b9d42d6154f5f98696a67097cc93dac027398d76,GPflow/sgpr.py,SGPR,build_likelihood,#SGPR#,39

Before Change

        L = tf.cholesky(Kuu)

        // Compute intermediate matrices
        A = tf.matrix_triangular_solve(L, Kuf, lower=True)*tf.sqrt(1./self.likelihood.variance)
        AAT = tf.matmul(A, tf.transpose(A))
        B = AAT + eye(num_inducing)
        LB = tf.cholesky(B)
        c = tf.matrix_triangular_solve(LB, tf.matmul(A, err), lower=True) * tf.sqrt(1./self.likelihood.variance)

        //compute log marginal bound
        bound = -0.5*tf.cast(num_data*output_dim, tf.float64)*np.log(2*np.pi)
        bound += -tf.cast(output_dim, tf.float64)*tf.reduce_sum(tf.log(tf.user_ops.get_diag(LB)))

After Change

        Kuf = self.kern.K(self.Z, self.X)
        Kuu = self.kern.K(self.Z) + eye(num_inducing) * 1e-6
        L = tf.cholesky(Kuu)
        sigma = tf.sqrt(self.likelihood.variance)

        // Compute intermediate matrices
        A = tf.matrix_triangular_solve(L, Kuf, lower=True) / sigma
        AAT = tf.matmul(A, tf.transpose(A))
        B = AAT + eye(num_inducing)
        LB = tf.cholesky(B)
        Aerr = tf.matmul(A, err)
        c = tf.matrix_triangular_solve(LB, Aerr, lower=True) / sigma

        // compute log marginal bound
        bound = -0.5*tf.cast(num_data*output_dim, tf.float64)*np.log(2*np.pi)
        bound += -output_dim * tf.reduce_sum(tf.log(tf.user_ops.get_diag(LB)))

Instances