Circular kernel

RELEASE-NOTES.md

@@ -11,6 +11,7 @@
 
 ### New features
 - `sample_posterior_predictive_w` can now feed on `xarray.Dataset` - e.g. from `InferenceData.posterior`. (see [#4042](https://github.com/pymc-devs/pymc3/pull/4042))
+- Added `pymc3.gp.cov.Circular` kernel for Gaussian Processes on circular domains, e.g. the unit circle (see [#4082](https://github.com/pymc-devs/pymc3/pull/4082)).
 - Add MLDA, a new stepper for multilevel sampling. MLDA can be used when a hierarchy of approximate posteriors of varying accuracy is available, offering improved sampling efficiency especially in high-dimensional problems and/or where gradients are not available (see [#3926](https://github.com/pymc-devs/pymc3/pull/3926))
 - Change SMC metropolis kernel to independent metropolis kernel [#4115](https://github.com/pymc-devs/pymc3/pull/3926))
 

pymc3/gp/cov.py

@@ -299,6 +299,61 @@ def full(self, X, Xs=None):
             return tt.alloc(0.0, X.shape[0], Xs.shape[0])
 
 
+class Circular(Covariance):
+    R"""
+    Circular Kernel.
+
+    .. math::
+
+        k_g(x, y) = W_\pi(\operatorname{dist}_{\mathit{geo}}(x, y)),
+
+    with
+
+    .. math::
+
+        W_c(t) = \left(1 + \tau \frac{t}{c}\right)\left(1-\frac{t}{c}\right)^\tau_+
+
+    where :math:`c` is maximum value for :math:`t` and :math:`\tau\ge 4`.
+    :math:`\tau` controls for correlation strength, larger :math:`\tau` leads to less smooth functions
+    See [1]_ for more explanations and use cases.
+
+    Parameters
+    ----------
+    period : scalar
+        defines the circular interval :math:`[0, \mathit{bound})`
+    tau : scalar
+        :math:`\tau\ge 4` defines correlation strength, the larger,
+        the smaller correlation is. Minimum value is :math:`4`
+
+    References
+    ----------
+    .. [1] Espéran Padonou, O Roustant, "Polar Gaussian Processes for Predicting on Circular Domains"
+    https://hal.archives-ouvertes.fr/hal-01119942v1/document
+    """
+
+    def __init__(self, input_dim, period, tau=4, active_dims=None):
+        super().__init__(input_dim, active_dims)
+        self.c = tt.as_tensor_variable(period / 2)
+        self.tau = tau
+
+    def dist(self, X, Xs):
+        if Xs is None:
+            Xs = tt.transpose(X)
+        else:
+            Xs = tt.transpose(Xs)
+        return tt.abs_((X - Xs + self.c) % (self.c * 2) - self.c)
+
+    def weinland(self, t):
+        return (1 + self.tau * t / self.c) * tt.clip(1 - t / self.c, 0, np.inf) ** self.tau
+
+    def full(self, X, Xs=None):
+        X, Xs = self._slice(X, Xs)
+        return self.weinland(self.dist(X, Xs))
+
+    def diag(self, X):
+        return tt.alloc(1.0, X.shape[0])
+
+
 class Stationary(Covariance):
     r"""
     Base class for stationary kernels/covariance functions.

pymc3/tests/test_gp.py

@@ -1189,3 +1189,31 @@ def test_plot_gp_dist_warn_nan(self):
             pm.gp.util.plot_gp_dist(ax, x=np.linspace(0, 50, X), samples=samples)
         plt.close()
         pass
+
+
+class TestCircular:
+    def test_1d_tau1(self):
+        X = np.linspace(0, 1, 10)[:, None]
+        etalon = 0.600881
+        with pm.Model():
+            cov = pm.gp.cov.Circular(1, 1, tau=5)
+        K = theano.function([], cov(X))()
+        npt.assert_allclose(K[0, 1], etalon, atol=1e-3)
+        K = theano.function([], cov(X, X))()
+        npt.assert_allclose(K[0, 1], etalon, atol=1e-3)
+        # check diagonal
+        Kd = theano.function([], cov(X, diag=True))()
+        npt.assert_allclose(np.diag(K), Kd, atol=1e-5)
+
+    def test_1d_tau2(self):
+        X = np.linspace(0, 1, 10)[:, None]
+        etalon = 0.691239
+        with pm.Model():
+            cov = pm.gp.cov.Circular(1, 1, tau=4)
+        K = theano.function([], cov(X))()
+        npt.assert_allclose(K[0, 1], etalon, atol=1e-3)
+        K = theano.function([], cov(X, X))()
+        npt.assert_allclose(K[0, 1], etalon, atol=1e-3)
+        # check diagonal
+        Kd = theano.function([], cov(X, diag=True))()
+        npt.assert_allclose(np.diag(K), Kd, atol=1e-5)