Natural Computing最新文献

Recent research into analog computing has introduced new notions of computing real numbers. Huang, Klinge, Lathrop, Li, and Lutz defined a notion of computing real numbers in real-time with chemical reaction networks (CRNs), introducing the classes (mathbb {R}_text {LCRN}) (the class of all Lyapunov CRN-computable real numbers) and (mathbb {R}_text {RTCRN}) (the class of all real-time CRN-computable numbers). In their paper, they show the inclusion of the real algebraic numbers (text { ALG} subseteq mathbb {R}_text {LCRN}subseteq mathbb {R}_text {RTCRN}) and that (text { ALG} subsetneqq mathbb {R}_text {RTCRN}) but leave open whether the inclusion is proper. In this paper, we resolve this open problem and show that ({ ALG} = mathbb {R}_text {LCRN}) and, as a consequence, (mathbb {R}_text {LCRN}subsetneqq mathbb {R}_text {RTCRN}). However, the definition of real-time computation by Huang et al. is fragile in the sense that it is sensitive to perturbations in initial conditions. To resolve this flaw, we further require a CRN to withstand these perturbations. In doing so, we arrive at a discrete model of memory. This approach has several benefits. First, a bounded CRN may compute values approximately in finite time. Second, a CRN can tolerate small perturbations of its species’ concentrations. Third, taking a measurement of a CRN’s state only requires precision proportional to the exactness of these approximations. Lastly, if a CRN requires only finite memory, this model and Turing machines are equivalent under real-time simulations.

Due to the compensating function of neutral grounded arc suppression coil, fault line selection in distribution network is still facing challenges: the classical models have insufficient learning ability in extracting fault features, and there is an imbalance in the original data used, resulting in low accuracy in fault line selection. In order to address this issue, this paper proposes a novel variant of spiking neural P (SNP) systems called integrated dynamic SNP (IDSNP) systems, which consist of gated neurons, rule neurons, and weighed neurons with different designed rules. Furthermore, according to the IDSNP systems, an IDSNP(FL) model is developed for fault line selection in distribution network, where the number of layers for transmitting weighted neuron spiking information could be dynamically changeable depending on the complexity of the number of stations in the power system. Finally, the proposed model is evaluated on a real-time dispatch dataset of a real power system. The experimental results show that the IDSNP(FL) model achieves the best performance compared with several classical models in deep learning, verifying the effectiveness of the proposed model for fault line selection tasks in distribution network.

Reaction systems are a formal model for computational processing in which reactions operate on sets of entities (molecules) providing a framework for dealing with qualitative aspects of biochemical systems. This paper is concerned with reaction systems in which entities can have discrete concentrations, and so reactions operate on multisets rather than sets of entities. The resulting framework allows one to deal with quantitative aspects of reaction systems, and a bespoke linear-time temporal logic allows one to express and verify a wide range of key behavioural system properties. In practical applications, a reaction system with discrete concentrations may only be partially specified, and the possibility of an effective automated calculation of the missing details provides an attractive design approach. With this idea in mind, the current paper discusses parametric reaction systems with parameters representing unknown parts of hypothetical reactions. The main result is a method aimed at replacing the parameters in such a way that the resulting reaction system operating in a specified external environment satisfies a given temporal logic formula.This paper provides an encoding of parametric reaction systems in smt, and outlines a synthesis procedure based on bounded model checking for solving the synthesis problem. It also reports on the initial experimental results demonstrating the feasibility of the novel synthesis method.

This paper forges a strong connection between two well known computational frameworks for representing biological systems, in order to facilitate the seamless transfer of techniques between them. Boolean networks are a well established formalism employed from biologists. They have been studied under different (synchronous and asynchronous) update semantics, enabling the observation and characterisation of distinct facets of system behaviour. Recently, a new semantics for Boolean networks has been proposed, called most permissive semantics, that enables a more faithful representation of biological phenomena. Reaction systems offer a streamlined formalism inspired by biochemical reactions in living cells. Reaction systems support a full range of analysis techniques that can help for gaining deeper insights into the underlying biological phenomena. Our goal is to leverage the available toolkit for predicting and comprehending the behaviour of reaction systems within the realm of Boolean networks. In this paper, we first extend the behaviour of reaction systems to several asynchronous semantics, including the most permissive one, and then we demonstrate that Boolean networks and reaction systems exhibit isomorphic behaviours under the synchronous, general/fully asynchronous and most permissive semantics.

Chemical systems have the potential to direct the next generation of dynamic materials if they can be integrated with a material while acting as the material’s own regulatory network. Chemical networks that use DNA and RNA strand displacement coupled with RNA synthesis and degradation, such as genelets, are promising chemical systems for this role. Genelets can produce a range of dynamic behaviors that respond to unique sets of environmental inputs. While a number of networks that generate specific types of outputs which vary in both time and amplitude have been developed, there are fewer examples of networks that recognize specific types of inputs in time and amplitude. Advanced chemical circuits in biology are capable of reading a given substrate concentration with relatively high accuracy to direct downstream function, demonstrating that such a chemical circuit is possible. Taking inspiration from this, we designed a genelet circuit which responds to a range of inputs by delivering a binary output based on the input concentration, and tested the network’s performance using an in silico model of circuit behavior. By modifying the concentrations of two circuit elements, we demonstrated that such a network topography could yield various target input concentration profiles to which a given circuit is sensitive. The number of unique elements in the final network topography as well as the individual circuit element concentrations are commensurate with properties of circuits that have been demonstrated experimentally. These factors suggest that such a network could be built and characterized in the laboratory.

This work introduces the new class of pure reaction automata, as well as a new update manner, called maximal reactive manner, that can also be applied to standard reaction automata. Pure reaction automata differ from the standard model in that they don’t have permanence: the entities that are not consumed by the reactions happening at a certain state are not conserved in the result states. We prove that the set of languages accepted by the new class under the maximal reactive manner contains the set of languages accepted by standard reaction automata under the same manner or under the maximal parallel manner. We also prove that a strict subclass of pure reaction automata can compute any partial recursive function.

Reaction systems (RSs) are a computational framework inspired by biochemical mechanisms. A RS defines a finite set of reactions over a finite set of entities. Typically each reaction has a local scope, because it is concerned with a small set of entities, but complex models can involve a large number of reactions and entities, and their computation can manifest unforeseen emerging behaviours. When a deviation is detected, like the unexpected production of some entities, it is often difficult to establish its causes, e.g., which entities were directly responsible or if some reaction was misconceived. Slicing is a well-known technique for debugging, which can point out the program lines containing the faulty code. In this paper, we define the first dynamic slicer for RSs and show that it can help to detect the causes of erroneous behaviour and highlight the involved reactions for a closer inspection. To fully automate the debugging process, we propose to distil monitors for starting the slicing whenever a violation from a safety specification is detected. We have integrated our slicer in BioResolve, written in Prolog which provides many useful features for the formal analysis of RSs. We define the slicing algorithm for basic RSs and then enhance it for dealing with quantitative extensions of RSs, where timed processes and linear processes can be represented. Our framework is shown at work on suitable biologically inspired RS models.

In this paper, we investigate shape-assembling power of a tile-based model of self-assembly called the Signal-Passing Tile Assembly Model (STAM). In this model, the glues that bind tiles together can be turned on and off by the binding actions of other glues via “signals”. Specifically, the problem we investigate is “shape replication” wherein, given a set of input assemblies of arbitrary shape, a system must construct an arbitrary number of assemblies with the same shapes and, with the exception of size-bounded junk assemblies that result from the process, no others. We provide the first fully universal shape replication result, namely a single tile set capable of performing shape replication on arbitrary sets of any 3-dimensional shapes without requiring any scaling or pre-encoded information in the input assemblies. Our result requires the input assemblies to be composed of signal-passing tiles whose glues can be deactivated to allow deconstruction of those assemblies, which we also prove is necessary by showing that there are shapes whose geometry cannot be replicated without deconstruction. Additionally, we modularize our construction to create systems capable of creating binary encodings of arbitrary shapes, and building arbitrary shapes from their encodings. Because the STAM is capable of universal computation, this then allows for arbitrary programs to be run within an STAM system, using the shape encodings as input, so that any computable transformation can be performed on the shapes. This is the full version, containing all construction and proof details, of a previously published extended abstract version that had most details omitted.

Chemical reaction networks (CRNs) are an important tool for molecular programming. This field is rapidly expanding our ability to deploy computer programs into biological systems for various applications. However, CRNs are also difficult to work with due to their massively parallel nature, leading to the need for higher-level languages that allow for more straightforward computation with CRNs. Recently, research has been conducted into various higher-level languages for deterministic CRNs but modeling CRN parallelism, managing error accumulation, and finding natural CRN representations are ongoing challenges. We introduce Reactamole, a higher-level language for deterministic CRNs that utilizes the functional reactive programming (FRP) paradigm to represent CRNs as a reactive dataflow network. Reactamole equates a CRN with a functional reactive program, implementing the key primitives of the FRP paradigm directly as CRNs. The functional nature of Reactamole makes reasoning about molecular programs easier, and its strong static typing allows us to ensure that a CRN is well-formed by virtue of being well-typed. In this paper, we describe the design of Reactamole and how we use CRNs to represent the common datatypes and operations found in FRP. We demonstrate the potential of this functional reactive approach to molecular programming by giving an extended example where a CRN is constructed using FRP to modulate and demodulate an amplitude-modulated signal. We also show how Reactamole can be used to specify abstract CRNs whose structure depends on the reactions and species of its input, allowing users to specify more general CRN behaviors.