General Purpose Problem Research Articles

Параллельная реализация алгоритма поиска минимальных остовных деревьев с использованием центрального и графического процессоров

Solution of the finding a minimum spanning tree problem is common in various areas of research: recognition of different objects, computer vision, analysis and construction of networks (eg, telephone, electrical, computer, travel, etc.), chemistry and biology, and many others. Processing large graphs is a quite time-consuming task for the central processor (CPU), and in high demand at the present moment. The usage of Graphics processing units (GPUs) as a mean to solve general-purpose problems grows every day, because GPUs have more computing power than CPUs. This article describes the methods of compression and conversion of graphs in standard formats to increase the efficiency of their processing. The search algorithm of minimum spanning trees has been used for researching the proposed approaches. The possibility of a hybrid implementation of this algorithm has been investigated. The highest results were obtained on the large R-MAT and SSCA2 graphs.

Session details: Complexity column

Cook proved in 1971 that SAT is NP complete, thus beginning the amazingly fruitful study of NP completeness. I've been teaching all my career that it is well-believed that SAT is intractable in the worst case. Over the years, however, SAT solvers have improved incredibly; they are now extensively used as general-purpose problem solvers.

Evolution, creativity, intelligence, and madness: "Here Be Dragons".

One of the great joys of being a scien-tist is the hunt for an elusive signal withinthe noise of data, opinions, biases, andother human foibles associated with thepursuit of knowledge. It is inevitable thatthis imperfect quest will result in manyfalse starts along the way when looking“through a glass, darkly.” Our imperfectand incomplete knowledge of the worldmust look like an unpolished mirror,reflecting gibberish, at times. However, italsoreflectsanunderlyingsignalthatbearsfurther scrutiny, in spite of our instinctto discard a flawed image of reality. Thepursuit of the neural underpinnings ofcreative cognition is certainly that “darkglass” we peer into so intently, attempt-ing to grasp, through our meager instru-ments, some hidden truth. Many thinkersand researchers have found that creativityand madness seem somehow to be inter-twined, but the signal is weak, the imageblurry, and the propensity toward roman-tic stereotypes is high. And yet, as scien-tists, we can only follow the data, tryingtomakesenseofwhatittellsus.So,ratherthan entertain the premise outright let metake you on a bit of a journey (which willend back at madness, I promise).First: What if evolutionary processesselected for two types of reasoning?CosmidesandToobyhypothesizeda“ded-icated intelligence” that “refers to theability of a computational system tosolve predefined, target set of prob-lems.” These problems often involve wellestablished rules—like your mundanelife, and Raven’s Matrices problems, andacquiring a language (Pinker, 1991). Theother problems require “improvisationalintelligence” referring to “the ability of acomputational system to improvise solu-tions to novel problems” (Cosmides andTooby, 2002). These problems are moretransient and involve contingencies thatmay or may not persist over time—likefiguring out how to get into your car, hav-ing locked your keys inside. Philosopherscall the former type of problem solving“deductive reasoning”—the observationsnecessarily result in a conclusion beingmade based on the evidence. They are rulebased, deterministic, and the cause leadsnaturallytoeffect.Thelatterproblemsolv-ing is called “abductive reasoning”—thereare an infinite number of possible solu-tionstothemyriadchallengesfacedinthe world; therefore a theory best explainsthe observation, given the evidence. Thisreasoningisprobabilistic,involvesapprox-imation,and(importantly)guessing.Bothmethods are adaptive: one for problemsthat are familiar, the other for problemsthat have never been encountered before.Kanazawa (2004) views intelligence(incorrectly), the pinnacle of deductivereasoning, as THE domain-specific adap-tation to solving novel problems in theenvironment.However,itismycontentionthat intelligence and creativity occupy twoextremes of a dichotomy: intelligence sup-plies a “dedicated reasoning capacity” forproblems that possess rule-based, cause-effect relationships. Others have coveredwell, and provide empirical support for,the “general purpose problem solving”capacity of intelligence and “g” (Kaufmanet al., 2011): I am merely saying here thatthe mechanism is rather “dedicated” tocause-effectrelationships—acapacitywithbroad applicability to deductive reason-ing tasks. In contrast, creativity emergedas an adaptive cognitive mechanism forlow frequency, “improvisational reason-ing,” where solutions to problems areunsighted (Simonton, 2013), and proba-bilistic approximation could lead to novelsolutions. Creative reasoning solves theminority of problems that are unforeseenand yet of high adaptability: “The light-ning has struck the tree near the campand set it on fire. The fire is now spread-ing to the dry underbrush. What should Ido?” (Kanazawa, 2004). In this conceptu-alization, creativity is an evolved cognitivemechanism to abstract, to synthesize, tosolve non-recurrent problems in the envi-ronment. Finally, intelligence should beseen as a rather stable evolved mechanismover the last 1.6 million years (i.e., thesingular“innovation”beingtheAcheuleanhand ax), while creativity appears to haveappeared, in humans at least, in thelast ∼30,000 years (Gabora and Kaufman,2010). Intelligence may not be evolution-arily novel, but creativity certainly is.Perhaps the most parsimonious the-ory of creative cognition to incorporateevolutionary principles is that of BlindVariation and Selective Retention (BVSR)(Campbell, 1960). Indeed, his the-ory posits that creativity in humans“represent(s) cumulated inductiveachievements, stage by stage expansionsof knowledge beyond what could havebeen

A Neural Network: Family Competition Genetic Algorithm and Its Applications in Electromagnetic Optimization

This study proposes a neural network-family competition genetic algorithm (NN-FCGA) for solving the electromagnetic (EM) optimization and other general-purpose optimization problems. The NN-FCGA is a hybrid evolutionary-based algorithm, combining the good approximation performance of neural network (NN) and the robust and effective optimum search ability of the family competition genetic algorithms (FCGA) to accelerate the optimization process. In this study, the NN-FCGA is used to extract a set of optimal design parameters for two representative design examples: the multiple section low-pass filter and the polygonal electromagnetic absorber. Our results demonstrate that the optimal electromagnetic properties given by the NN-FCGA are comparable to those of the FCGA, but reducing a large amount of computation time and a well-trained NN model that can serve as a nonlinear approximator was developed during the optimization process of the NN-FCGA.

Informal description and analysis of geographic requirements: an approach based on problems

Software requirements describe a problem in the real world that a software system is intended to solve. Describing requirements is challenging because usually too much attention is given to the final software product instead of concentrating on the problem itself and the real world. The area of geographic applications is no exception. Existing approaches to software development that are specific to the geographic area, for example, GIS tools, spatial databases, geographic query languages, and spatial data structures, are suitable for designing and implementing geographic applications and are, therefore, solution-oriented. We present a problem-oriented approach for requirements description of geographic applications. Most geographic applications are composed of well-known geographic subproblems. The proposed approach provides classes of common geographic subproblems that can be used to promote analysis and description of real-world problems. Each class of problems is presented as a problem frame showing domain properties, requirements and specifications. The problem frames discussed in this work are based on Jackson’s general purpose problem frames and are tailored here for the geographic area. The approach is validated through a case study.

Nonlinear optimization and parallel computing

The new computational technologies are having a very strong influence on numerical optimization, in several different ways. Many researchers have been stimulated by the need to either conform the existing numerical techniques to the new parallel architectures or to devise completely new parallel solution approaches. A mini-symposium on Parallel Computing in Nonlinear Optimization was held in Naples, Italy, September 2001, during the International Conference ParCo2001, in order to bring together researchers active in this field and to discuss and share their findings. Some of the papers presented during the mini-symposium, as well as additional contributions from other researchers are collected in this special issue. Clearly, two different trends, well representative for most of the current research activities, can be identified. Firstly, there is an attempt to encapsulate parallel linear algebra software and algorithms into optimization codes, particularly codes implementing interior point strategies for which the linear algebra issues are very critical, and secondly, there is an effort to devise new parallel solution strategies in global optimization, either for specific or general purpose problems, motivated by the large size and the combinatorial nature of them. In the present paper we review the literature on these trends and classify the contributed papers within this framework.

A statistical approach to adaptive problem solving

Domain independent general purpose problem solving techniques are desirable from the standpoints of software engineering and human computer interaction. They employ declarative and modular knowledge representations and present a constant homogeneous interface to the user, untainted by the peculiarities of the specific domain of interest. Unfortunately, this very insulation from domain details often precludes effective problem solving behavior. General approaches have proven successful in complex real-world situations only after a tedious cycle of manual experimentation and modification. Machine learning offers the prospect of automating this adaptation cycle, reducing the burden of domain specific tuning and reconciling the conflicting needs of generality and efficacy. A principal impediment to adaptive techniques is the utility problem: even if the acquired information is accurate and is helpful in isolated cases, it may degrade overall problem solving performance under difficult to predict circumstances. We develop a formal characterization of the utility problem and introduce COMPOSER, a statistically rigorous learning approach which avoids the utility problem. COMPOSER has been successfully applied to learning heuristics for planning and scheduling systems. This article includes theoretical results and an extensive empirical evaluation. The approach is shown to outperform significantly several other leading approaches to the utility problem.