A large part of science involves searching for the analytical expression of physical laws that govern various systems, which often depend on various symmetries and invariants. Attempts to automate the process have in the past proved unsuccessful, but now two researchers seem to have succeeded where others have failed.
Michael Schmidt and Hod Lipson of Cornell University have developed computer code that looks at data from physical systems ranging from simple harmonic oscillators to chaotic double pendula and proceeds to infer the appropriate Hamiltonians, Lagrangians and conservation laws – and all without any prior knowledge. "Might this process diminish the role of future scientists?" the researchers ask at the end of their paper. "Quite the contrary: scientists may use processes such as this to help focus on interesting phenomena more rapidly and to interpret their meaning." The software could provide valuable tools to help scientists work out the dynamics underlying all sorts of otherwise-confusing systems.
Further reading
M Schmidt and H Lipson 2009 Science 324 81.