​MIT says new algorithm speeds up answers to complex problems

The algorithm aims to hone in on a target answer by estimating values of each unknown parameter in a problem.

Researchers at the Massachusetts Institute of Technology say they have developed an algorithm that will speed up the time to find an answer by as much as 200 times.

The MIT algorithm can be applied to any computational model. The algorithm aims to hone in on a target answer by estimating values of each unknown parameter in a problem.

In a statement, MIT described the algorithm:

The algorithm may be thought of as a shrinking bull's-eye that, over several runs of a model, and in combination with some relevant data points, incrementally narrows in on its target: a probability distribution of values for each unknown parameter.

Gartner analysts told CIOs in October that algorithms would become the most important assets within enterprises. These algorithms would be developed commercially, bought and sold on exchanges and seen as key assets.

Academics will also have a big role in developing core algorithms.

Related special reports: AI and the Future of Business | AI, Automation, and Tech Jobs | IT Budgets 2016: A CIO's Guide | Innovation: How to be a World-Changer

The findings were published in the American Statistical Association journal.

What the algorithm ultimately tries to do is speed up the sampling process and offer an alternative to Monte Carlo analysis, a well-known statistical sampling model. The approach can be used on any complex model to find the probability distribution and run models with various inputs. The algorithm will also use relevant data to provide values for unknown parameters.

The research was funded in part by the Department of Energy. The algorithm was used to simulate the movement of sea ice in Antarctica. The problem included 24 unknown parameters. The algorithm will be used next to model combustion systems for supersonic jets.