The \(K\)-process on a tree as a scaling limit of the GREM-like trap model (Q2448697)

Trap models on a finite volume \(k\)-level tree are introduced as a class of Markov processes whose state space is constituted of leaves of the tree. By extending the previous results on the random energy model (REM) to its generalized version (GREM), the \(p\)-spin dynamics is addressed. Since rigorous results for GREM are scarce, the present paper addresses in some detail a specific GREM-like trap model introduced by \textit{M. Sasaki} and \textit{K. Nemoto} [``Analysis on aging in the generalized random energy model'', J. Phys. Soc. Jpn. 69, 3045--3050 (2000), \url{arxiv:cond-mat/0003484}]. Under suitable conditions on the trap model, an infinite volume limit is established, which is named the \(K\)-process on an infinite \(k\)-level tree. Next, this process is described in the scaling limit of the GREM trap model on extreme time scales, after first sending the volume to infinity in a fine tuned sequence of steps. The latter procedure is the main outcome of the paper, since the original description of the trap model on a tree, at first glance, does not seem to validate the infinite volume limit at all.