The TorchLight Tool: Analyzing Search Topology Without Running Any Search

نویسنده

  • JÖRG HOFFMANN
چکیده

The ignoring delete lists relaxation is of paramount importance for both satisficing and optimal planning. In earlier work [Hoffmann, JAIR’05], it was observed that the optimal relaxation heuristic h has amazing qualities in many classical planning benchmarks, in particular pertaining to the complete absence of local minima. The proofs of this are hand-made, raising the question whether such proofs can be lead automatically by domain analysis techniques. The TorchLight tool answers this question in the affirmative. TorchLight is based on a connection between causal graph structure and h topology. It distinguishes between global analysis and local analysis. Global analysis shows the absence of local minima once and for all, for the entire state space of a given planning task. Local analysis computes what we call the success rate, which estimates the percentage of individual sample states not on local minima and thus allows to make finer distinctions. Finally, diagnosis summarizes structural reasons for analysis failure, thus indicating domain aspects that may cause local minima. TorchLight Results Overview lo ca l m in im a ed < = c

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Analyzing Search Topology Without Running Any Search: On the Connection Between Causal Graphs and h+

The ignoring delete lists relaxation is of paramount importance for both satisficing and optimal planning. In earlier work, it was observed that the optimal relaxation heuristic h has amazing qualities in many classical planning benchmarks, in particular pertaining to the complete absence of local minima. The proofs of this are hand-made, raising the question whether such proofs can be lead aut...

متن کامل

CONFIGURATION OPTIMIZATION OF TRUSSES USING A MULTI HEURISTIC BASED SEARCH METHOD

Different methods are available for simultaneous optimization of cross-section, topology and geometry of truss structures. Since the search space for this problem is very large, the probability of falling in local optimum is considerably high. On the other hand, different types of design variables (continuous and discrete) lead to some difficulties in the process of optimization. In this articl...

متن کامل

Where Ignoring Delete Lists Works, Part II: Causal Graphs

The ignoring delete lists relaxation is of paramount importance for both satisficing and optimal planning. In earlier work (Hoffmann 2005), it was observed that the optimal relaxation heuristic h has amazing qualities in many classical planning benchmarks, in particular pertaining to the complete absence of local minima. The proofs of this are hand-made, raising the question whether such proofs...

متن کامل

Using Kullback-Leibler distance for performance evaluation of search designs

This paper considers the search problem, introduced by Srivastava cite{Sr}. This is a model discrimination problem. In the context of search linear models, discrimination ability of search designs has been studied by several researchers. Some criteria have been developed to measure this capability, however, they are restricted in a sense of being able to work for searching only one possibl...

متن کامل

OPTIMAL DESIGN OF SINGLE-LAYER BARREL VAULT FRAMES USING IMPROVED MAGNETIC CHARGED SYSTEM SEARCH

The objective of this paper is to present an optimal design for single-layer barrel vault frames via improved magnetic charged system search (IMCSS) and open application programming interface (OAPI). The IMCSS algorithm is utilized as the optimization algorithm and the OAPI is used as an interface tool between analysis software and the programming language. In the proposed algorithm, magnetic c...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2011