._stats_mstats_common import LinregressResult
+
+    # HACK: get scipy's linear regression result type
+    from scipy.stats import linregress
+    LinregressResult = type(linregress([0, 1], [0, 1]))
+
     (best_intercept,), _ = leastsq(func=err, x0=0.0)
     return LinregressResult(required_slope, best_intercept, None, None, None, intercept_stderr=False)
 

{sinter-1.14.dev1718081493 → sinter-1.14.dev1719955661}/src/sinter.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: sinter
-Version: 1.14.dev1718081493
+Version: 1.14.dev1719955661
 Summary: Samples stim circuits and decodes them using pymatching.
 Author: Craig Gidney
 Author-email: craig.gidney@gmail.com