Algorithm Draws Scientific Laws from Experiment Data

Experts at the Cornell University (CU) have created a remarkable, new algorithm for computers, which is able to derive basic natural laws from raw scientific data. That is to say, rather than people clogging around a table over coffee and determining why, for example, gravity exists, the entire task will be performed automatically. It would then be up to the scientists supervising the algorithm to act as peer-reviewers, in very much the same way experts do when a scientific paper is submitted for publishing in a peer-reviewed journal.

It is very interesting to note that the computer running the program does not have any prior scientific knowledge, and that it is able to extract soon-to-be-established laws from events it records by simply searching for patterns in what is happening around it. Newton acted in the same way when he discovered gravity – he noted a series of repeating phenomena, and began to ask questions about them. In the end, he created his laws on the matter. This is basically what the new algorithm is capable of.

The end-product of the computer is not a fortune cookie-like piece of paper, on which the new scientific law is written. Rather, the algorithm produces a series of “invariants,” a number of mathematic statements derived from the observed phenomena that always remain true. Experts analyzing the data can then conclude if these invariants have any merit in setting the foundation for new laws to govern things as movement, behavior, and so on.

“Even though it looks like it's changing erratically, there is always something deeper there that is always constant. That's the hint to the underlying physics. You want something that doesn't change, but the relationship between the variables in it changes in a way that's similar to [what we see in] the real system,” CU associate professor of mechanical and aerospace engineering Hod Lipson, who is also the author of a new paper detailing the find, published in the April 3rd issue of the journal Science, explains.