Fast Methods for Finding Multiple Effective Influencers in Real Networks.

We present scalable first hitting time methods for finding a collection of nodes that enables the fastest time for the spread of consensus in a network. That is, given a graph = () and a natural number , these methods find vertices in that minimize the sum of hitting times (expected number of steps of random walks) from all remaining vertices. Although computationally challenging for general graphs, we exploited the characteristics of real networks and utilized Monte Carlo methods to construct fast approximation algorithms that yield near-optimal solutions.

An Algorithm for Identifying Optimal Spreaders in a Random Walk Model of Network Communication.

In a model of network communication based on a random walk in an undirected graph, what subset of nodes (subject to constraints on the set size), enables the fastest spread of information? The dynamics of spread is described by a process dual to the movement from informed to uninformed nodes. In this setting, an optimal set minimizes the sum of the expected first hitting times (), of random walks that start at nodes outside the set. Identifying such a set is a problem in combinatorial optimization that is probably NP hard.

Constructing sequence alignments from a Markov decision model with estimated parameter values.

Current methods for aligning biological sequences are based on dynamic programming algorithms. If large numbers of sequences or a number of long sequences are to be aligned, the required computations are expensive in memory and central processing unit (CPU) time. In an attempt to bring the tools of large-scale linear programming (LP) methods to bear on this problem, we formulate the alignment process as a controlled Markov chain and construct a suggested alignment based on policies that minimise the expected total cost of the alignment.

The Role of Rendering in the Competence Project in Measurement Science for Optical Reflection and Scattering.

Computer rendering is used to simulate the appearance of lighted objects for applications in architectural design, for animation and simulation in the entertainment industry, and for display and design in the automobile industry. Rapid advances in computer graphics technology suggest that in the near future it will be possible to produce photorealistic images of coated surfaces from scattering data. This could enable the identification of important parameters in the coatings manufacturing process that lead to desirable appearance, and to the design of virtual surfaces by visualizing prospective coating formulations once their optical properties are known.