af286ba5280614ecc8371b71ed53cc6447d1183b,GPy/kern/src/static.py,Precomputed,Kdiag,#Precomputed#Any#,237

Before Change

            return self.variance * self.fixed_K[X[:,0].astype("int")][:,X2[:,0].astype("int")]

    def Kdiag(self, X):
        return self.variance * self.fixed_K[X[:,0].astype("int")][:,X[:,0].astype("int")].diagonal()

    def update_gradients_full(self, dL_dK, X, X2=None):
        if X2 is None:

After Change

        return self.variance * self._index(X, X2)

    def Kdiag(self, X):
        return self.variance * self._index(X,None).diagonal()

    def update_gradients_full(self, dL_dK, X, X2=None):
        self.variance.gradient = np.einsum("ij,ij", dL_dK, self._index(X, X2))

Instances