Merge d1ef49d into 473431c

arch/tests/univariate/test_rescale.py

@@ -0,0 +1,53 @@
+import numpy as np
+from numpy.testing import assert_allclose
+import pandas as pd
+import pytest
+
+from arch.univariate import GARCH, Normal, ZeroMean
+
+
+@pytest.fixture(scope="module")
+def small_data():
+    rs = np.random.RandomState([2389280, 238901, 382908031])
+    mod = ZeroMean(None, volatility=GARCH(), distribution=Normal(random_state=rs))
+    sim = mod.simulate([1e-4, 0.05, 0.90], nobs=1000)
+    return sim.data
+
+
+@pytest.fixture(scope="module")
+def small_data2():
+    rs = np.random.RandomState([2389280, 238901, 382908031])
+    mod = ZeroMean(None, volatility=GARCH(), distribution=Normal(random_state=rs))
+    sim = mod.simulate([1e-4, 0.05, 0.90], nobs=1000)
+    return sim.data
+
+
+@pytest.fixture(scope="module")
+def std_data():
+    rs = np.random.RandomState([2389280, 238901, 382908031])
+    mod = ZeroMean(None, volatility=GARCH(), distribution=Normal(random_state=rs))
+    sim = mod.simulate([1e-1, 0.05, 0.90], nobs=1000)
+    return sim.data
+
+
+def test_reproducibility(small_data, small_data2):
+    pd.testing.assert_series_equal(small_data, small_data2)
+
+
+def test_blank(small_data, std_data):
+    small_mod = ZeroMean(small_data, volatility=GARCH(), rescale=False)
+    small_res = small_mod.fit(disp="off")
+    mod = ZeroMean(std_data, volatility=GARCH(), rescale=False)
+    res = mod.fit(disp="off")
+    assert_allclose(1e3 * small_res.params[0], res.params[0], rtol=5e-3)
+
+
+def test_rescale_fit(small_data, std_data):
+    small_mod = ZeroMean(small_data, volatility=GARCH(), rescale=True)
+    small_res = small_mod.fit(disp="off")
+    direct_mod = ZeroMean(10 * small_data, volatility=GARCH())
+    direct_res = direct_mod.fit(disp="off")
+    assert_allclose(small_res.loglikelihood, direct_res.loglikelihood)
+    small_fcast = small_res.forecast(start=0)
+    direct_fcast = direct_res.forecast(start=0)
+    assert_allclose(small_fcast.variance, direct_fcast.variance)

arch/univariate/distribution.py

@@ -776,7 +776,7 @@ def starting_values(self, std_resid: ArrayLike1D) -> NDArray:
     def _simulator(self, size: Union[int, Tuple[int, ...]]) -> NDArray:
         # No need to normalize since it is already done in parameterization
         assert self._parameters is not None
-        return self.ppf(stats.uniform.rvs(size=size), self._parameters)
+        return self.ppf(self._random_state.random_sample(size=size), self._parameters)
 
     def simulate(
         self, parameters: Union[int, float, Sequence[Union[float, int]], ArrayLike1D]

doc/source/changes/4.0.txt

@@ -6,6 +6,9 @@ Since 4.12
 ==========
 - Issue warnings when unit root tests are mutated. Will raise after 5.0
   is released.
+- Fixed a bug in :class:`arch.univariate.SkewStudent` which did not use the
+  user-provided `RandomState` when one was provided. This prevented
+  reproducing simulated values (:issue:``).
 
 Release 4.12
 ============