1992-93 ACADEMIC ACTIVITIES REPORT NAME: TIMOTHY J. HICKEY DEPARTMENT: Computer Science I. INSTRUCTIONAL ACTIVITY (Summer 1992, Fall 1992, Spring 1993) a) Semester Course Number and Title Class Contact Enrollment Hours UnderG. Graduate TYP Weekly 1. Autumn 1992 CS140 Logic Programming 6 4 7 2. Autumn 1992 Math2 TYP 3 8 3. Autumn 1992 CS98a Independent Study 1 1 0 4. Autumn 1992 CS200a Readings in CoSci 1 0 1 5. Autumn 1993 CS300a Masters Study 1 0 1 6. Autumn 1992 CS406 Dissertation Rsh 5 0 1 1. Spring 1993 CS150 Compiler Design 2. Spring 1993 CS150 Compiler Design 6 10 7 3. Spring 1993 Math2 TYP 3 9 4. Spring 1993 CS98b Independent Study 1 1 0 5. Spring 1993 CS200b Readings 1 0 1 6. Spring 1993 CS300b Masters Study 1 0 1 7. Spring 1993 CS406 Dissertation Rsh 5 0 1 b) Advising (total contact hours per week: 3) 1) number of general or freshman advisees: 21 2) number of undergraduate departmental advisees: 12 3) number of graduate advisees: 3 c) Please describe your involvement in the direction of senior theses, graduate dissertations and other student research projects. 1. Advisor of 1 Ph.D. students (D. Smith). Co-advisor of 1 Ph.D. student (S. Mudambi) with J. Cohen. 2. Member of Defense Committee for Senior Honors Thesis in Economics (Deborah Shufrin) 3. On Dissertation committee of one Ph.D. student (Marc Feeley) II. RESEARCH, PUBLICATIONS, ARTISTIC CREATION (use additional page if necessary.) a) Describe current research activities or work in progress: My current research is in two areas: Massively Parallel Programming, and Logic Programming. In Logic Programming I am currently pursuing several lines of research: INTERVAL ARITHMETIC, CLP(BNR), AND PARALLELISM. In this project we are studying an exciting new approach to solving non-linear arithmetic constraints and we are incorporating this constraint solver into a CLP language, CLP(BNR), which is being implemented on sequential and parallel machines. The approach is to represent partial solutions to a system of equations by associating an interval [x_lo,x_hi] to each variable and to use the constraints to narrow these intervals. In practice this technique has been demonstrated to yield solutions to non-linear optimization problems exceptionally quickly, *and* because of the logical basis of the constraint solver, one actually obtains a proof that the given solution is optimal (modulo some rather weak assumptions). We have recently submitted a paper for publication which presents an approach to parallelizing the constraint solver and provides initial benchmarks suggesting that one may be able to attain linear speedup with this approach on certain classes of problems. DESIGNING CLP LANGUAGES FOR NATURAL LANGUAGE PROCESSING. I am still in the intial phases of this project. The goal is to work with computational linguists to determine the types of constraints that arise naturally in natural lanugage processing and to design constraint domains which model these constraints, and CLP languages based on these domains which can be used to specify and implement natural language processing tools (e.g., parsers). In Massively Parallel Programming I have two current projects: PARALLEL IMPLEMENTATIONS OF DIVIDE AND CONQUER ALGORITHMS In Massively Parallel Programming I am working on several projects with George Mou. The main thrust of our work is to develop languages for specifying divide and conquer algorithms on massively parallel machines. We have spent much of the last year examining the class of k-dimensional meshes. Our current joint research plan is to study the problem of designing and implementing high level languages for specifying D&C algorithms. ANALYSIS OF MIMD PROGRAMS. I am also investigating the problem of analyzing the average case execution time of programs executing on MIMD machines. The fundamental difficulty is in dealing with the non-algebraicity of the MAX operator on distributions. b) Manuscript (s) or artistic work accepted for publication (List journal or publisher and anticipated publication date.) 1. T. Hickey, ``Functional Constraints in CLP Languages,'' to appear in "Constraint Logic Programming", (eds. Alain Colmerauer, F. Benhamou), MIT Press, August 1993. c) Publications since June, 1992 (with inclusive page reference for articles). Please use standard form: author(s) or editor(s), title, number of pages, publisher, location, date. 1. Z.G. Mou, A. Huang, T. Hickey, ``Parallel Recurrence Transformation,'' Proceedings of the DAGS Parallel Computing Symposium, pp. 63-73, Hanover, NH, July, 1992. 2. Z.G. Mou, C. Constantinescu, T. Hickey, "Optimal Mappings of Divide-and-Conquer Algorithms to Mesh Connected Architectures", Proceedings of the Int'l. Computer Symposium, pp. 273-284, Taiwan, Decemeber, 1992, 3. Z.G. Mou, X. Wang, T. Hickey, "Divide-and-Conquer Algorithms with Recursive Broadcast Communication on Reconfigurable Arbitrary Dimensional Mesh," Sixth SIAM Conference on Parallel Processing for Scientific Computing, pp. 784-787, March 1993. d) Artistic creation (please describe) III. AWARDS AND HONORS (dates) Grant Support 1. NSF research grant in Logic Programming CCR-9115326, co-PI with J. Cohen. 2. NSF research grant in Microanalysis CCR-9207964, co-PI w J. Cohen, Z.G. Mou, IV. PROFESSIONAL ACTIVITIES OUTSIDE THE UNIVERSITY (lectures, activities in professional societies, legislative testimony, paid consulting, equity arrangements, etc.) 1. Program Committee member for the International Conference on Logic Programming. V. PARTICIPATION IN DEPARTMENTAL ACTIVITIES AND ADMINISTRATION 1. Seminar coordinator 2. Participated in the Senior Position search 3. Participated in revision of the department curriculum 4. Participated in the NSF CISE Infrastructure grant 5. Participated in selection of graduate students VI. PARTICIPATION IN UNIVERSITY ACTIVITIES, COMMITTEES, ETC. 1. Member of Wien Advisory Committee 2. Member of the Summer Odyssey Advisory Committee 3. Department Liason to Admissions Department.