Efficient reordering of C-PROLOG

Abstract

PROLOGuses a depih-first search of an AND/OR graph to satisfy queries against its database. It searches sequentially through the clauses of a predicate whose head matches the query, trying to satisfy the goals in the clause body in a sequential left-to-right order. The ordering of clauses and goals is a major factor in the eficiency of a PROLOG program. We have developed a profiler for C-PROLOG that collects statistics including the failure rate of clauses and goals in a C-PROLOG program. These statistics are used by any of several reordering predicates capable of local or global reordering. The intent is to construct a reordered PROLOG program that outputs an equivalent set of answers, and is more efficient. Test results are promising.

Optimizing Queries in a Logic-based Information Integration System

“…As a consequence, the order of members in a conjunction radically affects performance. It is well known that putting a member with a small expected number of solutions to the first place leads to a better execution time than putting a member with a plenty of solutions [3]. This optimization technique is similar to the techniques used by the query planners of database engines [4], [5], [6], but while the database has exact knowledge about the tables (keys, indexes, table sizes, number of different values in columns), we have to make do with some statistical data.…”

Shift-Scale Invariance of Electromagnetic Radiation

“…Such an approach, with minor modifications, is employed in the codes B2E (Elleaume & Marechal, 1991, 1997, UR (Dejus, 1994;Dejus & Luccio, 1994), YAUP (Boyanov et al, 1994) and SPECTRA (Tanaka & Kitamura, 2001) and is also integrated in the code RADID (C. Wang & Jin, 1992;C. Wang & Xiao, 1992;C. Wang, 1993).…”

“…Wang et al, 1994), UR (Dejus, 1994;Dejus & Luccio, 1994) and SpontLight (Geisler et al, 1994) (see also (Tatchyn et al, 1986;Yagi et al, 1995)). The problem with this approach is that the large number of individual computations for single-electron radiation can sometimes be too time consuming and impractical (Tatchyn et al, 1986;C. Wang, 1993;Lumpkin et al, 1995).…”

“…The validity of the off-axis approximation method has, however, been proven in the far-field region only (Jacobsen & Rarback, 1986;Rarback et al, 1988;Chapman et al, 1989), since the equations controlling the pattern of radiation in the near-field region are rather cumbersome. Because of this, it has been argued that in a real-life geometry, when the distance to the observer is limited and does not tends to infinity, the radiation does not have the scale properties of standard synchrotron radiation (Chapman et al, 1989) and thus, the off-axis approximation method fails in the near-field region (C. Wang & Xiao, 1992;C. Wang, 1993).…”