Fix overflow errors uncovered by Coverage-guided Fuzzing.

.github/workflows/tests.yml

@@ -37,6 +37,6 @@ jobs:
           architecture: x64
       - name: Install dependencies
         run: |
-          python -m pip install --upgrade pip setuptools numpy
+          python -m pip install --upgrade pip setuptools numpy hypothesis
           pip install --progress-bar off .
       - run: python -m unittest

.gitignore

@@ -7,6 +7,7 @@ __pycache__/
 .eggs/
 *.egg-info/
 .tox/
+.hypothesis
 
 tmp/
 benchmark/*.json

cmaes/_cma.py

@@ -10,6 +10,7 @@
 
 
 _EPS = 1e-8
+_MEAN_MAX = 1e32
 _SIGMA_MAX = 1e32
 
 
@@ -78,6 +79,10 @@ def __init__(
     ):
         assert sigma > 0, "sigma must be non-zero positive value"
 
+        assert np.all(
+            np.abs(mean) < _MEAN_MAX
+        ), f"Abs of all elements of mean vector must be less than {_MEAN_MAX}"
+
         n_dim = len(mean)
         assert n_dim > 1, "The dimension of mean must be larger than 1"
 
@@ -280,8 +285,12 @@ def _repair_infeasible_params(self, param: np.ndarray) -> np.ndarray:
 
     def tell(self, solutions: List[Tuple[np.ndarray, float]]) -> None:
         """Tell evaluation values"""
-        if len(solutions) != self._popsize:
-            raise ValueError("Must tell popsize-length solutions.")
+
+        assert len(solutions) == self._popsize, "Must tell popsize-length solutions."
+        for s in solutions:
+            assert np.all(
+                np.abs(s[0]) < _MEAN_MAX
+            ), f"Abs of all param values must be less than {_MEAN_MAX} to avoid overflow errors"
 
         self._g += 1
         solutions.sort(key=lambda s: s[1])

requirements-dev.txt

@@ -5,6 +5,9 @@ numpy>=1.20.0
 matplotlib
 scipy
 
+# tests
+hypothesis
+
 # lint
 mypy
 flake8

tests/test_fuzzing.py

@@ -0,0 +1,35 @@
+import hypothesis.extra.numpy as npst
+import unittest
+from hypothesis import given, strategies as st
+
+from cmaes import CMA
+
+
+class TestFuzzing(unittest.TestCase):
+    @given(
+        data=st.data(),
+    )
+    def test_cma_tell(self, data):
+        dim = data.draw(st.integers(min_value=2, max_value=100))
+        mean = data.draw(npst.arrays(dtype=float, shape=dim))
+        sigma = data.draw(st.floats(min_value=1e-16))
+        n_iterations = data.draw(st.integers(min_value=1))
+        try:
+            optimizer = CMA(mean, sigma)
+        except AssertionError:
+            return
+        popsize = optimizer.population_size
+        for _ in range(n_iterations):
+            tell_solutions = data.draw(
+                st.lists(
+                    st.tuples(npst.arrays(dtype=float, shape=dim), st.floats()),
+                    min_size=popsize,
+                    max_size=popsize,
+                )
+            )
+            optimizer.ask()
+            try:
+                optimizer.tell(tell_solutions)
+            except AssertionError:
+                return
+            optimizer.ask()