Gavriel Yarmish

A distributed, scaleable simplex method

We present a simple, scaleable, distributed simplex implementation for large linear programs. It is designed for coarse-grained computation, particularly, readily available networks of workstations. Scalability is achieved by using the standard form of the simplex rather than the revised method. Virtually all serious implementations are based on the revised method because it is much faster for sparse LPs, which are most common. However, there are advantages to the standard method as well. First, the standard method is effective for dense problems. Although dense problems are uncommon in general, they occur frequently in some important applications such as wavelet decomposition, digital filter design, text categorization, and image processing. Second, the standard method can be easily and effectively extended to a coarse grained, distributed algorithm. Such an implementation is presented here. The effectiveness of the approach is supported by experiment and analysis.

Revisiting novice programmer errors

Although programmer errors have been investigated, only a limited range of error types typically made by novices have been scrutinized. In this paper we present an expanded classification of the types of errors considered in previous research. In particular, problems which require the use of more difficult program constructs such as nested loops, arrays, recursion and functions have been somewhat neglected. We hope this paper will encourage other researchers to further analyze the types of errors advanced novices will make and the types of misunderstandings which underlie such errors.

A description and study of intermediate student programmer errors

To date there has been considerable investigation into the study of novice programmer errors. The research has analyzed both syntactic and semantic errors. However, the next level of programmers, who make more sophisticated errors, the internmediate level programmers, have been somewhat neglected. In this paper, we focus on the nature of the errors which intermediate level programmers make. The basis of our study is the semantic approach. Here, we the study problems which require more difficult program constructs such as nested loops, arrays, recursion, and functions.

Parallel Clustering of Protein Structures Generated via Stochastic Monte Carlo

The problem of efficient clustering of candidate protein structures into a limited number of groups is addressed. Such clustering can be expensive and is rarely used in practice due to its computational complexity. We present a parallel algorithm for the efficient clustering of proteins into groups. The input consists of thousands of candidate proteins structures that have been stochastically generated Monte-Carlo style. The first step is to make a Root Mean Square Deviation (RMSD) comparison matrix. The second step is to utilize parallel processors to calculate a hierarchal cluster of these proteins based on the RMSD matrix and using the Lance-Williams update algorithm. The final output is a Dendrogram of clusters. We have implemented our algorithm and have found it to be scalable.