poissonprocess_inhomogeneus

En esta página

poissonprocess_inhomogeneus#

Generates points according to an inhomogeneous Poisson point process in a square region using the thinning algorithm. An inhomogeneous Poisson point process is characterized by an intensity function lambda(x,y) that varies spatially, determining the likelihood of finding points at different locations.

Uses the thinning (acceptance-rejection) algorithm:

  1. Find Maximum Intensity: Numerically minimize -lambda(x,y) to find lambda_max = max lambda(x,y) over the simulation window.

  2. Generate Homogeneous Proposal Points: Generate N_prop ~ Poisson(lambda_max × area) proposal points uniformly in [0, limit]².

  3. Thinning: For each proposal point (x_i, y_i), keep it with probability p(x_i, y_i) = lambda(x_i, y_i) / lambda_max by generating u_i ~ U(0,1) and keeping the point if u_i < p(x_i, y_i).

The resulting retained points follow the inhomogeneous Poisson process defined by lambda(x,y).

Parameters#

  • fun_lambda (callable): A function taking two arguments (x, y) and returning the intensity at that point. Example: lambda x, y: x + y

  • n_sim (int): This parameter is currently unused in the implementation.

  • limit (float): The side length of the square simulation window [0, limit]².

  • seed (int, optional): A seed for the random number generator. If None, uses entropy from the OS.

Returns#

  • SetPoints: Instance with points following the inhomogeneous Poisson process.

Example#

import proximitygraphs as pg

pts = pg.SetPoints.poissonprocess_inhomogeneus(fun_lambda=lambda x, y: x + y, limit=1, seed=7)

pts.draw(figsize=(8, 8), v_color='#8c564b')

Example point set

En esta página