http://hdl.handle.net/1721.1/3651 http://dspace.mit.edu:80/retrieve/3313 http://hdl.handle.net/1721.1/3651 http://hdl.handle.net/1721.1/35807 Finite Energy Survey Propagation for Constraint Satisfaction Problems Chieu, Hai Leong The Survey Propagation (SP) algorithm [1] has recently been shown to work well in the hard region for random K-SAT problems. SP has its origins in sophisticated arguments in statistical physics, and can be derived from an approach known as the cavity method, when applied at what is called the one-step replica symmetry breaking level. In its most general form, SP can be applied to general constraint satisfaction problems, and can also be used in the unsatisfiable region, where the aim is to minimize the number of violated constraints. In this paper, we formulate the SP-Y algorithm for general constraint satisfaction problems, applicable for minimizing the number of violated constraints. This could be useful, for example, in solving approximate subgraph isomorphism problems. Preliminary results show that SP can solve a few instances of induced subgraph isomorphism for which belief propagation failed to converge. Fri, 29 Dec 2006 22:58:59 GMT http://hdl.handle.net/1721.1/35779 Provably Efficient Adaptive Scheduling for Parallel Jobs He, Yuxiong Hsu, Wen Jing Leiserson, Charles E. Scheduling competing jobs on multiprocessors has always been an important issue for parallel and distributed systems. The challenge is to ensure global, system-wide efficiency while offering a level of fairness to user jobs. Various degrees of successes have been achieved over the years. However, few existing schemes address both efficiency and fairness over a wide range of work loads. Moreover, in order to obtain analytical results, most of them require prior information about jobs, which may be difficult to obtain in real applications. This paper presents two novel adaptive scheduling algorithms -- GRAD for centralized scheduling, and WRAD for distributed scheduling. Both GRAD and WRAD ensure fair allocation under all levels of workload, and they offer provable efficiency without requiring prior information of job's parallelism. Moreover, they provide effective control over the scheduling overhead and ensure efficient utilization of processors. To the best of our knowledge, they are the first non-clairvoyant scheduling algorithms that offer such guarantees. We also believe that our new approach of resource request-allotment protocol deserves further exploration. Specifically, both GRAD and WRAD are O(1)-competitive with respect to mean response time for batched jobs, and O(1)-competitive with respect to makespan for non-batched jobs with arbitrary release times. The simulation results show that, for non-batched jobs, the makespan produced by GRAD is no more than 1.39 times of the optimal on average and it never exceeds 4.5 times. For batched jobs, the mean response time produced by GRAD is no more than 2.37 times of the optimal on average, and it never exceeds 5.5 times. Fri, 29 Dec 2006 22:58:59 GMT http://hdl.handle.net/1721.1/35778 How to Do a Million Watchpoints: Efficient Debugging Using Dynamic Instrumentation Zhao, Qin Amarasinghe, Saman P. Rabbah, Rodric M. Rudolph, Larry Wong, Weng Fai Application debugging is a tedious but inevitable chore in any software development project. An effective debugger can make programmers more productive by allowing them to pause execution and inspect the state of the process, or monitor writes to memory to detect data corruption. The latter is a notoriously difficult category of bugs to diagnose and repair especially in pointer-heavy applications. The debugging challenges will increase with the arrival of multicore processors which require explicit parallelization of the user code to get any performance gains. Parallelization in turn can lead to more data debugging issues such as the detection of data races between threads. This paper leverages the increasing efficiency of runtime binary interpreters to provide a new concept of Efficient Debugging using Dynamic Instrumentation, or EDDI. The paper demonstrates for the first time the feasibility of using dynamic instrumentation on demand to accelerate software debuggers, especially when the available hardware support is lacking or inadequate. As an example, EDDI can simultaneously monitor millions of memory locations, without crippling the host processing platform. It does this in software and hence provides a portable debugging environment. It is also well suited for interactive debugging because of the low associated overheads. EDDI provides a scalable and extensible debugging framework that can substantially increase the feature set of standard off the shelf debuggers. Fri, 29 Dec 2006 22:58:59 GMT http://hdl.handle.net/1721.1/35777 Collaborative Data Publishing and Searching System Ooi, Beng Chin Yu, Bei Li, Guoliang In this paper, we present a folksonomy-based collaborative data publishing and searching system. The system accepts data objects described with user-created metadata, called data units. The system supports flexible structure on the data units, and places no restrictions on the vocabulary used. We devise a generic table model for storing and representing the data units of various structures. We propose a framework for managing the data units and providing browsing, searching and querying services over them. We present our current approaches and discuss relevant research issues. Fri, 29 Dec 2006 22:58:59 GMT