← All papers
First page of Weak Convergence and Gaussian Limits for a General-Dimensional Origin-Invariant Cramér–von Mises Statistic

Weak Convergence and Gaussian Limits for a General-Dimensional Origin-Invariant Cramér–von Mises Statistic

Marco Mandap

stat.ME May 18, 2026 · v3 math.ST
Lean 4 was used to formally verify the proofs of the statistic's computing formulas and asymptotic results, building on Lean empirical-process work.
We introduce a general-dimensional origin-invariant Cramér–von Mises statistic $\barω_d^2$ for testing complete spatial randomness, defined by averaging corner-oriented empirical-distribution-function discrepancies over all $2^d$ corners of $[0,1]^d$. The statistic has a closed-form $O(n^2)$ computing formula. For $d=1$ it reduces to the classical rank-based Cramér–von Mises statistic, while for $d=2$ its computing formula agrees with Zimmerman's origin-invariant statistic. Under iid complete spatial randomness, the associated empirical process converges in $\ell^\infty$ to a centered Gaussian process with an explicit cross-corner covariance kernel. The continuous mapping theorem yields a quadratic Gaussian limit governed by a positive trace-class covariance operator with trace $2^{-d}-3^{-d}$. We also connect fixed-count complete spatial randomness with the homogeneous Poisson formulation. For strictly stationary alpha-mixing sequences with uniform marginals, we establish covariance summability, the long-run variance limit, and a finite-dimensional all-corners Gaussian limit. Monte Carlo experiments illustrate null calibration and sensitivity to selected alternatives. The statistic is consistent against fixed iid alternatives whose distribution functions differ from uniformity on a set of positive measure.

Multivariate Cramér–von Mises statistics for testing complete spatial randomness depend on which corner of the domain is taken as the origin. Zimmerman's origin-invariant statistic covered only d=2, and a general-dimensional version with established asymptotic theory was missing.

The statistic is defined by averaging corner-oriented EDF discrepancies over all 2^d corners of [0,1]^d, and a closed-form O(n^2) computing formula is derived via Fubini reductions. Weak convergence of the corner-indexed empirical process is established using VC-class arguments, and the quadratic limit follows from the continuous mapping theorem. The paper also treats alpha-mixing dependent sequences and the link to the homogeneous Poisson formulation. The author reports that Lean 4 was used for formal proof verification.

The statistic reduces to the classical CvM statistic at d=1 and agrees with Zimmerman's formula at d=2. Under iid CSR it converges to a weighted chi-square law governed by a trace-class covariance operator with trace 2^{-d}-3^{-d}. Monte Carlo and real-data examples illustrate null calibration; on the biological-cells pattern the observed value lies below all 50,000 null simulations, giving a corrected lower-tail p-value of 0.00002.

Datasetnn·ω̄²₂p_upperp_lower
Swedish pines710.10070.65210.3479
Redwood seedlings620.21410.13320.8668
Biological cells420.01981.00000.00002
Observed statistics and Monte Carlo p-values (B=50,000) for spatial datasets