Martyn Amos

Enhancing data parallelism for Ant Colony Optimization on GPUs

Graphics Processing Units (GPUs) have evolved into highly parallel and fully programmable architecture over the past five years, and the advent of CUDA has facilitated their application to many real-world applications. In this paper, we deal with a GPU implementation of Ant Colony Optimization (ACO), a population-based optimization method which comprises two major stages: tour construction and pheromone update. Because of its inherently parallel nature, ACO is well-suited to GPU implementation, but it also poses significant challenges due to irregular memory access patterns. Our contribution within this context is threefold: (1) a data parallelism scheme for tour construction tailored to GPUs, (2) novel GPU programming strategies for the pheromone update stage, and (3) a new mechanism called I-Roulette to replicate the classic roulette wheel while improving GPU parallelism. Our implementation leads to factor gains exceeding 20x for any of the two stages of the ACO algorithm as applied to the TSP when compared to its sequential counterpart version running on a similar single-threaded high-end CPU. Moreover, an extensive discussion focused on different implementation paths on GPUs shows the way to deal with parallel graph connected components. This, in turn, suggests a broader area of inquiry, where algorithm designers may learn to adapt similar optimization methods to GPU architecture.

Topics in the theory of DNA computing

DNA computing, or, more generally, molecular computing, is an exciting fast developing interdisciplinary area. Research in this area concerns theory, experiments, and applications of DNA computing. In this paper, we demonstrate the theoretical developments by discussing a number of selected topics. We also give an introduction to the basic structure of DNA and the basic DNA processing tools.

Parallelization strategies for ant colony optimisation on GPUs

Ant Colony Optimisation (ACO) is an effective population-based meta-heuristic for the solution of a wide variety of problems. As a population-based algorithm, its computation is intrinsically massively parallel, and it is therefore theoretically well-suited for implementation on Graphics Processing Units (GPUs). The ACO algorithm comprises two main stages: textit{Tour construction} and textit{Pheromone update}. The former has been previously implemented on the GPU, using a task-based parallelism approach. However, up until now, the latter has always been implemented on the CPU. In this paper, we discuss several parallelisation strategies for {it both} stages of the ACO algorithm on the GPU. We propose an alternative {it data-based} parallelism scheme for textit{Tour construction}, which fits better on the GPU architecture. We also describe novel GPU programming strategies for the textit{Pheromone update} stage. Our results show a total speed-up exceeding 28x for the textit{Tour construction} stage, and 20x for textit{Pheromone update}, and suggest that ACO is a potentially fruitful area for future research in the GPU domain.

Multicellular Computing Using Conjugation for Wiring

Recent efforts in synthetic biology have focussed on the implementation of logical functions within living cells. One aim is to facilitate both internal “re-programming” and external control of cells, with potential applications in a wide range of domains. However, fundamental limitations on the degree to which single cells may be re-engineered have led to a growth of interest in multicellular systems, in which a “computation” is distributed over a number of different cell types, in a manner analogous to modern computer networks. Within this model, individual cell type perform specific sub-tasks, the results of which are then communicated to other cell types for further processing. The manner in which outputs are communicated is therefore of great significance to the overall success of such a scheme. Previous experiments in distributed cellular computation have used global communication schemes, such as quorum sensing (QS), to implement the “wiring” between cell types. While useful, this method lacks specificity, and limits the amount of information that may be transferred at any one time. We propose an alternative scheme, based on specific cell-cell conjugation. This mechanism allows for the direct transfer of genetic information between bacteria, via circular DNA strands known as plasmids. We design a multi-cellular population that is able to compute, in a distributed fashion, a Boolean XOR function. Through this, we describe a general scheme for distributed logic that works by mixing different strains in a single population; this constitutes an important advantage of our novel approach. Importantly, the amount of genetic information exchanged through conjugation is significantly higher than the amount possible through QS-based communication. We provide full computational modelling and simulation results, using deterministic, stochastic and spatially-explicit methods. These simulations explore the behaviour of one possible conjugation-wired cellular computing system under different conditions, and provide baseline information for future laboratory implementations.

A reconfigurable NAND/NOR genetic logic gate

BackgroundEngineering genetic Boolean logic circuits is a major research theme of synthetic biology. By altering or introducing connections between genetic components, novel regulatory networks are built in order to mimic the behaviour of electronic devices such as logic gates. While electronics is a highly standardized science, genetic logic is still in its infancy, with few agreed standards. In this paper we focus on the interpretation of logical values in terms of molecular concentrations.ResultsWe describe the results of computational investigations of a novel circuit that is able to trigger specific differential responses depending on the input standard used. The circuit can therefore be dynamically reconfigured (without modification) to serve as both a NAND/NOR logic gate. This multi-functional behaviour is achieved by a) varying the meanings of inputs, and b) using branch predictions (as in computer science) to display a constrained output. A thorough computational study is performed, which provides valuable insights for the future laboratory validation. The simulations focus on both single-cell and population behaviours. The latter give particular insights into the spatial behaviour of our engineered cells on a surface with a non-homogeneous distribution of inputs.ConclusionsWe present a dynamically-reconfigurable NAND/NOR genetic logic circuit that can be switched between modes of operation via a simple shift in input signal concentration. The circuit addresses important issues in genetic logic that will have significance for more complex synthetic biology applications.

Dynamic load balancing on heterogeneous clusters for parallel ant colony optimization

Ant colony optimisation (ACO) is a nature-inspired, population-based metaheuristic that has been used to solve a wide variety of computationally hard problems. In order to take full advantage of the inherently stochastic and distributed nature of the method, we describe a parallelization strategy that leverages these features on heterogeneous and large-scale, massively-parallel hardware systems. Our approach balances workload effectively, by dynamically assigning jobs to heterogeneous resources which then run ACO implementations using different search strategies. Our experimental results confirm that we can obtain significant improvements in terms of both solution quality and energy expenditure, thus opening up new possibilities for the development of metaheuristic-based solutions to “real world” problems on high-performance, energy-efficient contemporary heterogeneous computing platforms.

A novel genetic algorithm for the layout optimization problem

In this paper we present a new algorithm for the Layout Optimization Problem: this concerns the placement of circular, weighted objects inside a circular container, the two objectives being to minimize imbalance of mass and to minimize the radius of the container. This problem carries real practical significance in industrial applications (such as the design of satellites), as well as being of significant theoretical interest. We present a genetic algorithm solution and compare it with two existing nature-inspired methods, one of which is the best published algorithm for this problem. Experimental results show that our approach out-performs these existing methods in terms of both solution quality and execution time.

Enhancing GPU parallelism in nature-inspired algorithms

We present GPU implementations of two different nature-inspired optimization methods for well-known optimization problems. Ant Colony Optimization (ACO) is a two-stage population-based method modelled on the foraging behaviour of ants, while P systems provide a high-level computational modelling framework that combines the structure and dynamic aspects of biological systems (in particular, their parallel and non-deterministic nature). Our methods focus on exploiting data parallelism and memory hierarchy to obtain GPU factor gains surpassing 20x for any of the two stages of the ACO algorithm, and 16x for P systems when compared to sequential versions running on a single-threaded high-end CPU. Additionally, we compare performance between GPU generations to validate hardware enhancements introduced by Nvidia’s Fermi architecture.

Biological and Chemical Information Technologies

Biological and chemical information technologies (bio/chem IT) have the potential to reshape the scientific and technological landscape. In this paper we briefly review the main challenges and opportunities in the field, before presenting several case studies based on ongoing FP7 research projects.

Continuous computation in engineered gene circuits

In this paper we consider the problem of representation and measurement in genetic circuits, and investigate how they can affect the reliability of engineered systems. We propose a design scheme, based on the notion of continuous computation, which addresses these issues. We illustrate the methodology by showing how a concept from computer architecture (namely, branch prediction) may be implemented in vivo, using a distributed approach. Simulation results confirm the in-principle feasibility of our method, and offer valuable insights into its future laboratory validation.