Hi Alice. 🔮
Admin I really wanted to touch the topic of quantum mechanics but didn't really want to blow your mind for I still need you🙃

But recently a study came out about how machine learning can help in quantum mechanics 🧬. And I want to tell about it briefly (the material will be particularly incomplete and without many of those details for better understanding read the article below). It is divided into two parts, 1) without technical details and 2) with a little of technical details.
Here we go!

Part 1) Structure without the details.

The usual components are good. But often we can hardly imagine how they will help us in quantum mechanics, and quantum computers will appear soon, but they are not there yet.
Alas already now people creating new quantum states in the laboratory may not be able to understand them.

But by machine learning it is possible to get classical data to explain what is going on and to understand quantum mechanics more deeply.
This paper builds on the classical shadow formalism and develops classical machine learning algorithms with polynomial time for quantum problems .
We will touch on the prediction of classical quantum states. Specifically, the family of .
Hamiltonians, where the Hamiltonian H(x) depends smoothly on m real parameters (denoted by x).
The AI is trained on a set of training data consisting of sample values of x, each accompanied by the corresponding classical shadow for the ground state p(x) H(x). These training data can be obtained either in a classical simulation or by moh favorite quantum experiments.

In the prediction step, the AI algorithm predicts a classical representation of p(x) for new values of x other than those in the training data. The ground state
properties can then be estimated using the predicted classical representation.
This learning algorithm is effective as long as the ground state properties to be predicted do not change too rapidly as a function of x
Indeed, sufficient upper bounds on the gradient,and can be obtained for any family of Hamiltonians in any finite spatial dimension if the property of interest is the expectation of the sum of several observables.
Effective transformation of a quantum state into a classical one.
The classical shadow of a quantum state, constructed by measuring a very small number of copies of the state, can be used to predict many properties of the state with a strict performance guarantee guarantee .
After training on the data obtained in quantum experiments, the classical AI model predicts a classical representation of the ground state p(x) of the Hamiltonian H(x) for the parameters x spanning the entire phase.
Such a representation makes it possible to obtain estimates of the properties of p(x), avoiding the need to perform classical calculations or quantum experiments.
After training, the AI obtains a classical representation of the quantum state and predicts the phase from which the state
taken. To predict the ground states, the AI obtains p(x) for each value of x sampled during training. For predicting quantum matter phases, the training data consists of classical
representations of quantum states accompanied by labels identifying the phase to which each state belongs with small average error, where the amount of training data and the classical computation time are polynomial
m and linear in the size of the system...
fuuuh hope it is clear .
Roughly speaking we took the data from the quantum experiments and we left them for further study .🔆
This concludes the first stage and I think we can move on to the more technical part.
#quantum_mechanics #neural_networks