A limit theorem for moments in space of the increments of Brownian local time (Q2012244)

If \((L_t^x)\) is the local time of Brownian motion, \textit{X. Chen} et al. [Ann. Probab. 38, No. 1, 396--438 (2010; Zbl 1190.60071)] proved that \[ \frac{1}{h^{3/2}}\left(\int_{-\infty}^{\infty}(L_t^ {x+h}-L_t^x)^2 dx - 4 ht \right) \] converges in distribution to \(\sqrt{\frac{64}{3}\int_{-\infty}^{\infty}(L_t^x)^2dxZ},\) where \(Z\) is a standard Gaussian random variable, independent of \((L_t^x)_{x \in \mathcal{R}}.\) The proof of this was through the method of moments, and subsequently two different proofs were given: one by \textit{J. Rosen} [Lect. Notes Math. 2006, 95--104 (2011; Zbl 1219.60028)] using stochastic calculus and Brownian self-intersection local times and the other by \textit{Y. Hu} and \textit{D. Nualart} [Electron. Commun. Probab. 14, 529--539 (2009; Zbl 1193.60074)] using Malliavin calculus and Pitman and Yor's asymptotic version of the Ray-Knight theorem. The author of the article under review takes a fundamentally different approach to exclusively work in the space variable \(x\) instead of approaching the problem through the time domain \(t\) and states and proves a result for arbitrary integer powers in place of the power 2 in the cited result. This generalizes the two proofs and also is amenable for further generalization as the time \(t\) can be replaced with a suitable stopping time \(\tau.\) The result also proves a conjecture of Rosen. The author's proof starts with Perkin's semimartingale decomposition of Brownian local time in space and uses an iterative application of the Kailath-Segall formula.