Ultrametric Diffusion, Rugged Energy Landscapes and Transition Networks

TL;DR

This paper introduces ultrametric networks as p-adic analogues of Markov models for complex hierarchical energy landscapes, providing explicit solutions and revealing fast transition modes that act as kinetic hubs.

Contribution

It develops a novel p-adic framework for Markov processes on hierarchical energy landscapes, explicitly solving the master equation and identifying fast transition modes.

Findings

Explicit solution to the master equation for ultrametric networks

Identification of fast transition modes as kinetic hubs

Long-term behavior governed by Markov chains or absorbing states

Abstract

In this article we introduce the ultrametric networks which are p-adic continuous analogues of the standard Markov state models constructed using master equations. A p-adic transition network (or an ultrametric network) is a model of a complex system consisting of a hierarchical energy landscape, a Markov process on the energy landscape, and a master equation. We focus on networks where the transition rates between two different basins are constant functions, and the jumping process inside of each basin is controlled by a p-adic radial function. We solve explicitly the Cauchy problem for the master equation attached to this type of networks. The solution of this problem is the network response to a given initial concentration. If the Markov process attached to the network is conservative, the long term response of the network is controlled by a Markov chain. If the process is not conservative the network has absorbing states. We define an absorbing time, which depends on the initial concentration, if this time is finite the network reaches an absorbing state in a finite time. We identify in the response of the network the terms responsible for bringing the network to an absorbing state, we call them the fast transition modes. The existence of the fast transition modes is a consequence of the assumption that the energy landscape is ultrametric (hierarchical), and to the best of our understanding this result cannot be obtained using standard methods of Markov state models. Nowadays, it is widely accepted that protein native states are kinetic hubs that can be reached quickly from any other state. The existence of fast transition modes implies that certain states on an ultrametric network work as kinetic hubs.