An efficient composition of bidirectional programs by memoization and lazy update
Publikation: Bidrag til bog/antologi/rapport › Konferencebidrag i proceedings › Forskning › fagfællebedømt
Standard
An efficient composition of bidirectional programs by memoization and lazy update. / Tsushima, Kanae; Trong, Bach Nguyen; Glück, Robert; Hu, Zhenjiang.
Functional and Logic Programming.: 15th International Symposium, FLOPS 2020 Akita, Japan, September 14–16, 2020 Proceedings. red. / Keisuke Nakano; Konstantinos Sagonas. Springer, 2020. s. 159-178 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Bind 12073 LNCS).Publikation: Bidrag til bog/antologi/rapport › Konferencebidrag i proceedings › Forskning › fagfællebedømt
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - GEN
T1 - An efficient composition of bidirectional programs by memoization and lazy update
AU - Tsushima, Kanae
AU - Trong, Bach Nguyen
AU - Glück, Robert
AU - Hu, Zhenjiang
PY - 2020
Y1 - 2020
N2 - Bidirectional transformations (BX) are a solution to the view update problem and widely used for synchronizing data. The semantics and correctness of bidirectional programs have been investigated intensively during the past years, but their efficiency and optimization are not yet fully understood. In this paper, as a first step, we study different evaluation methods to optimize their evaluation. We focus on the interpretive evaluation of BX compositions because we found that these compositions are an important cause of redundant computations if the compositions are not right associative. For evaluating BX compositions efficiently, we investigate two memoization methods. The first method, minBiGUL m , uses memoization, which improves the runtime of many BX programs by keeping intermediate results for later reuse. A disadvantage is the familiar tradeoff for keeping and searching values in a table. When inputs become large, the overhead increases and the effectiveness decreases. To deal with large inputs, we introduce the second method, xpg, that uses tupling, lazy update and lazy evaluation as optimizations. Lazy updates delay updates in closures and enables them to use them later. Both evaluation methods were fully implemented for minBiGUL. The experimental results show that our methods are faster than the original method of BiGUL for the non-right associative compositions.
AB - Bidirectional transformations (BX) are a solution to the view update problem and widely used for synchronizing data. The semantics and correctness of bidirectional programs have been investigated intensively during the past years, but their efficiency and optimization are not yet fully understood. In this paper, as a first step, we study different evaluation methods to optimize their evaluation. We focus on the interpretive evaluation of BX compositions because we found that these compositions are an important cause of redundant computations if the compositions are not right associative. For evaluating BX compositions efficiently, we investigate two memoization methods. The first method, minBiGUL m , uses memoization, which improves the runtime of many BX programs by keeping intermediate results for later reuse. A disadvantage is the familiar tradeoff for keeping and searching values in a table. When inputs become large, the overhead increases and the effectiveness decreases. To deal with large inputs, we introduce the second method, xpg, that uses tupling, lazy update and lazy evaluation as optimizations. Lazy updates delay updates in closures and enables them to use them later. Both evaluation methods were fully implemented for minBiGUL. The experimental results show that our methods are faster than the original method of BiGUL for the non-right associative compositions.
KW - Bidirectional transformation
KW - Efficiency
KW - Implementation technique
KW - Optimization
KW - Tupling
U2 - 10.1007/978-3-030-59025-3_10
DO - 10.1007/978-3-030-59025-3_10
M3 - Article in proceedings
AN - SCOPUS:85091327917
SN - 9783030590246
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 159
EP - 178
BT - Functional and Logic Programming.
A2 - Nakano, Keisuke
A2 - Sagonas, Konstantinos
PB - Springer
T2 - 15th International Symposium on Functional and Logic Programming, FLOPS 2020
Y2 - 14 September 2020 through 16 September 2020
ER -
ID: 249395716