# Selected Publications

Note: The documents distributed by this server have been provided by the contributing authors as a means to ensure timely dissemination of scholarly and technical work on a noncommercial basis. Copyright and all rights therein are maintained by the authors or by other copyright holders, notwithstanding that they have offered their works here electronically. It is understood that all persons copying this information will adhere to the terms and constraints invoked by each author's copyright. These works may not be reposted without the explicit permission of the copyright holder.

For the fuller list of publications including those in Japanese (we write quite a few of them, mostly with my students and colleagues), please visit here.

I. Refereed or Invited Papers
II. Doctoral Thesis
III. Books and Proceedings
IV. Technical Reports, Workshop Papers, etc.

## I. Refereed or Invited Papers

1. Muhammad Hataba, Ahmed El-Mahdy and Kazunori Ueda: Generation of Efficient Obfuscated Code Through Just-In-Time Compilation. IEICE Transactions on Information and Systems, Vol.E102-D, No.3, 2019 (to appear).

2. [pdf] Alimujiang Yasen and Kazunori Ueda, Name Binding is Easy with Hypergraphs. IEICE Transactions on Information and Systems, Vol.E101-D, No.4, 2018, pp.1126-1140, DOI: 10.1587/transinf.2017EDP7257.

We develop a technique for representing variable names and name binding which is a mechanism of associating a name with an entity in many formal systems including logic, programming languages and mathematics. The idea is to use a general form of graph links (or edges) called hyperlinks to represent variables, graph nodes as constructors of the formal systems, and a graph type called hlground to define substitutions.
Our technique is based on simple notions of graph theory in which graph types ensure correct substitutions and keep bound variables distinct. We encode strong reduction of the untyped lambda-calculus to introduce our technique. Then we encode a more complex formal system called System F<:, a polymorphic lambda-calculus with subtyping that has been one of important theoretical foundations of functional programming languages. The advantage of our technique is that the representation of terms, definition of substitutions, and implementation of formal systems are all straightforward. We formalized the graph type hlground, proved that it ensures correct substitutions in the lambda-calculus, and implemented hlground in HyperLMNtal, a modeling language based on hypergraph rewriting. Experiments were conducted to test this technique. By this technique, one can implement formal systems simply by following the steps of their definitions as described in papers.

3. [pdf] Yutaro Tsunekawa, Taichi Tomioka and Kazunori Ueda: Implementation of LMNtal Model Checkers: a Metaprogramming Approach. The Journal of Object Technology, Vol.17, No.1, 2018, pp.1:1-28.

LMNtal is a modeling language based on hierarchical graph rewriting, and its implementation SLIM features state space search and an LTL model checker. Several variations and extensions of the SLIM have been developed, and all of them achieve their functionalities by modifying SLIM written in C. If a model checker is implemented in the modeling language itself, it should be easy to develop prototypes of various model checkers without changing the base implementation of the model checker. This approach is called metaprogramming which has been taken extensively in Lisp and Prolog communities.
In this paper, we design a framework for implementing extendable model checkers. First, we define first-class rewrite rules to extend a modeling language. Second, we design an API to operate on the states of programs. These features enable programmers to handle state transition graphs as first-class objects and implement diverse variants of a model checker without changing SLIM. We demonstrate it by implementing an LTL model checker and its variant and a CTL model checker. Furthermore, we show how easy it is to extend these model checkers in our framework by extending the CTL model checker to handle fairness constraints. The overhead of metainterpretation is around an order of magnitude or less. All these results demonstrate the viability of the resulting framework based on metainterpreters that handle explicit state space in a flexible manner.

4. [pdf] Alimujiang Yasen and Kazunori Ueda: Unification of Hypergraph Lambda-Terms. In Proc. Second IFIP International Conference on Topics in Theoretical Computer Science (TTCS 2017), LNCS 10608, Springer-Verlag, 2017, pp.106-124, DOI:10.1007/978-3-319-68953-1_9.

We developed a technique for modeling formal systems involving name binding in a modeling language based on hypergraph rewriting. A hypergraph consists of graph nodes, edges with two endpoints and edges with multiple endpoints. The idea is that hypergraphs allow us to represent terms containing bindings and that our notion of a graph type keeps bound variables distinct throughout rewriting steps.
We previously encoded the untyped lambda-calculus and the evaluation and type checking of System F_sub, but the encoding of System F_sub type inference requires a unification algorithm. We studied and successfully implemented a unification algorithm modulo alpha-equivalence for hypergraphs representing untyped lambda-terms. The unification algorithm turned out to be similar to nominal unification despite the fact that our approach and nominal approach to name binding are very different. However, some basic properties of our framework are easier to establish compared to the ones in nominal unification. We believe this indicates that hypergraphs provide a nice framework for encoding formal systems involving binders and unification modulo alpha-equivalence.

5. [pdf] Kazunozi Ueda: Logic/Constraint Programming and Concurrency: The Hard-Won Lessons of the Fifth Generation Computer Project. Science of Computer Programming, Vol.164 (2018), pp.3-17, http://dx.doi.org/10.1016/j.scico.2017.06.002.

The technical goal of the Fifth Generation Computer Systems (FGCS) project (1982-1993) was to develop Parallel Inference technologies, namely systematized technologies for realizing knowledge information processing on top of parallel computer architecture. The Logic Programming paradigm was adopted as the central working hypothesis of the project. At the same time, building a large-scale Parallel Inference Machine (PIM) meant to develop a novel form of general-purpose computing technologies that are powerful enough to express various parallel algorithms and to describe a full operating system of PIM. Accordingly, the research goal of the Kernel Language was set to designing a concurrent and parallel programming language under the working hypothesis of Logic Programming. The aim of this article is to describe the design process of the Kernel Language (KL1) in the context of related programming models in the 1980's, the essence of Concurrent Logic Programming and Constraint-Based Concurrency, and how the technologies we developed in those days evolved after their conception.

6. [pdf] Hiroaki Akutsu, Takahiro Yamamoto, Kazunori Ueda and Hideo Saito: MEC: Network Optimized Multi-stage Erasure Coding for Scalable Storage Systems. In Proc. 22nd IEEE Pacific Rim International Symposium on Dependable Computing (PRDC 2017), Christchurch, New Zealand, January 2017, IEEE Computer Society, 2017, pp. 292-300, DOI:10.1109/PRDC.2017.54.

In scalable storage systems, there are two kinds of methods for data redundancy: mirroring and parity. Each has its pros and cons. Mirroring creates a large amount of redundancy data, resulting in less usable space. Write performance degrades proportionally to the redundancy level due to an increase in communication. Parity-based methods partition data into multiple pieces, add parity information, and distribute the pieces of data and parity information. Parity-based methods are not often used with memory class media that are faster than the network, because distributing the data across servers results in low read performance. This research aims to establish an efficient data protection method that can be applied to fast, memory class media. We propose a new parity-based method called Multi-stage Erasure Coding (MEC), which creates two different erasure codes: one at the data transmission source server, and the other at the destination server. We show that our method reduces the space required to achieve redundancy while achieving high performance by making the amount of write communication independent of the redundancy level. We built a prototype program using MEC on a commodity cluster server. We show that compared with conventional parity-based methods with redundancy level 2, read I/O throughput is over one order of magnitude higher thanks to local reads and that write I/O throughput is almost the same due to network bottleneck.

7. [pdf] Yutaro Tsunekawa, Taichi Tomioka and Kazunori Ueda: Implementation of LMNtal Model Checkers: a Metaprogramming Approach. First Workshop on Meta-Programming Techniques and Reflection (META'16), Amsterdam, Oct. 2016.

LMNtal is a modeling language based on hierarchical graph rewriting, and its implementation SLIM features an LTL model checker. Several variations and extensions of the SLIM model checker have been developed, and all of them achieve their functionalities by modifying SLIM. If a model checker is implemented in the modeling language itself, it should be easy to develop prototypes of various model checkers without changing the implementation of the model checker. This approach is called metaprogramming which has been taken conventionally in Lisp and Prolog. In this paper, we define first-class rewrite rules to enable metaprogramming in LMNtal. In addition, we design APIs to operate states of LMNtal programs. These features allow us to implement diverse variants of a model checker without changing SLIM.We demonstrate it by implementing an LTL model checker and a CTL model checker in LMNtal. The resulting framework should be sufficient to develop metacir- cular interpreters that treat program states flexibly.

8. [pdf] Kenichi Betsuno, Shota Matsumoto and Kazunori Ueda: Symbolic Analysis of Hybrid Systems Involving Numerous Discrete Changes Using Loop Detection. In Proc. Sixth Workshop on Design, Modeling and Evaluation of Cyber Physical Systems (CyPhy'16), Pittsburg, USA, October 2016, LNCS 10107, Springer-Verlag, 2017, pp.17-30, DOI:10.1007/978-3-319-51738-4_2.

Hybrid systems are dynamical systems that include both continuous and discrete changes. Some hybrid systems involve a large or infinite number of discrete changes with in an infinitesimal-width region of phase space. Systems with sliding mode are typical examples of such hybrid systems. It is difficult to analyze such hybrid systems through ordinary numerical simul ation, since the time required for simulation increases in proportion to the number of discrete changes. In this paper, we propose a method to symbolically analyze such models involvin g numerous discrete changes by detecting loops and checking loop invariants of the model's behavior. The method handles parameterized hybrid systems and checks inclusion of paramet erized states focusing on the values of a switching function that dominate the dynamics of sliding mode. We implemented the main part of the method in our symbolic hybrid system simula tor HyLaGI, and conducted analysis of example models.

9. [pdf] Shota Matsumoto and Kazunori Ueda: Symbolic Simulation of Parametrized Hybrid Systems with Affine Arithmetic. In Proc. 23rd International Symposium on Temporal Representation and Reasoning (TIME 2016), Copenhagen, IEEE Computer Society, Oct. 2016, pp.4-11, DOI:10.1109/TIME.2016.8.

The purpose of this research is to develop a highly reliable simulator of hybrid systems, i.e., systems involving both discrete change and continuous evolution. In particular, we aim at rigorous simulation of parametrized hybrid systems, which enables not only the analysis of modelfs possible behavior but also the design of parameters that realize desired properties. Simula- tors with interval arithmetic can reliably compute a reachable set of states, but preserving the dependency of uncertain quantities in models is still challenging. In this paper, we discuss a simulation method that is based on symbolic computation and cooperates with the interval Newton method and affine arithmetic, which is able to preserve first- order dependency of uncertain quantities. We implemented the algorithm on the symbolic simulator we have been developing and evaluated the performance of the method with example models.

10. [pdf] Hiroaki Akutsu,Kazunori Ueda,Takeru Chiba, Tomohiro Kawaguchi, Norio Shimozono: Reliability and Failure Impact Analysis of Distributed Storage Systems with Dynamic Refuging. IEICE Transactions on Information and Systems, Vol.E99-D, No.9, 2016, pp.2259-2268, DOI:10.1587/transinf.2016EDP7139.

In recent data centers, large-scale storage systems storing big data comprise thousands of large-capacity drives. Our goal is to establish a method for building highly reliable storage systems using more than a thousand low-cost large-capacity drives. Some large- scale storage systems protect data by erasure coding to prevent data loss. As the redundancy level of erasure coding is increased, the probability of data loss will decrease, but the increase in normal data write operation and additional storage for coding will be incurred. We therefore need to achieve high reliability at the lowest possible redundancy level. There are two concerns regarding reliability in large- scale storage systems: (i) as the number of drives increases, systems are more subject to multiple drive failures and (ii) distributing stripes among many drives can speed up the rebuild time but increase the risk of data loss due to multiple drive failures. If data loss occurs by multiple drive failure, it affects many users using a storage system. These concerns were not addressed in prior quantitative reliability studies based on realistic settings. In this work, we analyze the reliability of large-scale storage systems with distributed stripes, focusing on an effective rebuild method which we call Dynamic Refuging. Dynamic Refuging rebuilds failed blocks from those with the lowest redundancy and strategically selects blocks to read for repairing lost data. We modeled the dynamic change of amount of storage at each redundancy level caused by multiple drive failures, and performed reliability analysis with Monte Carlo simulation using realistic drive failure characteristics. We showed a failure impact model and a method for localizing the failure. When stripes with redundancy level 3 were sufficiently distributed and rebuilt by Dynamic Refuging, the proposed technique turned out to scale well, and the probability of data loss decreased by two orders of magnitude for systems with a thousand drives compared to normal RAID. The appropriate distribution level could localize the failure.

11. [pdf] Alimujiang Yasen and Kazunori Ueda. Alimujiang Yasen and Kazunori Ueda: Hypergraph Representation of Lambda-Terms. In Proc. 10th International Symposium on Theoretical Aspects of Software Engineering, IEEE Compueter Society, 2016, pp.113-116, DOI:10.1109/TASE.2016.25.

Substitution in the lambda-calculus is a subtle operation. In a formal description, Barendregt's variable convention is assumed to avoid variable capture, but such an assumption is not well suited for implementation on computers.We introduce graph representation and manipulation of lambda-terms, in which bound and free variables are encoded by using hyperlinks with different attributes. A graph type called hlground is generalized to identify the scope of a term n in substitution m[x := n], which enables bound variables and free variables to have suitable behavior during the substitution. Our representation of the lambda-terms are readable and the definition of substitution in this technique is free from any side conditions on the freeness and freshness of variables.

12. [pdf] Kazunori Ueda: Logic/Constraint Programming and Concurrency: The Hard-Won Lessons of the Fifth Generation Computer Project. In Proc. 13th International Symposium on Functional and Logic Programming (FLOPS 2016), Oleg Kiselyov and Andy King (eds.), LNCS 9613, Springer-Verlang, 2016, pp.1-11, DOI:10.1007/978-3-319-29604-3_1.

The technical goal of the Fifth Generation Computer Systems (FGCS) project (1982-1993) was to develop Parallel Inference technologies, namely systematized technologies for realizing knowledge information processing on top of parallel computer architecture. The Logic Programming paradigm was adopted as the central working hypothesis of the project. At the same time, building a large-scale Parallel Inference Machine (PIM) meant to develop a novel form of general-purpose computing technologies that are powerful enough to express various parallel algorithms and to describe a full operating system of PIM. Accordingly, the research goal of the Kernel Language was set to designing a concurrent and parallel programming language under the working hypothesis of Logic Programming. The aim of this article is to describe the design process of the Kernel Language (KL1) in the context of related programming models in the 1980's, the essence of Concurrent Logic Programming and Constraint-Based Concurrency, and how the technologies we developed in those days evolved after their conception.

13. [pdf] Kazuhiro Miyahara and Kazunori Ueda: Optimized Canonical Labeling Algorithm for Graph Rewriting Systems. Computer Software, Vol.33, No.1 (2015), pp.126-149.

In graph-based model checking, systems are modeled with graph structures which are highly expressive and feature a symmetry reduction mechanism. However, it involves frequent isomorphism checking of graphs generated in the course of model checking. Canonical labeling of graphs, which gives a unique representation to isomorphic graphs, is expected to be an effective method to check isomorphism among many graphs efficiently. It is therefore important to efficiently compute canonical forms of graphs in graph rewriting systems. For this purpose, we propose an optimization technique for McKay's canonical labeling algorithm that reuses information of graph structures that does not change by rewriting. To enable reuse, we reformulated McKay's algorithm to clarify what substructures of graphs are utilized in its execution, and designed an algorithm for successive graph canonicalization that organizes graph information in such a way that recomputation may be minimized. We evaluated the performance of the proposed algorithm, and found that it achieved sublinear time complexity with respect to the number of vertices for many cases of re-canonicalization.

14. [pdf] Shota Matsumoto, Fumihiko Kono, Teruya Kobayashi and Kazunori Ueda: HyLaGI: Symbolic Implementation of a Hybrid Constraint Language,
Electronic Notes in Theoretical Computer Science, Vol.317 (2015), pp.109-115. (Also in Proc. NSV 2015: 8th International Workshop on Numerical Software Verification, Seattle, USA, April 13, 2015, pp.70-75.)

A modeling language for hybrid systems HydLa and its implementation HyLaGI are described. HydLa is a constraint-based language that can handle uncertainties of models smoothly. HyLaGI calculates trajectories by symbolic formula manipulation to exclude errors resulting from floating-point arithmetic. HyLaGI features a nondeterministic simulation algorithm so it can calculate all possible qualitative different trajectories of models with uncertainties.

15. [pdf] Hiroaki Akutsu, Kazunori Ueda, Takeru Chiba, Tomohiro Kawaguchi and Norio Shimozono: Reliability Analysisof Highly Redundant Distributed Storage Systems with Dynamic Refuging, in Proc. 23rd Euromicro International Conference on Parallel, Distributed and Network-based Processing, Turk, Finland, March 4, 2015, pp.261-268, DOI:10.1109/PDP.2015.32.

In recent data centers, large-scale storage systems storing big data comprise thousands of large-capacity drives. Our goal is to establish a method for building highly reliable storage systems using more than a thousand low-cost large-capacity drives. Some large-scale storage systems protect data by erasure coding to prevent data loss. As the redundancy level of erasure coding is increased, the probability of data loss will decrease, but the increase in normal data write operation and additional storage for coding will be incurred. We therefore need to achieve high reliability at the lowest possible redundancy level. There are two concerns regarding reliability in large-scale storage systems: (i) as the number of drives increases, systems are more subject to multiple drive failures and (ii) distributing stripes among many drives can speed up the rebuild time but increase the risk of data loss due to multiple drive failures. These concerns were not addressed in prior quantitative reliability studies based on realistic settings. In this work, we analyze the reliability of large-scale storage systems with distributed stripes, focusing on an effective rebuild method which we call Dynamic Refuging. Dynamic Refuging rebuilds failed storage areas from those with the lowest redundancy and strategically selects blocks to read for repairing lost data. We modeled the dynamically changing amount of storage at each redundancy level due to multiple drive failures, and performed reliability analysis with Monte Carlo simulation using realistic drive failure characteristics. When stripes with redundancy level 3 were sufficiently distributed and rebuilt by Dynamic Refuging, we found that the probability of data loss decreased by two orders of magnitude for systems with 384 or more drives compared to normal RAID. This technique turned out to scale well, and a system with 1536 inexpensive drives attained lower data loss probability than RAID 6 with 16 enterprise-class drives.

16. [pdf] Ryo Yasuda, Taketo Yoshida, and Kazunori Ueda: Reduction of the Number of States and the Acceleration of LMNtal Parallel Model Checking. Trans. Japanese Society for Artificial Intelligence, Vol.29, No.1 (2014), pp.182-187.

SLIM is an LMNtal runtime. LMNtal is a programming and modeling language based on hierarchical graph rewriting. SLIM features automata-based LTL model checking that is one of the methods to solve accepting cycle search problems. Parallel search algorithms OWCTY and MAP used by SLIM generate a large number of states for problems having and accepting cycles. Moreover, they have a problem that performance seriously falls for particular problems. We propose a new algorithm that combines MAP and Nested DFS to remove states for problems including accepting cycles. We experimented the algorithm and confirmed improvements both in performance and scalability.

17. [pdf] Shota Matsumoto and Kazunori Ueda: Hyrose: A Symbolic Simulator of the Hybrid Constraint Language HydLa. Computer Software, Vol.30, No.4 (2013), pp.18-35.

Hybrid systems are dynamical systems with continuous changes of states and discrete changes of states and governing equations. Hybrid systems are appliable to diverse fields including physics. HydLa is a declarative language for describing hybrid systems using constraints, and we have developed a symbolic simulator of HydLa named Hyrose. Hyrose is our first publicly available implementation of HydLa. Features of Hyrose include the simulation of systems with parameters, search based on automatic case analysis and bounded model checking. In this paper, we first describe the architecture and features of Hyrose, as well as semantical issues that had to be clarified to implement Hyrose. Then, in order to demonstrate the power of symbolic execution, we show several examples of the analysis of systems whose behavior exhibits quantitative and qualitative changes depending on parameters.

18. [pdf] Kazunori Ueda: Towards a Substrate Framework of Computation. In Concurrent Objects and Beyond, Gul Agha et al. (eds.), LNCS 8665, Springer-Verlag, 2014, pp.341-366, DOI:10.1007/978-3-662-44471-9_15.

A grand challenge in computing is to establish a substrate computational model that encompasses diverse forms of non-sequential computation. This paper demonstrates how a hypergraph rewriting framework nicely integrates various forms and ingredients of concurrent computation and how simple static analyses help the understanding and optimization of programs. Hypergraph rewriting treats processes and messages in a unified manner, and treats message sending and parameter passing as symmetric reaction between two entities. Specifically, we show how fine-grained strong reduction of the lambda-calculus can be concisely encoded into hypergraph rewriting with a small set of primitive operations.

19. [pdf] Kazunori Ueda, Shota Matsumoto, Akira Takeguchi, Hiroshi Hosobe, Daisuke Ishii: HydLa: A High-Level Language for Hybrid Systems. In Proc. Second Workshop on Logics for System Analysis (LfSA 2012, affiliated with CAV 2012), July 2012, pp.3-17.

We have been working on the design and implementation of HydLa, a high-level modeling language for hybrid systems. Its objectives and features include a constraint-based (as opposed to automata-based) formalism, proper handling of uncertainties, and nondeterministic execution. The talk will describe an overview of the language with live demonstration of our prototype implementation.

20. [pdf] Kazunori Ueda and Seiji Ogawa: HyperLMNtal: An Extension of a Hierarchical Graph Rewriting Model. Künstliche Intelligenz, Vol.26, No.1 (2012), pp.27-36. DOI: 10.1007/s13218-011-0162-3.

LMNtal is a language model based on hierarchical graph rewriting that uses point-to-point links to represent connectivity and membranes to represent hierarchy. LMNtal was designed to be a substrate language of various computational models, especially those addressing concurrency, mobility and multiset rewriting. Although point-to-point links and membranes could be used together to represent multipoint connectivity, our experiences with LMNtal told that \textit{hyperlinks} would be a important and useful extension to the language.
We have accordingly expanded LMNtal into a hierarchical hypergraph rewriting language model, HyperLMNtal. HyperLMNtal enabled concise description of computational models involving flexible and diverse forms of references between data; in particular, it enabled efficient encoding of a constraint processing language CHR in terms of both performance and computational complexity. This paper describes the design and implementation of HyperLMNtal as a case study of language evolution.

21. [pdf] Masato Gocho, Taisuke Hori and Kazunori Ueda: Evolution of the LMNtal Runtime to a Parallel Model Checker. Computer Software, Vol.28, No.4 (2011), pp.137-157.

Model checking is a verification technique based on the exhaustive search of erroneous behaviors of state transition systems, which is attracting growing interest. The LMNtal language allows us to represent state transition systems using hierarchical graph rewriting, and its runtime system, SLIM, has recently evolved into an explicit-state model checker. Hierarchical graphs are a powerful data structure that features a built-in symmetry reduction mechanism that plays a key role in state space search. However, model checking is still prone to state space explosion and has called for time- and space-efficient methods for state management. To address this problem and make our model checker more useful, we have extended the SLIM runtime further to a parallel model checker to run on shared-memory multiprocessors and developed various optimization techniques of state management. These two enhancements drastically improved the scale of problems the LMNtal model checker can handle as well as its performance.

22. [pdf] Daisuke Ishii, Kazunori Ueda, Hiroshi Hosobe: An Interval-based SAT Modulo ODE Solver for Model Checking Nonlinear Hybrid Systems. Int. J. Softw. Tools. Technol. Transfer, Vol.13, No.5 (2011), pp.449-461, DOI: 10.1007/s10009-011-0193-y.

This paper presents a bounded model checking (BMC) tool called hydlogic for hybrid systems. It translates a reachability problem of a nonlinear hybrid system into a predicate logic formula involving arithmetic constraints, and checks the satisfiability of the formula based on a satisfiability modulo theories (SMT) method. We tightly integrate (i) an incremental SAT solver to enumerate the possible sets of constraints and (ii) an interval-based solver for hybrid constraint systems (HCSs) to solve the constraints described in the formulas. The HCS solver verifies the occurrence of a discrete change by enclosing continuous states that may cause the discrete change by a set of boxes.We adopt the existence property of a unique solution in the boxes computed by the HCS solver as (i) a proof of the reachability of a model, and (ii) a guide in the over-approximation refinement procedure. Our hydlogic implementation successfully handled several examples including those with nonlinear constraints.

23. [pdf] Shun Shibuya, Kenshiro Takata, Hiroshi Hosobe and Kazunori Ueda: An Execution Algorithm for the Hybrid System Modeling Language HydLa. Computer Software, Vol.28, No.3 (2011), pp. 167-172.

Hybrid systems are dynamical systems with continuous changes of states and discrete changes of states and governing equations. HydLa is a constraint-based modeling language for hybrid systems, and its implementation aims for system verification by means of simulation with guaranteed accuracy. However, how to simulate HydLa programs with initial values given as ranges or with symbolic parameters has not been clarified. This paper gives a nondeterministic execution algorithm that correctly handles discrete and continuous changes in HydLa programs.

24. [pdf] Toshiki Kawabata, Fumiyoshi Kobayashi and Kazunori Ueda: Speedup of OWCTY Model Checking Algorithm Using Strongly Connected Components. Trans. Japanese Society for Artificial Intelligence, Vol.26, No.2 (2011), pp.341-346.

verification. Automata-based LTL model checking is one of the methods to solve accepting cycle search problems. Model checking is prone to state-space explosion, and we may expect that parallel processing would be a promising approach. However, the optimal sequential algorithm is P-complete and is difficult to parallelize. Alternative parallel algorithms have been proposed, and OWCTY reversed is one of them. OWCTY reversed is known to be a stable and fast algorithm for models without bugs, but it does not use the characteristics of the automata used in LTL model checking. We propose a new algorithm named SCCOWCTY that exploits the SCCs (strongly connected components) of property automata. The algorithm removes states that are judged not to form accepting cycles faster than OWCTY reversed. We experimented and compared the two algorithms using DiVinE, and confirmed improvements both in performance and scalability.

25. [pdf] Kazunori Ueda, Hiroshi Hosobe and Daisuke Ishii: Declarative Semantics of the Hybrid Constraint Language HydLa. Computer Software, Vol.28, No.1 (2011), pp.306-311.

Hybrid systems are dynamical systems with continuous evolution of states and discrete evolution of states and governing equations. We have worked on the design and implementation of HydLa, a constraint-based modeling language for hybrid systems, with a view to the proper handling of uncertanties and the integration of simulation and verification. HydLa's constraint hierarchies facilitate the description of constraints with adequate strength, but its semantical foundations are not obvious due to the interaction of various language constructs. This paper gives the declarative semantics of HydLa and discusses its properties and consequences by means of examples.

26. [pdf] Takayuki Ayano, Taisuke Hori, Hiroki Iwasawa, Seiji Ogawa and Kazunori Ueda: LMNtal Model Checking using an Integrated Development Environment. Computer Software, Vol.27, No.4 (2010), pp.197-214.

LMNtal was designed and implemented as a unifying computational model based on hierarchical graph rewriting. In pursuit of its application to the fields of verification and search, the system has recently evolved into a model checker that employs LMNtal as a modeling language, and we have been accumulating experiences with modeling and verification. This paper describes LMNtalEditor, an integrated development environment (IDE) featuring both state space search and visualization of verification, and gives various examples of model description, verification and visualization. LMNtalEditor features the visualization and browsing of state space and counterexamples, the searching of states of interest, and so on. We have successfully run diverse examples taken from the fields including concurrency and AI search, and found that the IDE plays an essential role in understanding the models and counterexamples and thus greatly eases the task of verification. Furthermore, through the encoding of models in Promela, MSR and Coloured Petri Nets into LMNtal, we have extended the expressive power of LMNtal to the field of verification.

27. [pdf] Daisuke Ishii, Kazunori Ueda, Hiroshi Hosobe and Alexandre Goldsztejn, Interval-based solving of hybrid constraint systems. In Proc. Third IFAC Conference on Analysis and Design of Hybrid Systems (ADHS'09), 2009, pp.144-149.

An approach for reliable modeling, simulation and verification of hybrid systems is interval arithmetic, which guarantees that the resulting intervals do not exceed specified size. Interval-based computation on hybrid systems are often difficult, especially when the systems are described by nonlinear ordinary differential equations (ODEs) and nonlinear algebraic equations. We formulate the problem of detecting a discrete change in hybrid systems as a hybrid constraint system (HCS), consisting of a flow constraint on trajectories (i.e. continuous functions over time) and a guard constraint on states causing discrete changes. We also implemented a technique for solving HCSs by coordinating (i) interval-based solving of nonlinear ODEs, and (ii) a constraint programming technique for reducing of interval enclosures of solutions. The proposed technique reliably solves HCSs with nonlinear constraints. Our technique employs the interval Newton method to accelerate the reduction of interval enclosures, while guaranteeing that the enclosure contains a solution.

28. [pdf] Kazunori Ueda, Takayuki Ayano, Taisuke Hori, Hiroki Iwasawa and Seiji Ogawa, Hierarchical Graph Rewriting as a Unifying Tool for Analyzing and Understanding Nondeterministic Systems. In Proc. Sixth International Colloquium on Theoretical Aspects of Computing (ICTAC 2009), LNCS 5684, Springer, 2009, pp.349-355.

We have designed and implemented LMNtal (pronounced "elemental"), a language based on hierarchical graph rewriting that allows us to encode diverse computational models involving concurrency, mobility and multiset rewriting. Towards its novel applications, the system has recently evolved into a model checker that employs LMNtal as the modeling language and PLTL as the specification language. The strengths of our LMNtal model checker are its powerful data structure, highly nondeterministic computation it can express, and virtually no discrepancy between programming and modeling languages. Models expressed in Promela, MSR, and Coloured Petri Nets can be easily encoded into LMNtal. The visualizer of the LMNtal IDE turned out to be extremely useful in understanding models by state space browsing. The LMNtal IDE has been used to run and visualize diverse examples taken from the fields of model checking, concurrency and AI search.

29. [pdf] Kei Ohmura and Kazunori Ueda, c-sat: A Parallel SAT Solver for Clusters. In Proc. Twelfth International Conference on Theory and Applications of Satisfiability Testing (SAT 2009), LNCS 5584, Springer, 2009, pp.524-537.

Parallelizing modern SAT solvers for clusters such as Beowulf is an important challenge both in terms of performance scalability and stability. This paper describes a SAT Solver c-sat, a parallelization of MiniSat using MPI. It employs a layered master-worker architecture, where the masters handle lemma exchange, deletion of redundant lemmas and the dynamic partitioning of search trees, while the workers do search using different decision heuristics and random number seeds. With careful tuning, c-sat showed good speedup over MiniSat with reasonably small communication overhead on various clusters. On an eight-node cluster with two Dual-Core Opterons on each node (32 PEs), c-sat ran at least 23 times faster than MiniSat using 31 PEs (geometric mean; at least 31 times for satisfiable problems) for 189 large-scale problems from SAT Competition and two SAT-Races.

30. [pdf] Kazunori Ueda, LMNtal as a Hierarchical Logic Programming Language. Theoretical Computer Science, Vol.410, No.46 (2009), pp.4784-4800.
doi:10.1016/j.tcs.2009.07.043

LMNtal (pronounced "elemental") is a simple language model based on hierarchical graph rewriting that uses logical variables to represent connectivity and membranes to represent hierarchy. LMNtal is an outcome of the attempt to unify constraint-based concurrency and Constraint Handling Rules (CHR), the two notable extensions to concurrent logic programming. LMNtal is intended to be a substrate language of various computational models, especially those addressing concurrency, mobility and multiset rewriting. Although the principal objective of LMNtal was to provide a unifying computational model, it is of interest to equip the formalism with a precise logical interpretation. In this paper, we show that it is possible to give LMNtal a simple logical interpretation based on intuitionistic linear logic and a flattening technique. This enables us to call LMNtal a hierarchical, concurrent linear logic language.

31. [pdf] Daisuke Ishii, Kazunori Ueda and Hiroshi Hosobe: An Interval-based Consistency Technique for Reliable Simulation of Hybrid Systems. IPSJ Transactions on Mathematical Modeling and its Applications, Vol.1, No.1 (2008), pp.149-159.

Hybrid systems are systems consisting of discrete changes and continuous changes over time. Problems in various fields such as physics can be modeled as hybrid systems. In a simulation of a hybrid system, it is difficult to obtain rigorous solution of the model because of computation errors in the handling of continuous states. Our proposed method integrates interval-based techniques for initial value problems for ordinary differential equations and consistency techniques for nonlinear problems. The method obtains complete interval enclosures of hybrid trajectories efficiently with a given accuracy. We implemented the method and evaluated its effectiveness. Using the method, we can simulate hybrid systems reliably with guaranteeing its accuracy.

32. [pdf] Kazunori Ueda, Encoding the Pure Lambda Calculus into Hierarchical Graph Rewriting. In Proc. 19th International Conference on Rewriting Techniques and Applications (RTA 2008), LNCS 5117, Springer, 2008, pp.392-408.

Fine-grained reformulation of the lambda calculus is expected to solve several difficulties with the notion of substitutions--definition, implementation and cost properties. However, previous attempts including those using explicit substitutions and those using Interaction Nets were not ideally simple when it came to the encoding of the pure (as opposed to weak) lambda calculus. This paper presents a novel, fine-grained, and highly asynchronous encoding of the pure lambda calculus using LMNtal, a hierarchical graph rewriting language, and discusses its properties. The major strength of the encoding is that it is significantly simpler than previous encodings, making it promising as an alternative formulation, rather than just the encoding, of the pure lambda calculus. The membrane construct of LMNtal plays an essential role in encoding colored tokens and operations on them. The encoding has been tested using the publicly available LMNtal implementation.

33. [pdf] Kei Murayama, Shintaro Kudo, Ken Sakurai, Ken Mizuno, Norio Kato and Kazunori Ueda: Implementation of the Hierarchical Graph Rewriting Language LMNtal. Computer Software, Vol.25, No.2 (2008), pp.47-77.

LMNtal is a language model based on hierarchical graph rewriting that allows the representation and the manipulation of connectivity expressed using links and hierarchy expressed using membranes. The model allows concise description of programs involving dynamic data structures and multiset processing. Since LMNtal features fine-grained concurrency whose atomic operation is the application of a rewrite rule, how to implement it correctly and efficiently is far from obvious. This motivated us to implement LMNtal using Java and establish implementation techniques that are both correct and reasonably efficient. Our implementation consists of a compiler into intermediate code, an interpreter of the intermediate code, and a translator from the intermediate code to Java source code. It also provides a variety of useful features including foreign-language interface. The cruxes of the implementation techniques include the handling of links that cross membranes and the asynchronous execution of rewrite rules belonging to different membranes. This paper describes the design and implementation of our publicly available LMNtal system, focusing on technical issues raised in the course of the development. Those issues include the representation of hierarchical graph structures, intermediate code, and safe asynchronous execution of multiple tasks.

34. [pdf] Atsuyuki Inui, Shintaro Kudo, Koji Hara, Ken Mizuno, Norio Kato and Kazunori Ueda: LMNtal: The Unifying Programming Language Based on Hierarchical Graph Rewriting. Computer Software, Vol.25, No.1 (2008), pp.124-150.

LMNtal is a simple language model based on hierarchical graph rewriting that uses logical variables to represent connectivity and membranes to represent hierarchy. The major goals of LMNtal have been to unify various computational models addressing multiset rewriting, concurrency and mobility, and at the same time to provide a practical programming language based on hierarchical graph rewriting and demonstrate its versatility. The purpose of this paper is to present the aspects of LMNtal as a full-fledged programming language and demonstrate its expressive power using a number of examples. We have designed important language features such as arithmetics, rule application control, modularity and foreign-language interface and integrated them into the hierarchical graph rewriting model. To demonstrate the expressive power of the language, we have successfully encoded diverse related computational models including the lambda-calculous, the pi-calculus, the ambient calculus and CHR and run them on our LMNtal system.

35. [pdf] Kazunori Ueda, Encoding Distributed Process Calculi into LMNtal. Electronic Notes in Theoretical Computer Science, Vol.209 (2008), pp.187-200.

Towards a unifying model of concurrency, we have designed and implemented LMNtal (pronounced "elemental"), a model and language based on hierarchical graph rewriting that uses logical variables to represent connectivity and membranes to represent hierarchy. Diverse computational models including the pi-calculus and the lambda-calculus have been encoded into LMNtal and tested on our LMNtal system. This paper describes the encoding of the ambient calculus with running examples. The technical crux is the distributed management of names in the presence of locality an mobility. We have designed and implemented a self-adjusting management technique of names in which the name management and mobility operations are performed concurrently.

36. [pdf] Daisuke Ishii, Kazunori Ueda, Hiroshi Hosobe, An Interval-Based Approximation Method for Discrete Changes in Hybrid CC. In Frederic Benhamou, Narendra Jussien, and Barry O'Sullivan (Eds.), Trends in Constraint Programming, pp. 245-255, ISTE, London, 2007 (Re-reviewed post-proceedings of the CP2006 Workshops).

Hybrid cc is a constraint-based programming language aimed for simple modeling and rigorous simulation of hybrid systems. The existing implementation of Hybrid cc uses intervals to handle partially specified models. Although the implementation solves nonlinear equations and ordinary differential equations (ODEs) based on interval arithmetic, discrete changes are not properly computed with intervals. We propose a method for enclosing every solution of a hybrid system with a set of boxes. Our method computes continuous state changes by solving ODEs with initial time and values expressed by intervals. Then the method detects and encloses every state that will cause a discrete change. We devise algorithms for pruning and splitting intervals to obtain tight enclosures. The experimental results show the efficiency of the algorithms.

37. [pdf] Kazunori Ueda, Norio Kato, Koji Hara and Ken Mizuno, LMNtal as a Unifying Declarative Language. In Proc. Third workshop on Constraing Handling Rules (CHR 2006), pp.1-15 (invited talk).

LMNtal (pronounced "elemental") is a simple language model based on hierarchical graph rewriting that uses logical variables to represent connectivity and membranes to represent hierarchy. LMNtal is an outcome of the attempt to unify constraint-based concurrency and Constraint Handling Rules (CHR), the two notable extensions to concurrent logic programming. LMNtal is intended to be a substrate language of various computational models, especially those addressing concurrency, mobility and multiset rewriting. Another important goal of LMNtal has been to put hierarchical graph rewriting into practice and demonstrate its versatility by designing and implementing a full-fledged, monolithic programming language. In this paper, we demonstrate the practical aspects of LMNtal using a number of examples taken from diverse areas of computer science. Also, we discuss the relationship between LMNtal and CHR, which exhibit both commonalities and differences in various respects.

38. [pdf] Kazunori Ueda, Logic Programming and Concurrency: a Personal Perspective. The ALP NewsLetter, Vol.19, No.2, May 2006.

39. [pdf] Kazunori Ueda and Norio Kato, LMNtal: a language model with links and membranes. In Proc. Fifth Int. Workshop on Membrane Computing (WMC 2004), LNCS 3365, Springer, 2005, pp.110-125. (invited lecture)

LMNtal (pronounced "elemental") is a simple language model based on graph rewriting that uses logical variables to represent links and membranes to represent hierarchies. The two major goals of LMNtal are (i) to unify various computational models based on multiset rewriting and (ii) to serve as the basis of a truly general-purpose language covering various platforms ranging from wide-area to embedded computation. Another important contribution of the model is it greatly facilitates programming with dynamic data structures.

40. [pdf] Kazunori Ueda and Norio Kato, The Language Model LMNtal. In Proc. 19th Int. Conf. on Logic Programming (ICLP'03), LNCS 2916, Springer-Verlag, pp.517-518, 2003.

41. [pdf] Kazunori Ueda, A Pure Meta-Interpreter for Flat GHC, A Concurrent Constraint Language.
In Computational Logic: Logic Programming and Beyond (Essays in Honour of Robert A. Kowalski, Part I), A.C. Kakas, F. Sadri (Eds.), Lecture Notes in Artificial Intelligence 2407, Springer-Verlag, 2002, pp.138-161.

This paper discusses the construction of a meta-interpreter of Flat GHC, one of the simplest and earliest concurrent constraint languages.
Meta-interpretation has a long history in logic programming, and has been applied extensively to building programming systems, adding functionalities, modifying operational semantics and evaluation strategies, and so on. Our objective, in contrast, is to design the pair of (i) a representation of programs suitable for code mobility and (ii) a pure interpreter (or virtual machine) of the represented code, bearing networked applications of concurrent constraint programming in mind. This is more challenging than it might seem; indeed, meta-interpreters of many programming languages achieved their objectives by adding small primitives into the languages and exploiting their functionalities. A meta-interpreter in a pure, simple concurrent language is useful because it is fully amenable to theoretical support including partial evaluation.
After a number of trials and errors, we have arrived at treecode, a ground-term representation of Flat GHC programs that can be easily interpreted, transmitted over the network, and converted back to the original syntax. The paper describes how the interpreter works, where the subtleties lie, and what its design implies. It also describes how the interpreter, given the treecode of a program, is partially evaluated to the original program by the unfold/fold transformation system for Flat GHC.

42. [pdf] Yasuhiro Ajiro and Kazunori Ueda, Kima: an Automated Error Correction System for Concurrent Logic Programs. Automated Software Engineering, Vol.9, No.1 (2002), pp.67-94.

We have implemented Kima, an automated error correction system for concurrent logic programs. Kima corrects near-misses such as wrong variable occurrences in the absence of explicit declarations of program properties.
Strong moding/typing and constraint-based analysis are turning out to play fundamental roles in debugging concurrent logic programs as well as in establishing the consistency of communication protocols and data types. Mode/type analysis of Moded Flat GHC is a constraint satisfaction problem with many simple mode/type constraints, and can be solved efficiently. We proposed a simple and efficient technique which, given a non-well-moded/typed program, diagnoses the reasons'' of inconsistency by finding minimal inconsistent subsets of mode/type constraints. Since each constraint keeps track of the symbol occurrence in the program, a minimal subset also tells possible sources of program errors.
Kima realizes automated correction by replacing symbol occurrences around the possible sources and recalculating modes and types of the rewritten programs systematically. As long as bugs are near-misses, Kima proposes a rather small number of alternatives that include an intended program. Search space is kept small because the minimal subset confines possible sources of errors in advance. This paper presents the basic algorithm and various optimization techniques implemented in Kima, and then discusses its effectiveness based on quantitative experiments.

43. Kazunori Ueda, A Close Look at Constraint-Based Concurrency (a tutorial).
In Proc. 17th Int. Conf. on Logic Programming (ICLP'01), Codognet, P. (ed.), Lecture Notes in Computer Science 2237, Springer, 2001, p.9.
PDF file of the talk slides

Constraint-based concurrency, also known as the cc (concurrent constraint) formalism, is a simple framework of concurrency that features (i) asynchronous message passing, (ii) polyadicity and data structuring mechanisms, (iii) channel mobility, and (iv) nonstrictness (or computing with partial information). Needless to say, all these features originate from the use of constraints and logical variables for interprocess communication and data representation. Another feature of constraint-based concurrency is its remarkable stability; all the above features were available essentially in its present form by mid 1980's in concurrent logic programming languages.
Another well-studied framework of concurrency is name-based concurrency represented by the family of pi-calculi, in which names represent both channels and tokens conveyed by channels. Some variants of the original pi-calculus featured asynchronous communication, and some limited the use of names in pursuit of nicer semantical properties. These modifications are more or less related to the constructs of constraint-based concurrency. Integration of constraint-based and name-based concurrency can be found in proposals of calculi such as the gamma-calculus, the rho-calculus and the Fusion calculus, all of which incorporate constraints or name equation into name-based concurrency.
This tutorial takes a different, analytical approach in relating the two formalisms; we compare the roles of logical variables and names rather than trying to integrate one into the other. Although the comparison under their original, untyped setting is not easy, once appropriate type systems are incorporated to both camps, name-based communication and constraint-based communication exhibit more affinities. The examples of such type systems are linear types for the pi-calculus and the mode/linearity/capability systems for Guarded Horn Clauses (GHC). Both are concerned with the polarity and multiplicity of communication, and prescribe the ways in which communication channels can be used. They help in-depth understanding of the phenomena occurring in the course of computation.
The view of constraint-based concurrency that the tutorial intends to provide will complement the usual, abstract view based on ask and tell operations on a constraint store. For instance, it reveals the highly local nature of a constraint store (both at linguistic and implementation levels) which is often understood to be a global, shared entity. It also brings resource-consciousness into the logical setting. This is expected to be a step towards the deployment of cc languages as a common platform of non-sequential computation including parallel, distributed, and embedded computing.

44. [pdf] Kazunori Ueda, Resource-Passing Concurrent Programming.
In Proc. Fourth Int. Symp. on Theoretical Aspects of Computer Software, Kobayashi, N. and Pierce, B. (Eds.), Lecture Notes in Computer Science 2215, Springer, 2001, pp.95-126.
PDF file of the talk slides

The use of types to deal with access capabilities of program entities is becoming increasingly popular.
In concurrent logic programming, the first attempt was made in Moded Flat GHC in 1990, which gave polarity structures (modes) to every variable occurrence and every predicate argument. Strong moding turned out to play fundamental roles in programming, implementation and the in-depth understanding of constraint-based concurrent computation.
The moding principle guarantees that each variable is written only once and encourages capability-conscious programming. Furthermore, it gives less generic modes to programs that discard or duplicate data, thus providing the view of "data as resources." A simple linearity system built upon the mode system distinguishes variables read only once from those read possibly many times, enabling compile-time garbage collection. Compared to linear types studied in other programming paradigms, the primary issue in constraint-based concurrency has been to deal with logical variables and highly non-strict data structures they induce.
In this paper, we put our resource-consciousness one step forward and consider a class of ecological' programs which recycle or return all the resources given to them while allowing concurrent reading of data structures via controlled aliasing. This completely recyclic subset enforces us to think more about resources, but the resulting programs enjoy high symmetry which we believe has more than aesthetic implications to our programming practice in general.
The type system supporting recyclic concurrent programming gives a [-1,+1] capability to each occurrence of variable and function symbols (constructors), where positive/negative values mean read/write capabilities, respectively, and fractions mean non-exclusive read/write paths. The capabilities are intended to be statically checked or reconstructed so that one can tell the polarity and exclusiveness of each piece of information handled by concurrent processes. The capability type system refines and integrates the mode system and the linearity system for Moded Flat GHC. Its arithmetic formulation contributes to the simplicity.
The execution of a recyclic program proceeds so that every variable has zero-sum capability and the resources (i.e., constructors weighted by their capabilities) a process absorbs match the resources it emits. Constructors accessed by a process with an exclusive read capability can be reused for other purposes.
The first half of this paper is devoted to a tutorial introduction to constraint-based concurrency in the hope that it will encourage cross-fertilization of different concurrency formalisms.

45. [ps.gz] Yasuhiro Ajiro and Kazunori Ueda, Kima -- an Automated Error Correction System for Concurrent Logic Programs.
In Proc. Fourth International Workshop on Automated Debugging (AADEBUG 2000).

We have implemented Kima, an automated error correction system for concurrent logic programs. Kima corrects near-misses such as wrong variable occurrences in the absence of explicit declarations of program properties.
Strong moding/typing and constraint-based analysis are turning to play fundamental roles in debugging concurrent logic programs as well as in establishing the consistency of communication protocols and data types. Mode/type analysis of Moded Flat GHC is a constraint satisfaction problem with many simple mode/type constraints, and can be solved efficiently. We proposed a simple and efficient technique which, given a non-well-moded/typed program, diagnoses the "reasons" of inconsistency by finding minimal inconsistent subsets of mode/type constraints. Since each constraint keeps track of the symbol occurrence in the program, a minimal subset also tells possible sources of program errors.
Kima realizes automated correction by replacing symbol occurrences around the possible sources and recalculating modes and types of the rewritten programs systematically. As long as bugs are near-misses, Kima proposes a rather small number of alternatives that include an intended program. Search space is kept small because the minimal subset confines possible sources of errors in advance. This paper presents the basic algorithm and various optimization techniques implemented in Kima, and then discusses its effectiveness based on quantitative experiments.

46. [ps] Kazunori Ueda, Linearity Analysis of Concurrent Logic Programs.
In Proc. International Workshop on Parallel and Distributed Computing for Symbolic and Irregular Applications, Ito, T. and Yuasa, T. (eds.), World Scientific, 2000, pp.253-270.

Automatic memory management and the hiding of the notion of pointers are the prominent features of symbolic processing languages. They make programming easy and guarantee the safety of memory references. For the memory management of linked data structures, copying garbage collection is most widely used because of its simplicity and desirable properties. However, if certain properties about runtime storage allocation and the behavior of pointers can be obtaind by static analysis, a compiler may be able to generate object code closer to that of procedural programs. In the fields of parallel, distributed and real-time computation, it is highly desirable to be able to identify data structures in a program that can be managed without using garbage collection. To this end, this paper proposes a framework of linearity analysis for a concurrent logic language Moded Flat GHC, and proves its basic property. The purpose of linearity analysis is to distinguish between fragments of data structures that may be referenced by two or more pointers and those that cannot be referenced by two or more pointers. Data structures with only one reader are amenable to compile-time garbage collection or local reuse. The proposed framework of linearity analysis is constraint-based and involves both equality and implicational constraints. It has been implemented as part of klint v2, a static analyzer for KL1 programs.

47. [ps] Kazunori Ueda, Concurrent Logic/Constraint Programming: The Next 10 Years.
In The Logic Programming Paradigm: A 25-Year Perspective, K. R. Apt, V. W. Marek, M. Truszczynski, and D. S. Warren (eds.), Springer-Verlag, 1999, pp.53-71.
PDF file of the talk slides

Concurrent logic/constraint programming is a simple and elegant formalism of concurrency that can potentially address a lot of important future applications including parallel, distributed, and intelligent systems. Its basic concept has been extremely stable and has allowed efficient implementations. However, its uniqueness makes this paradigm rather difficult to appreciate. Many people consider concurrent logic/constraint programming to have rather little to do with the rest of logic programming. There is certainly a fundamental difference in the view of computation, but careful study of the differences will lead to the understanding and the enhancing of the whole logic programming paradigm by an analytic approach. As a model of concurrency, concurrent logic/constraint programming has its own challenges to share with other formalisms of concurrency as well. They are: (1) a counterpart of lambda-calculus in the field of concurrency, (2) a common platform for various non-sequential forms of computing, and (3) type systems that cover both logical and physical aspects of computation.

48. [ps] Yasuhiro Ajiro, Kazunori Ueda and Kenta Cho, Error-Correcting Source Code.
In Proc. International Conference on Principles and Practice of Constraint Programming (CP98), Michael Maher and Jean-Francois Puget (eds.), Lecture Notes in Computer Science 1520, Springer-Verlag, 1998, pp.40-54.

We study how constraint-based static analysis can be applied to the automated and systematic debugging of program errors.
Strongly moding and constraint-based mode analysis are turning out to play fundamental roles in debugging concurrent logic/constraint programs as well as in establishing the consistency of communication protocols and in optimization. Mode analysis of Moded Flat GHC is a constraint satisfaction problem with many simple mode constraints, and can be solved efficiently by unification over feature graphs. We have proposed a simple and efficient technique which, given a non-well-moded program, diagnoses the "reasons" of inconsistency by finding minimal inconsistent subsets of mode constraints. Since each constraint keeps track of the symbol occurrence in the program that imposed the constraint, a minimal subset also tells possible sources of program errors. The technique is quite general and can be used with other constraint-based frameworks such as strong typing.
Based on the above idea, we study the possibility of automated debugging in the absence of mode/type declarations. The mode constraints are usually imposed redundantly, and the constraints that are considered correct can be used for correcting wrong symbol occurrences found by the diagnosis. As long as bugs are near-misses, the automated debugger can propose a rather small number of alternatives that include the intended program. Search space is kept small because constraints effectively prune many irrelevant alternatives. The paper demonstrates the technique by way of examples.

49. [ps] Kazunori Ueda and Ryoji Tsuchiyama, Optimizing KLIC Generic Objects by Static Analysis.
In Proc. 11th International Conference on Applications of Prolog, Prolog Association of Japan, 1998, pp.27-33.

The KLIC system has achieved both high portability and extensibility by employing C as an intermediate language and featuring generic objects that allow users to define new classes of data. It is also efficient for an untyped and unmoded language with fine-grained concurrency, but its flexibility incurs runtime overhead that could be reduced by static analysis. This paper studies how constraint-based static analysis and abstract interpretation can be used to reduce dynamic data checking and to optimize loops. We applied the proposed technique to the optimization of floating-point numbers and their arrays. The optimized KL1 programs turned out to be only 34%-70% slower than the comparable C programs.

50. [ps] Kenta Cho and Kazunori Ueda, Diagnosing Non-Well-Moded Concurrent Logic Programs.
In Proc. 1996 Joint International Conference and Symposium on Logic Programming (JICSLP'96), Michael Maher (ed.), The MIT Press, 1996, pp.215-229.

Strong moding and constraint-based mode analysis are expected to play fundamental roles in debugging concurrent logic/constraint programs as well as in establishing the consistency of communication protocols and in optimization. Mode analysis of Moded Flat GHC is a constraint satisfaction problem with many simple mode constraints, and can be solved efficiently by unification over feature graphs. In practice, however, it is important to be able to analyze non-well-moded programs (programs whose mode constraints are inconsistent) and present plausible "reasons" of inconsistency to the programmers in the absence of mode declarations.
This paper discusses the application of strong moding to systematic and efficient static program debugging. The basic idea, which turned out to work well at least for small programs, is to find a minimal inconsistent subset from an inconsistent set of mode constraints and indicate the symbol( occurrence)s in the program text that imposed those constraints. A bug can be pinpointed better by finding more than one overlapping minimal subset. These ideas can be readily extended to finding multiple bugs at once. For large programs, stratification of predicates narrows search space and produces more intuitive explanations. Stratification plays a fundamental role in introducing mode polymorphism as well. Our experiments show that the sizes of minimal subsets are small enough for bug location and do not depend on the program size, which means that diagnosis can be done in almost linear time.

51. [ps] Kazunori Ueda, Experiences with Strong Moding in Concurrent Logic/Constraint Programming.
In Proc. International Workshop on Parallel Symbolic Languages and Systems (PSLS'95), Lecture Notes in Computer Science 1068, Springer-Verlag, Berlin, April 1996, pp.134-153.

Strong moding is turning out to play fundamental roles in concurrent logic programming (or in general, concurrent constraint programming) as strong typing does but in different respects. "Principal modes" can most naturally be represented as feature graphs and can be formed by unification. We built a mode analyzer, implementing mode graphs and their operations by means of concurrent processes and streams (rather than records and pointers). This is a non-trivial programming experience with complicated process structures and has provided us with several insights into the relationship between programming with dynamic data structures and programming with dynamic process structures. The mode analyzer was then applied to the analyzer itself to study the characteristics of the mode constraints it imposed and of the form of large mode graphs. Finally, we show how our framework based on principal moding can be extended to deal with (1) random-access data structures, (2) mode polymorphism, (3) higher-order constructs, and (4) various non-Herbrand constraint systems.

52. [pdf (slides)] Kazunori Ueda, Strong Moding in Concurrent Logic/Constraint Programming.
Tutorial given at the Twelfth International Conference on Logic Programming, Kanagawa, Japan, June 1995.

53. [ps] Kazunori Ueda, I/O Mode Analysis in Concurrent Logic Programming.
In Theory and Practice of Parallel Programming, Lecture Notes in Computer Science 907, Springer-Verlag, Berlin, pp.356-368, May 1995.

This paper briefly reviews concurrent logic programming and the I/O mode system designed for the concurrent logic language Flat GHC. The mode system plays fundamental roles both in programming and implementation in almost the same way as type systems do but in different respects. It provides us with the information on how data are generated and consumed and thus the view of "data as resources". It statically detects bugs resulting from ill-formed dataflow and advocates the "programming as wiring" paradigm. Well-modedness guarantees the safety of unification, the basic operation in concurrent logic programming. Information on the numbers of access paths to data can be obtained by slightly extending the framework, which can be used for compile-time garbage collection and the destructive update of structures.

54. [ps] Kazunori Ueda, Moded Flat GHC for Data-Parallel Programming.
In Proc. FGCS'94 Workshop on Parallel Logic Programming, ICOT, Tokyo, pp.27-35, December 1994.

Concurrent logic languages have been used mainly for the (parallel) processing of rather irregular symbolic applications. However, since concurrent logic languages are essentially general-purpose, they should be applicable to problems with regular structures and their data-parallel processing as well. This paper studies the possibility of massively parallel processing in concurrent logic programming, focusing on arrays and its data-parallel processing.

55. [ps] Kazunori Ueda and Masao Morita, Moded Flat GHC and Its Message-Oriented Implementation Technique.
New Generation Computing, Vol.13, No.1, pp.3-43, November 1994.

Concurrent processes can be used both for programming computation and for programming storage. Previous implementations of Flat GHC, however, have been tuned for computation-intensive programs, and perform poorly for storage-intensive programs (such as programs implementing reconfigurable data structures using processes and streams) and demand-driven programs. This paper proposes an optimization technique for programs in which processes are almost always suspended. The technique compiles unification for data transfer into message passing. Instead of reducing the number of process switching operations, the technique optimizes the cost of each process switching operation and reduces the number of cons operations for data buffering.

The technique is based on a mode system which is powerful enough to analyze bidirectional communication and streams of streams. The mode system is based on mode constraints imposed by individual clauses rather than on global dataflow analysis, enabling separate analysis of program modules. The introduction of a mode system into Flat GHC effectively subsets Flat GHC; the resulting language is called Moded Flat GHC. Moded Flat GHC programs enjoy two important properties under certain conditions: (1) reduction of a goal clause retains the well-modedness of the clause, and (2) when execution terminates, all the variables in an initial goal clause will be bound to ground terms. Practically, the computational complexity of all-at-once mode analysis can be made almost linear with respect to the size n of the program and the complexity of the data structures used, and the complexity of separate analysis is higher only by O(log n) time. Mode analysis provides useful information for debugging as well.

Benchmark results show that the proposed technique well improves the performance of storage-intensive programs and demand-driven programs compared with a conventional native-code implementation. It also improves the performance of some computation-intensive programs. We expect that the proposed technique will expand the application areas of concurrent logic languages.

56. [ps] Kazunori Ueda and Masao Morita, Message-Oriented Parallel Implementation of Moded Flat GHC.
New Generation Computing, Vol.11, No.3-4, pp.323-341, July 1993.

We proposed in [Ueda and Morita 1990] a new, message-oriented implementation technique for Moded Flat GHC that compiled unification for data transfer into message passing. The technique was based on constraint-based program analysis, and significantly improved the performance of programs that used goals and streams to implement reconfigurable data structures. In this paper we discuss how the technique can be parallelized. We focus on a method for shared-memory multiprocessors, called the shared-goal method, though a different method could be used for distributed-memory multiprocessors. Unlike other parallel implementations of concurrent logic languages which we call process-oriented, the unit of parallel execution is not an individual goal but a chain of message sends caused successively by an initial message send. Parallelism comes from the existence of different chains of message sends that can be executed independently or in a pipelined manner. Mutual exclusion based on busy waiting and on message buffering controls access to individual, shared goals. Typical goals allow last-send optimization, the message-oriented counterpart of last-call optimization. We are building an experimental implementation on Sequent Symmetry. In spite of the simple scheduling currently adopted, preliminary evaluation shows good parallel speedup and good absolute performance for concurrent operations on binary process trees.

57. [pdf] Kazunori Ueda, The Fifth Generation Project: Personal Perspectives.
Commun. ACM, Vol.36, No.3, pp.65-76, March 1993.
(The paper is avaiable from ACM Ditital Library as a joint article with Kazuhiro Fuchi, Robert Kowalski, Koichi Furukawa, Ken Kahn, Takashi Chikayama and Evan Tick under the title "Launching the New Era".

This article reviews the design process of KL1, the kernel language for the Parallel Inference Machine in the Fifth Generation Computer Systems (FGCS) project. The FGCS project chose logic programming as its central principle, but the shift to concurrent logic programming started very early in the R&D of the kernel language, which provoked many discussions and many criticisms. The author proposed a concurrent logic language GHC in December 1984 as a basis of KL1. GHC was recognized as a stable platform with a number of justifications, and the design of KL1 started to converge. The article focuses on the history of the design of KL1, and also presents personal principles behind the language design and perspectives on the future of GHC and KL1.

58. [pdf] Kazunori Ueda and Takashi Chikayama, Design of the Kernel Language for the Parallel Inference Machine.
The Computer Journal, Vol.33, No.6, pp.494-500, December 1990.

59. [ps] Kazunori Ueda, Designing a Concurrent Programming Language.
In Proceedings of an International Conference organized by the IPSJ to Commemorate the 30th Anniversary (InfoJapan'90), Information Processing Society of Japan, October 1990, pp.87-94.

This paper reviews the design and the evolution of a concurrent programming language Guarded Horn Clauses (GHC). GHC was born from a study of parallelism in logic programming, but turned out to be a simple and flexible concurrent programming language with a number of nice properties. We give both an abstract view of GHC computation based the notion of transactions and a concrete view, namely an operational semantics, based on reductions. Also, we discuss in detail the properties of GHC such as its atomic operations, which have much to do with the design of GHC.

60. [pdf] Kazunori Ueda, Parallelism in Logic Programming.
In Information Processing 89, Ritter, G.X. (ed.), North-Holland, pp.957-964, August 1989.

61. [ps] Kazunori Ueda, Guarded Horn Clauses: A Parallel Logic Programming Language with the Concept of a Guard.
ICOT Technical Report TR-208, Institute for New Generation Computer Technology (ICOT), Tokyo, 1986.
Revised version in Programming of Future Generation Computers, Nivat, M. and Fuchi, K. (eds.), North-Holland, Amsterdam, pp.441-456, 1988.

62. [ps] Kazunori Ueda, Theory and Practice of Concurrent Systems: The Role of Kernel Language in the FGCS Project.
In Proc. Int. Conf. on Fifth Generation Computer Systems 1988 (FGCS'88), ICOT, Tokyo, November 1988, pp.165-166.

63. [ps] Kazunori Ueda and Koichi Furukawa, Transformation Rules for GHC Programs.
In Proc. Int. Conf. on Fifth Generation Computer Systems 1988 (FGCS'88), ICOT, Tokyo, November 1988, pp.582-591.

Transformation rules for (Flat) GHC programs are presented, which refine the previous rules proposed by one of the authors (Furukawa et al. 1987). The rules are based on unfolding/folding and are novel in that they are stated in terms of idempotent substitutions with preferred directions of bindings. They are more general than the old rules in that no mode system is assumed and that the rule of folding is included. The presentation of the rules suggests that idempotent substitutions with preferred directions of bindings are an appropriate tool for modeling information in (concurrent) logic programming. A semantic model is given which associates a multiset of goals with the set of possible finite sequences of transactions (via substitutions) with the multiset. A transformation preserves the set of transaction sequences that are made without the risk of the failure of unification. The model also deals with anomalous behavior such as the failure of unification and deadlock, so it can be shown with the same model that the transformation cannot introduce those anomalies. Applications of the transformation technique include the fusion of communicating parallel processes. Some familiarity with GHC is assumed.

64. [ps] Kazunori Ueda, Making Exhaustive Search Programs Deterministic, Part II.
In Proc. Fourth Int. Conf. on Logic Programming, The MIT Press, Cambridge, Mass., U.S.A., pp.356-375, May 1987.

65. [ps] Kazunori Ueda, Making Exhaustive Search Programs Deterministic.
New Generation Computing, Vol.5, No.1, pp.29-44, 1987.

66. [pdf] Kazunori Ueda and Takashi Chikayama, Concurrent Prolog Compiler on Top of Prolog.
In Proc. 1985 Symp. on Logic Programming, IEEE Computer Society, pp.119-126, July 1985.

A Concurrent Prolog compiler, whose target language is (sequential) Prolog, was implemented in Prolog. The object program obtained can further be compiled into machine codes by a Prolog compiler. Due to the similarity among the source, target and implementation languages, the compiler and the runtime support were small and very rapidly developed. Benchmark tests showed that (twice) compiled Concurrent Prolog programs ran 2.7 to 4.4 times faster and 2.7 to 5.3 times slower than comparable Prolog programs running on the interpreter and compiler, respectively, of the same Prolog system. After these experiments, the Concurrent Prolog compiler was modified to obtain a compiler of the new parallel logic programming language, GHC (Guarded Horn Clauses), and almost the same efficiency was achieved. These compilers will serve for practice of parallel logic programming.

67. [pdf] Kazunori Ueda, Guarded Horn Clauses.
ICOT Technical Report TR-103, Institute for New Generation Computer Technology (ICOT), Tokyo, 1985.
Revised version in Proc. Logic Programming '85, Wada, E.(ed.), Lecture Notes in Computer Science 221, Springer-Verlag, Berlin Heidelberg New York Tokyo, 1986, pp.168-179.
Also in Concurrent Prolog: Collected Papers, Shapiro, E.Y. (ed.), The MIT Press, Cambridge, 1987, pp.140-156.

A set of Horn clauses augmented with a guard' mechanism is shown to be a simple and yet powerful parallel logic programming language.
(The first paper on the concurrent logic programming langauge GHC (Guarded Horn Clauses).)

68. [pdf] Kazunori Ueda and Takashi Chikayama, Efficient Stream/Array Processing in Logic Programming Languages.
In Proc. Int. Conf. on Fifth Generation Computer Systems 1984 (FGCS'84), ICOT, Tokyo, pp. 317-326, November 1984.

The Concurrent Prolog predicate for merging n input streams is investigated, and a compilation technique tor getting its efficient code is presented. Using the technique, data are transferred with a delay independent of n. Furthermore, it is shown that the addition and the removal of an input stream can be done within an average time of 0(1). The predicate for distributing data on an input stream to n output streams can also be realized as efficiently as n-ary merge. The compilation technique for the distribute predicate can further be applied to the implementation of mutable arrays that allow constant-time accessing and updating. Although the efficiency stated above could be achieved by a sophisticated compiler, the codes should be provided directly by the system to get rid of the bulk of source programs and the time required to compile them.

## II. Doctoral Thesis

1. Kazunori Ueda, Guarded Horn Clauses. [pdf (1013898 bytes)]
The University of Tokyo, 1986.

This thesis introduces the programming language Guarded Horn Clauses which is abbreviated to GHC. Guarded Horn Clauses was born from the examination of existing logic programming languages and logic programming in general, with special attention paid to parallelism.

The main feature of GHC is its extreme simplicity compared with the other parallel programming languages. GHC is a restriction of a resolution-based theorem prover for Horn-clause sentences. The restriction has two aspects: One is the restriction on data flow caused by unification, and the other is the introduction of choice nondeterminism. The former is essential for a general-purpose language and it also provides GHC with a synchronization primitive. The latter is required by the intended applications which include a system interacting with the outside world. What is characteristic with GHC is that all the restrictions have been imposed as the semantics given to the sole additional syntactic construct, guard.

Although Guarded Horn Clauses can be classified into the family of logic programming languages, it has close relationship to other formalisms including dataflow languages, Communicating Sequential Processes, and functional languages for multiprocessing. Except for the lack of higher-order facilities, GHC can be viewed as a generalization of these frameworks. The simplicity and generality of GHC will make it suitable for a standard not only of parallel logic programming languages but of parallel programming languages. Moreover, it is simple enough to be regarded as a computation model as well as a programming language.

Attention has always been paid to the possibility of efficient implementation during the design stage of GHC. We showed that stream merging and distribution which are expected to be heavily used can be implemented with the same time-complexity as the time-complexity of many-to-one communication in procedural languages. Furthermore, we made available an efficient compiler-based implementation of a subset of GHC on top of Prolog.

GHC has lost the completeness as a theorem prover deliberately, not as a result of compromise. Nevertheless, it can be used for efficient implementation of exhaustive solution search for Horn-clause programs. We showed how to automatically compile a Horn-clause program for exhaustive search into a GHC program.

## III. Books and Proceedings

1. Kazunori Ueda (ed.), Programming Languages and Systems: 8th Asian Symposium, APLAS 2010. Lecture Notes in Computer Science 6461, Springer-Verlag, 2010.

2. Herbert Kuchen and Kazunori Ueda (eds.), Functional and Logic Programming---5th International Symposium on Functional and Logic Programming, FLOPS 2001. Lecture Notes in Computer Science 2024, Springer-Verlag, 2001.

3. R. K. Shyamasundar and Kazunori Ueda (eds.), Advances in Computing Science---ASIAN'97. Lecture Notes in Computer Science 1345, Springer-Verlag, 1997.

4. Logic Programming: Proceedings of the 1991 International Symposium
Vijay Saraswat and Kazunori Ueda (eds.), The MIT Press, Cambridge, Mass, U.S.A., October 1991.

## IV. Technical Reports, Workshop Papers, etc.

1. [pdf (slides)] Kazunori Ueda, HydLa: A High-Level Language for Hybrid Systems. Presented at the Shonan Meeting on Hybrid Systems: Theory and Practice, Seriously, April 2012.

We have been working on the design and implementation of HydLa, a high-level modeling language for hybrid systems. Its objectives and features include a constraint-based (as opposed to automata-based) formalism, proper handling of uncertanties, and nondeterministic execution. The talk will describe an overview of the language with live demonstration of our prototype implementation.

2. [pdf] Kazunori Ueda and Norio Kato, Programming with Logical Links: Design of the LMNtal language. In Proc. Third Asian Workshop on Programming Languages and Systems (APLAS 2002), pp.115-126, 2002.

We propose LMNtal, a simple language model based on the rewriting of hierarchical graphs that use logical variables to represent links. The two major goals of the model are (i) to unify various computational models based on multiset rewriting and (ii) to serve as the basis of a truly general-purpose language covering various platforms ranging from wide-area to embedded computation. Another important contribution of the model is it greatly facilitates programming with dynamic data structures.

3. Kazunori Ueda, The Mode System of Moded Flat GHC.
In Proc. DIMACS workshop on Specification of Parallel Algorithms,
Guy E. Blelloch, K. Mani Chandy, Suresh Jagannathan (eds.),
Volume 18 of DIMACS Series in Discrete Mathematics and Theoretical Computer Science, 1994.
4. [pdf] Kazunori Ueda, Introduction to Guarded Horn Clauses.
ICOT Technical Report TR-209, Institute for New Generation Computer Technology (ICOT), Tokyo, November 1986.

This paper informally introduces a programming language Guarded Horn Clauses (GHC) through program examples. GHC is a paral- lel programming language devised from investigation of the basic framework and practice of logic programming. It has introduced the guard construct with sim- ple semantics into logic programming to express interacting processes. A GHC program naturally expresses parallelism inherent in the original problem. The simple and uniform framework of GHC should be far easier to understand than the constructs of conventional parallel programming languages. We explain GHC in comparison with Prolog, the best-known logic programming language.

Last update: May 10, 2017
ueda -at- ueda -dot- info -dot- waseda -dot- ac -dot- jp