SEARCH · Engineering Papers
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Technique for solution of constraint-satisfaction problems uses definite-clause grammars of Prolog computer language. Exploits fact that grammar-rule notation viewed as "state-change notation". Facilitates development of dynamic representation performing informed as well as blind searches. Applicable to design, scheduling, and planning problems.
A new technique for solving constraint satisfaction problems using Prolog's definite clause grammars is presented. It exploits the fact that the grammar rule notation can be viewed as a state exchange notation. The novel feature of the technique is that it can perform informed as well as blind search. It provides the Prolog programmer with a new technique for application to a wide range of design, scheduling, and planning problems.
Researchers describe a newly-developed artificial neural network algorithm for solving constraint satisfaction problems (CSPs) which includes a learning component that can significantly improve the performance of the network from run to run. The network, referred to as the Guarded Discrete Stochastic (GDS) network, is based on the discrete Hopfield network but differs from it primarily in that auxiliary networks (guards) are asymmetrically coupled to the main network to enforce certain types of constraints. Although the presence of asymmetric connections implies that the network may not converge, it was found that, for certain classes of problems, the network often quickly converges to find satisfactory solutions when they exist. The network can run efficiently on serial machines and can find solutions to very large problems (e.g., N-queens for N as large as 1024). One advantage of the network architecture is that network connection strengths need not be instantiated when the network is established: they are needed only when a participating neural element transitions from off to on. They have exploited this feature to devise a learning algorithm, based on consistency techniques for discrete CSPs, that updates the network biases and connection strengths and thus improves the network performance.
At the Honeywell Technology Center (HTC), we have been working on a scheduling problem related to commercial avionics. This application is large, complex, and hard to solve. To be a little more concrete: 'large' means almost 20,000 activities, 'complex' means several activity types, periodic behavior, and assorted types of temporal constraints, and 'hard to solve' means that we have been unable to eliminate backtracking through the use of search heuristics. At this point, we can generate solutions, where solutions exist, or report failure and sometimes why the system failed. To the best of our knowledge, this is among the largest and most complex scheduling problems to have been solved as a constraint satisfaction problem, at least that has appeared in the published literature. This abstract is a preliminary report on what we have done and how. In the next section, we present our approach to treating scheduling as a constraint satisfaction problem. The following sections present the application in more detail and describe how we solve scheduling problems in the application domain. The implemented system makes use of Ginsberg's Dynamic Backtracking algorithm, with some minor extensions to improve its utility for scheduling. We describe those extensions and the performance of the resulting system. The paper concludes with some general remarks, open questions and plans for future work.
In this paper we report our work on a new constraint domain, where variables can take structured values. Earth-science data processing (ESDP) is a planning domain that requires the ability to represent and reason about complex constraints over structured data, such as satellite images. This paper reports on a constraint-based planner for ESDP and similar domains. We discuss our approach for translating a planning problem into a constraint satisfaction problem (CSP) and for representing and reasoning about structured objects and constraints over structures.
The Space Station era presents a highly-complex multi-mission planning and scheduling environment exercised over a highly distributed system. In order to automate the scheduling process, customers require a mechanism for communicating their scheduling requirements to NASA. A request language that a remotely-located customer can use to specify his scheduling requirements to a NASA scheduler, thus automating the customer-scheduler interface, is described. This notation, Flexible Envelope-Request Notation (FERN), allows the user to completely specify his scheduling requirements such as resource usage, temporal constraints, and scheduling preferences and options. The FERN also contains mechanisms for representing schedule and resource availability information, which are used in the inter-scheduler inconsistency resolution process. Additionally, a scheduler is described that can accept these requests, process them, generate schedules, and return schedule and resource availability information to the requester. The Request-Oriented Scheduling Engine (ROSE) was designed to function either as an independent scheduler or as a scheduling element in a network of schedulers. When used in a network of schedulers, each ROSE communicates schedule and resource usage information to other schedulers via the FERN notation, enabling inconsistencies to be resolved between schedulers. Individual ROSE schedules are created by viewing the problem as a constraint satisfaction problem with a heuristically guided search strategy.
Constraint satisfaction problems are an important area of computer science. Many of these problems are in the complexity class NP which is exponentially hard for all known methods, both for worst cases and often typical. Fundamentally, the lack of any guided local minimum escape method ensures the hardness of both exact and approximate optimization classically, but the intuitive mechanism for approximation hardness in quantum algorithms based on Hamiltonian time evolution is poorly understood. We explore this question using the prototypically hard MAX-3-XORSAT problem class. We conclude that the mechanisms for quantum exact and approximation hardness are fundamentally distinct. We qualitatively identify why traditional methods such as quantum adiabatic optimization are not good approximation algorithms. We propose a new spectral folding optimization method that does not suffer from these issues and study it analytically and numerically. We consider random rank-3 hypergraphs including extremal planted solution instances, where the ground state satisfies an anomalously high fraction of constraints compared to truly random problems. We show that, if we define the energy to be $E = N_{unsat}-N_{sat}$, then spectrally folded quantum optimization will return states with energy $E \leq A E_{GS}$ (where $E_{GS}$ is the ground state energy) in polynomial time, where conservatively, $A \simeq 0.6$. We thoroughly benchmark variations of spectrally folded quantum optimization for random classically approximation-hard (planted solution) instances in simulation, and find performance consistent with this prediction. We do not claim that this approximation guarantee holds for all possible hypergraphs, though our algorithm's mechanism can likely generalize widely. These results suggest that quantum computers are more powerful for approximate optimization than had been previously assumed.
In this paper, I discuss four different areas of my research. One portion of my research has focused on automatic synthesis of search control heuristics for constraint satisfaction problems (CSPs). I have developed techniques for automatically synthesizing two types of heuristics for CSPs: Filtering functions are used to remove portions of a search space from consideration. Another portion of my research is focused on automatic synthesis of hierarchic algorithms for solving constraint satisfaction problems (CSPs). I have developed a technique for constructing hierarchic problem solvers based on numeric interval algebra. Another portion of my research is focused on automatic decomposition of design optimization problems. We are using the design of racing yacht hulls as a testbed domain for this research. Decomposition is especially important in the design of complex physical shapes such as yacht hulls. Another portion of my research is focused on intelligent model selection in design optimization. The model selection problem results from the difficulty of using exact models to analyze the performance of candidate designs.
The work described in this paper was inspired by a surprisingly effective neural network developed for scheduling astronomical observations on the Hubble Space Telescope. Our heuristic constraint satisfaction problem (CSP) method was distilled from an analysis of the network. In the process of carrying out the analysis, we discovered that the effectiveness of the network has little to do with its connectionist implementation. Furthermore, the ideas employed in the network can be implemented very efficiently within a symbolic CSP framework. The symbolic implementation is extremely simple. It also has the advantage that several different search strategies can be employed, although we have found that hill-climbing methods are particularly well-suited for the applications that we have investigated. We begin the paper with a brief review of the neural network. Following this, we describe our symbolic method for heuristic repair.
This paper describes DTS, a decision-theoretic scheduler designed to employ state-of-the-art probabilistic inference technology to speed the search for efficient solutions to constraint-satisfaction problems. Our approach involves assessing the performance of heuristic control strategies that are normally hard-coded into scheduling systems, and using probabilistic inference to aggregate this information in light of features of a given problem. BPS, the Bayesian Problem-Solver, introduced a similar approach to solving single-agent and adversarial graph search problems, yielding orders-of-magnitude improvement over traditional techniques. Initial efforts suggest that similar improvements will be realizable when applied to typical constraint-satisfaction scheduling problems.
This paper describes a decision theoretic scheduler (DTS) designed to employ state-of-the-art probabilistic inference technology to speed the search for efficient solutions to constraint-satisfaction problems. Our approach involves assessing the performance of heuristic control strategies that are normally hard-coded into scheduling systems and using probabilistic inference to aggregate this information in light of the features of a given problem. The Bayesian Problem-Solver (BPS) introduced a similar approach to solving single agent and adversarial graph search patterns yielding orders-of-magnitude improvement over traditional techniques. Initial efforts suggest that similar improvements will be realizable when applied to typical constraint-satisfaction scheduling problems.
This paper describes a simple heuristic approach to solving large-scale constraint satisfaction and scheduling problems. In this approach one starts with an inconsistent assignment for a set of variables and searches through the space of possible repairs. The search can be guided by a value-ordering heuristic, the min-conflicts heuristic, that attempts to minimize the number of constraint violations after each step. The heuristic can be used with a variety of different search strategies. We demonstrate empirically that on the n-queens problem, a technique based on this approach performs orders of magnitude better than traditional backtracking techniques. We also describe a scheduling application where the approach has been used successfully. A theoretical analysis is presented both to explain why this method works well on certain types of problems and to predict when it is likely to be most effective.
It is shown that quantum resonance can be used for combinatorial optimization. The advantage of the approach is in independence of the computing time upon the dimensionality of the problem. As an example, the solution to a constraint satisfaction problem of exponential complexity is demonstrated.
Many AI tasks can be formalized as constraint satisfaction problems (CSPs), which involve finding values for variables subject to constraints. While solving a CSP is an NP-complete task in general, tractable classes of CSPs have been identified based on the structure of the underlying constraint graphs. Much effort has been spent on exploiting structural properties of the constraint graph to improve the efficiency of finding a solution. These efforts contributed to development of a class of CSP solving algorithms called decomposition algorithms. The strength of CSP decomposition is that its worst-case complexity depends on the structural properties of the constraint graph and is usually better than the worst-case complexity of search methods. Its practical application is limited, however, since it cannot be applied if the CSP is not decomposable. In this paper, we propose a graph based backtracking algorithm called omega-CDBT, which shares merits and overcomes the weaknesses of both decomposition and search approaches.
Tractable classes of constraint satisfaction problems are of great importance in artificial intelligence. Identifying and taking advantage of such classes can significantly speed up constraint problem solving. In addition, tractable classes are utilized in applications where strict worst-case performance guarantees are required, such as constraint-based plan execution. In this work, we present a formal framework for search-free (backtrack-free) constraint satisfaction. The framework is based on general procedures, rather than specific propagation techniques, and thus generalizes existing techniques in this area. We also relate search-free problem solving to the notion of decision sets and use the result to provide a constructive criterion that is sufficient to guarantee search-free problem solving.
Scheduling can be formalized as a Constraint Satisfaction Problem (CSP). Within this framework activities belonging to a plan are interconnected via temporal constraints that account for slack among them. Temporal representation must include methods for constraints propagation and provide a logic for symbolic and numerical deductions. In this paper we describe a support framework for opportunistic reasoning in constraint directed scheduling. In order to focus the attention of an incremental scheduler on critical problem aspects, some discrete temporal indexes are presented. They are also useful for the prediction of the degree of resources contention. The predictive method expressed through our indexes can be seen as a Knowledge Source for an opportunistic scheduler with a blackboard architecture.
This final report covers work performed under Contract NCC2-220 between NASA Ames Research Center and the Knowledge Systems Laboratory, Stanford University. The period of research was from March 1, 1987 to February 29, 1988. Topics covered were as follows: (1) concurrent architectures for knowledge-based systems; (2) methods for the solution of geometric constraint satisfaction problems, and (3) reasoning under uncertainty. The research in concurrent architectures was co-funded by DARPA, as part of that agency's Strategic Computing Program. The research has been in progress since 1985, under DARPA and NASA sponsorship. The research in geometric constraint satisfaction has been done in the context of a particular application, that of determining the 3-D structure of complex protein molecules, using the constraints inferred from NMR measurements.
Previously studied theoretical frameworks for dynamic constraint satisfaction problems (DCSPs) employ a small set of primitive operators to modify a problem instance. They do not address the desire to model problems using sophisticated global constraints, and do not address efficiency questions related to incremental constraint enforcement. In this paper, we extend a DCSP framework to incorporate global constraints with flexible scope. A simple approach to incremental propagation after scope modification can be inefficient under some circumstances. We characterize the cases when this inefficiency can occur, and discuss two ways to alleviate this problem: adding rejection variables to the scope of flexible constraints, and adding new features to constraints that permit increased control over incremental propagation.