Robotics discover alternative physics

Robotics at Columbia University discover alternative physics

Latent motifs of a frame colored by physical state variables. Credit: Boyuan Chen / Columbia Engineering

Energy, mass and velocity. These three variables make up Einstein’s iconic equation E = MC2. But how did Einstein learn about these concepts in the first place? A preliminary step to understanding physics is to identify the relevant variables. Without the concept of energy, mass, and velocity, not even Einstein could discover relativity. But can such variables be detected automatically? Doing so can greatly speed up scientific discovery.

That’s the question researchers at Columbia Engineering are asking a new artificial intelligence program. The program is designed for monitoring physical phenomena through a Video camera, then try to find the minimum set of basic variables that fully describe the observed dynamics. The study was published on July 25 in computational natural sciences.

The researchers began feeding the system raw video footage of phenomena they already knew the answer to. For example, they fed a video of a swinging double pendulum that is known to have exactly four “state variables” – angle and angular velocity from both arms. After a few hours of analysis, the AI ​​produced the answer: 4.7.






The picture shows a chaotic dynamic system swaying in motion. The work aims to identify and extract the minimum state variables needed to describe such a system directly from high-dimensional video footage. Credit: Yinuo Qin / Columbia Engineering

“We thought this answer was close enough,” said Hood Lipson, director of the Creative Machines Laboratory in the Department of Mechanical Engineering, where the work was primarily done. “Especially since all the AI ​​could access was raw video footage, without any knowledge of physics or engineering. But we wanted to know what the variables actually were, not just how many.”

The researchers then proceeded to visualize the actual variables identified by the program. Extracting the variables themselves was not easy, as the program cannot describe them in any intuitive way that can be understood by humans. After some investigation, it turned out that two of the variables chosen by the program correspond loosely to the angles of the arms, but the other two variables remain a mystery.

“We tried to relate the other variables to anything and everything we could think of: angular and linear velocities, kinetic velocities and potential energy, and various combinations of known quantities. A set of four variables, because they made good predictions, “but we didn’t yet understand the mathematical language he was speaking,” he explained.

After validating a number of other physical systems with known solutions, the researchers fed videos of systems to which they didn’t know the explicit answer. In the first videos, she appeared as an “air dancer” swaying in front of a local used car yard. After a few hours of analysis, the program returned eight variables. A video of the lava lamp also produced eight variants. Then they fed a flame video from a holiday fireplace episode, and the show brought back 24 variables.

A particularly interesting question was whether the set of variables was unique to each system, or whether a different set was produced each time the program was restarted.

“I’ve always wondered, if we ever met an intelligent alien race, would they discover the same laws of physics that we have, or would they describe the universe differently?” Lipson said. “Perhaps some phenomena seem vaguely complex because we are trying to understand them using the wrong set of variables. In experiments, the number of variables was the same every time the AI ​​restarted, but the specific variables were different each time. So yes, there are alternative approaches. to describe the universe and it is entirely possible that our choices may not be perfect.”

Researchers believe this type of AI could help scientists discover complex phenomena for which theoretical understanding doesn’t keep pace with vast amounts of data — fields ranging from biology to cosmology. “While we used video data in this work, any type of matrix data source can be used — for example radar arrays, or DNA arrays, for example,” explained Kwang Huang, PhD, who co-authored the paper.

The work is part of Chiang Du, a professor of mathematics at the Lipson and Fu Foundation, who has long been interested in creating algorithms that can turn data into scientific laws. Previous software systems, such as Lipson’s and Michael Schmidt’s Eureqa, could extract free physical laws from experimental data, but only if the variables were previously defined. But what if the variables are not yet known?

Lipson, who is also the James and Sally Scapa Professor of Innovation, argues that scientists may misinterpret or fail to understand many phenomena simply because they do not have a good set of variables to describe the phenomenon.

Lipson noted: “For thousands of years, people have known that objects move quickly or slowly, but only when the concept of velocity and acceleration was formally defined did Newton discover his famous law of motion F = MA.” The variables describing temperature and pressure must be identified before the laws of thermodynamics can be formulated, and so in every corner of the scientific world. Variables are a precursor to any theory.

“What other laws do we miss just because we don’t have Variablesasked Doe, who co-led the work.

The paper was also co-authored by Sunand Raghupathi and Ishaan Chandratreya, who helped collect data for the trials.


Astronomers discover dozens of new variable stars


more information:
Boyuan Chen et al, Automated detection of baseline variables hidden in experimental data, computational natural sciences (2022). DOI: 10.1038 / s43588-022-00281-6

the quote: Robotics Discover Alternative Physics (2022, July 26) Retrieved July 27, 2022 from https://phys.org/news/2022-07-roboticists-alternative-physics.html

This document is subject to copyright. Notwithstanding any fair dealing for the purpose of private study or research, no part may be reproduced without written permission. The content is provided for informational purposes only.