Hou-Duo Qi is a professor in optimization at the School of Mathematics, the University of Southampton. He mainly works on optimization problems over correlations and distances among multivariate variables. He develops both theory and algorithms that strike a balance between theoretical computational complexity and practical computational speed aiming to deal with large scale data. One such a problem of the prototype is computing the nearest correlation matrix arising from finance. His recent work is on Euclidean distance matrix optimization that has applications in sensor network localization, molecular conformation, dimension reduction and data visualization.
Prior to joining Southampton University in 2004, He was QEII (Queen Elizabeth II) Fellow at the University of New South Wales, supported by the Australian Research Council. Professor Qi graduated from Peking University in Statistics (1990), and obtained PhD in Operational Research and Control Theory in Chinese Academy of Sciences (1996).
Distances among a number of items decide their relative positions. Global Positioning System (GPS) is one of such examples. The procedure of converting distances information to locations involves a branch of mathematics called distance geometry. It is known that we only need some of pairwise distances in order to determine the intrinsic structure among items due to the rigidity theory in mathematics. However, practical applications only have partial noisy distances. For example, in a larger network, a node only communicates with its neighbors. The current of research of Houduo Qi is to reconstruct the true structure of networks determined by the available noisy distances. His is main tool is optimization, which optimizes certain quantities of interest while keeping the distances information as accurate as possible. His research has found applications in sensor network localization, molecular conformation, dimension reduction, embedding on spheres and data visualization.