Imperial mathematician sheds new light on 50 year old algorithm

An Imperial mathematician has found a new way of formulating a 50 year old algorithm, used when describing the world using mathematical models. It is anticipated that the proposed technique, published this week in the Proceedings of the National Academy of Sciences (PNAS), will pave the way for greatly accelerating the calculations involved when making predictions about the behaviour of complex systems in many different areas of science and engineering. "Often, scientists are interested in things they can't measure directly, either because it is too expensive or even because it is physically impossible to do so," explained Ben Calderhead, from the Department of Mathematics at Imperial College London. "For example, we can't measure every biological process simultaneously occurring in the human body, nor can we directly measure many of the physical properties of very distant objects in the universe. However, given a smaller collection of related measurements and a mathematical model, scientists can make predictions about these unobserved components via a process called Bayesian statistical inference." He added: "Rather than coming up with a single mathematical description of some observed natural phenomenon, scientists can consider the collection of plausible descriptions that are consistent with the measurements.