On 07/06/17 09:57, Fernanda De Bastiani wrote:
Dear Sir/Madam, I am trying to use function kppm() of spatstat package with anisotropic covariance option as the following: #------------------------------------------- kppm(redwood, ~x, "LGCP", statistic="pcf", covmodel=list(model="matern", nu=0.3, Aniso = matrix(nc=2, c(1.5, 3, -3, 4)))) #------------------------------------------- but it gives me an error message: Error in rfInit(model = p, x = x, y = y, z = z, T = T, grid = grid, distances = distances, : 'RFcov' : 'Aniso' : not of the required size: (2, 2) instead of (undetermined, 1) Please, could anyone help me to understand in which conditions nc=2 will work?
The kppm() function is designed for models whose pair correlation function is isotropic. It cannot be used to fit an anisotropic model. (Extending this to anisotropic models is conceivable but it would require new theory as well as completely new software). I?m sorry about the opaque error message. I?ll try to find a way to detect anisotropic models and give a more meaningful message from the spatstat code. Adrian Baddeley