Bart Massey

An implementation of narrowing strategies

This paper describes an implementation of narrowing, an essential component of implementations of modern functional logic languages. These implementations rely on narrowing, in particular on some optimal narrowing strategies, to execute functional logic programs. We translate functional logic programs into imperative (Java) programs without an intermediate abstract machine. A central idea of our approach is the explicit representation and processing of narrowing computations as data objects. This enables the implementation of operationally complete strategies (i.e., without backtracking) or techniques for search control (e.g., encapsulated search). Thanks to the use of an intermediate and portable represen tation of programs, our implementation is general enough to be used as a common back end for a wide variety of functional logic languages.

Search-based Job Scheduling for Parallel Computer Workloads

To balance performance goals and allow administrators to declaratively specify high-level performance goals, we apply complete search algorithms to design on-line job scheduling policies for workloads that run on parallel computer systems. We formulate a hierarchical two-level objective that contains two goals commonly placed on parallel computer systems: (1) minimizing the total excessive wait; (2) minimizing the average slowdown. Ten monthly workloads that ran on a Linux cluster (IA-64) from NCSA are used in our simulation of policies. A wide range of measures are used for performance evaluation, including the average slowdown, average wait, maximum wait, and new measures based on excessive wait. For the workloads studied, our results show that the best search-based scheduling policy (i.e., DDS/lxf/dynB) reported here simultaneously beats both FCFS-backfill and LXF-backfill, each roughly providing a lower bound on maximum wait and the average slowdown, respectively, among backfill policies

Longitudinal analysis of long-timescale open source repository data

One of the more unique features of open source software development is the continuity of projects over large time scales and incremental development efforts. For this reason, the open development process provides an interesting environment for investigation of the software development process. The problems of data collection and analysis of two particular long-running repositories, the X Window System and the Nickle Programming Language, are considered here as instructive examples. The use of uniform software tools (CVS/RCS) with open formats and interfaces makes it possible to collect data that provide unique analysis opportunities.

Improving the Efficiency of Non-Deterministic Computations

Abstract Non-deterministic computations greatly enhance the expressive power of functional logic programs, but are often computationally expensive. We analyze a programming technique that improves the time and memory efficiency of some non-deterministic computations. This technique relies on the introduction of a new symbol into the signature of a program. This symbol may be treated either as a polymorphic defined operation or as an overloaded constructor. Our programming technique may save execution time, by reducing the number of steps of a computation. The technique may also save memory, by reducing the number of terms constructed by a computation. We give some examples of the application of our technique, address its soundness and completeness, and informally reason about its impact on the efficiency of computations.

Fast perfect weighted resampling

We describe an algorithm for perfect weighted-random resampling of a population with time complexity O(m + n) for resampling to inputs to produce n outputs. This algorithm is an incremental improvement over standard resampling algorithms. Our resampling algorithm is parallelizable, with linear speedup. Linear-time resampling yields notable performance improvements in our motivating example of sequential importance resampling for Bayesian particle filtering.

Using Portable Radios to Operate Mobile Data Terminals

Recently, voice controlled software has begun to be used in public safety applications. The motivation for these efforts is to improve officer safety by allowing the officer to operate the police cruisers electronics and mobile data terminal (MDT) using voice commands so he can continue to focus on the road when driving. Project 54, which is representative of the work in this area, allows an officer to use a microphone and push-to-talk button on the steering wheel for voice interaction with vehicles MDT and electronics. This projects focus is on the officers access to the cruisers electronics and computer from outside the vehicle while responding to an unknown-risk call. Tactically, it is absolutely imperative that the officer be able to maintain visual contact of the potential threat. Accessing the electronics and MDT by voice avoids the division of attention that occurs if the officer is forced to look at a keyboard or computer screen. While Project54 appears to have been successful in making driving safer, its microphone technology limits its use to the interior and immediate vicinity of the cruiser making it unsuitable for tactical situations. We are investigating the use of the officers off-the-shelf, portable radio packset to interact with the cruisers MDT from outside the car. We do this by using the officers packset to communicate with a transceiver attached to the MDT which also has an audio and control interface to the cruisers in-car mobile radio. Transmissions received from the officers packset by the MDTs transceiver are processed by a software module running on the MDT. Ordinary radio transmissions are forwarded to the in-car mobile radio, which rebroadcasts them. Voice commands, such as a computer aided dispatch (CAD) command are executed and responses converted from text to speech and rebroadcast by the MDTs transceiver to the officers packset. A significant feature of this approach is that none of the hardware needs to be replaced. The only requirement is the addition of a transceiver (another packset like the one the officer carries), some cabling and software. Further, by simply changing channels on the officers packset and replacing the cables between the MDT and the mobile radio with a microphone and speaker, the MDT and the communications equipment can revert to their original configuration.