Artificial Intelligence infused in an algorithm can do wonders and can solve tons of mysteries that were initially a bit difficult to crack. It took astronomers a little more than expected but now they are able to resolve the most complex mystery. A machine- learning algorithm has made the entire picture clear and has been successful to find a whole new aspect to quantum physics.
This avant-garde discovery is something that could expose the world to new quantum physics laws. What this machine actually figured out, is that Sun be placed at the center of the Solar System. It came to this conclusion on the basis of monitoring the movements of the Sun and Mars from Earth.
What this could result in, are new laws of physics that could reformulate quantum mechanics, by looking out of patterns in large data sets.
This algorithm was devised Ion Zurich by Physicist Renato Renner along with his collaborators at the Swiss Federal Institute of Technology (ERH). What they wanted to design an algorithm which could fetch large data sets in just a few basic formulae. This was in emulation of succinct and successful pre-existing equations. This was carried out by creating a neural network design, which was inspired by the human brain structure.
Traditional neurons work by coming together in collaboration to identify and recognize objects. These objects include both sounds and images. Similarly what the physicists did to their newly formed neurons was identity and the features and attributes it sees, in to a mathematical ‘nodes’. These nodes render neurons in the AI terminology. But then these nodes instead of just passing away the information in just a couple of steps, the neural network delivers and expands its knowledge over a thousand and million other nodes. This manner in which the information spread is what difficult to make out
What Renner and his team of geniuses did, was that they came up with a ‘lobotomized’ neural network. This lobotomized neural network comprises of two separate sub networks which came together to be joined by yet another set of links. The function of the first sub network would be to earn new data by being exposed to it. The second sub network would make use of this experience and come up with predictions and possibilities and test them too. Due to the sensitivity of the links connecting both sides, the first network passed information onto the other in a sorted and condensed manner.
The initial step was giving this network of nodes, the mock information about the activity of Mars and the Sun in the sky, which have been recorded from the Earth. What’s revealing about Mars is it’s certain peculiar aspects, which involves going ‘retrograde’ and going backwards in its course.
For decades, the scientists were under the illusion that Earth was the centre of all the Universe and gave a very limp explanation about Mars’ activities. This they explained further by saying that the planets made moves in small circles, called epicycles. Somewhere around 1500,
Nicolaus Copernicus came forth and revealed that with simple formulas in place, the movements could be apprehended, to unveil if the Earth and all the planets were revolving around the Sun.
Similarly, based on Copernicus’ suggestion, and came up with their neural network in accordance with the Copernicus styled formulas for Mars’ trajectory.
Renner emphasized on the need of a diligent human brain to come up with and see to the interpretations of how planets revolved around the sun, even though it was the algorithm who devised the formula.
A roboticist at Columbia University, New York, named God Lipson, looks at these works and revelations of being of pressing matters of importance as they are able to bring out and describe the individual functionalities of a physical system. He comments, “I think that these kinds of techniques are our only hope of understanding and keeping pace with increasingly complex phenomena, in physics and beyond,”.
With the massive success that Renner and his team were able to bag with the help of algorithms and AI, their long term goal is to concoct machine-learning technologies that would render useful to physicists and help them make out and resolve contradictions in quantum mechanics.
Moreover, Renner and his visionary team is looking for a way that their neural network would not just rely on the data obtained through experiment but also perform its own experiments and test out their hypothesis.
Renner says, It’s possible that the current way [quantum mechanics is] formulated is in some way just a historical artefact,”.
Source – Nature