-
On the maxmin-$ω$ eigenspaces and their over-approximation by zones
Authors:
Muhammad Syifa'ul Mufid,
Ebrahim Patel,
Sergei Sergeev
Abstract:
Maxmin-$ω$ dynamical systems were previously introduced as a generalization of dynamical systems expressed by tropical linear algebra. To describe steady states of such systems one has to study an eigenproblem of the form $A\otimes_ω x=λ+x$ where $\otimes_ω$ is the maxmin-$ω$ matrix-vector multiplication. This eigenproblem can be viewed in more general framework of nonlinear Perron-Frobenius theor…
▽ More
Maxmin-$ω$ dynamical systems were previously introduced as a generalization of dynamical systems expressed by tropical linear algebra. To describe steady states of such systems one has to study an eigenproblem of the form $A\otimes_ω x=λ+x$ where $\otimes_ω$ is the maxmin-$ω$ matrix-vector multiplication. This eigenproblem can be viewed in more general framework of nonlinear Perron-Frobenius theory. However, instead of studying such eigenspaces directly we develop a different approach: over-approximation by zones. These are traditionally convex sets of special kind which proved to be highly useful in computer science and also relevant in tropical convexity. We first construct a sequence of zones over-approximating a maxmin-$ω$ eigenspace. Next, the limit of this sequence is refined in a heuristic procedure, which yields a refined zone and also the eigenvalue $λ$ with a high success rate. Based on the numerical experiments, in successful cases there is a column of the difference bound matrix (DBM) representation of the refined zone which yields an eigenvector.
△ Less
Submitted 23 July, 2024;
originally announced July 2024.
-
Solving Linear Equations Over Maxmin-$ω$ Systems
Authors:
Muhammad Syifa'ul Mufid,
Ebrahim Patel,
Sergei Sergeev
Abstract:
Maxmin-$ω$ dynamical systems were previously introduced as an ``all-in-one package'' that can yield a solely min-plus, a solely max-plus, or a max-min-plus dynamical system by varying a parameter $ω\in(0,1]$. With such systems in mind, it is natural to introduce and consider maxmin-$ω$ linear systems of equations of the type $A\otimes_ω x=b$. However, to our knowledge, such maxmin-$ω$ linear syste…
▽ More
Maxmin-$ω$ dynamical systems were previously introduced as an ``all-in-one package'' that can yield a solely min-plus, a solely max-plus, or a max-min-plus dynamical system by varying a parameter $ω\in(0,1]$. With such systems in mind, it is natural to introduce and consider maxmin-$ω$ linear systems of equations of the type $A\otimes_ω x=b$. However, to our knowledge, such maxmin-$ω$ linear systems have not been studied before and in this paper we present an approach to solve them. We show that the problem can be simplified by performing normalization and then generating a ``canonical'' matrix which we call the principal order matrix. Instead of directly trying to find the solutions, we search the possible solution indices which can be identified using the principal order matrix and the parameter $ω$. The fully active solutions are then immediately obtained from these solution indices. With the fully active solutions at hand, we then present the method to find other solutions by applying a relaxation, i.e., increasing or decreasing some components of fully active solutions. This approach can be seen as a generalization of an approach that could be applied to solve max-plus or min-plus linear systems. Our results also shed more light on an unusual feature of maxmin-$ω$ linear systems, which, unlike in the usual linear algebra, can have a finite number of solutions in the case where their solution is non-unique.
△ Less
Submitted 1 August, 2023;
originally announced August 2023.
-
A new threshold model based on tropical mathematics reveals network backbones
Authors:
Ebrahim L. Patel
Abstract:
Maxmin-$ω$ is a new threshold model, where each node in a network waits for the arrival of states from a fraction $ω$ of neighborhood nodes before processing its own state, and subsequently transmitting it to downstream nodes. Repeating this sequence of events leads to periodic behavior. We show that maxmin-$ω$ reduces to a smaller, simpler, system represented by tropical mathematics, which forges…
▽ More
Maxmin-$ω$ is a new threshold model, where each node in a network waits for the arrival of states from a fraction $ω$ of neighborhood nodes before processing its own state, and subsequently transmitting it to downstream nodes. Repeating this sequence of events leads to periodic behavior. We show that maxmin-$ω$ reduces to a smaller, simpler, system represented by tropical mathematics, which forges a useful link between the nodal state update times and circuits in the network. Thus, the behavior of the system can be analyzed directly from the smaller network structure and is computationally faster. We further show that these reduced networks: (i) are not unique; they are dependent on the initialisation time, (ii) tend towards periodic orbits of networks -- "attractor networks." In light of these features, we vary the initial condition and $ω$ to obtain statistics on types of attractor networks. The results suggest the case $ω=0.5$ to give the most stable system. We propose that the most prominent attractor networks of nodes and edges may be seen as a 'backbone' of the original network. We subsequently provide an algorithm to construct attractor networks, the main result being that they must necessarily contain circuits that are deemed critical and that can be identified without running the maxmin-$ω$ system. Finally, we apply this work to the C. elegans neuron network, giving insight into its function and similar networks. This novel area of application for tropical mathematics adds to its theory, and provides a new way to identify important edges and circuits of a network.
△ Less
Submitted 22 September, 2021;
originally announced September 2021.
-
A Max-Plus Model of Asynchronous Cellular Automata
Authors:
Ebrahim L. Patel,
David Broomhead
Abstract:
This paper presents a new framework for asynchrony. This has its origins in our attempts to better harness the internal decision making process of cellular automata (CA). Thus, we show that a max-plus algebraic model of asynchrony arises naturally from the CA requirement that a cell receives the state of each neighbour before updating. The significant result is the existence of a bijective mapping…
▽ More
This paper presents a new framework for asynchrony. This has its origins in our attempts to better harness the internal decision making process of cellular automata (CA). Thus, we show that a max-plus algebraic model of asynchrony arises naturally from the CA requirement that a cell receives the state of each neighbour before updating. The significant result is the existence of a bijective mapping between the asynchronous system and the synchronous system classically used to update cellular automata. Consequently, although the CA outputs look qualitatively different, when surveyed on "contours" of real time, the asynchronous CA replicates the synchronous CA. Moreover, this type of asynchrony is simple - it is characterised by the underlying network structure of the cells, and long-term behaviour is deterministic and periodic due to the linearity of max-plus algebra. The findings lead us to proffer max-plus algebra as: (i) a more accurate and efficient underlying timing mechanism for models of patterns seen in nature, and (ii) a foundation for promising extensions and applications.
△ Less
Submitted 13 February, 2015;
originally announced February 2015.
