bayespecon.logdet.logdet_interpolated¶

bayespecon.logdet.logdet_interpolated(rho, W_dense, rho_min=-1.0, rho_max=1.0, n_grid=200)[source]¶

Cubic spline interpolation of log|I - rho*W|.

Pre-computes values on a rho grid at construction time and evaluates a cubic-spline piecewise polynomial symbolically.

Parameters:¶

rho : pytensor scalar¶

Spatial autoregressive parameter symbol.

W_dense : np.ndarray¶

Dense spatial weights matrix.

rho_min : float, default=-1.0¶

Lower bound for rho grid.

rho_max : float, default=1.0¶

Upper bound for rho grid.

n_grid : int, default=200¶

Number of grid points.

Returns:¶

Interpolated symbolic log-determinant value.

Return type:¶

pytensor.tensor.TensorVariable