Google has released a library of programs that are designed to work with quantum neural networks. It can be used for real quantum computers and their simulations. The product is free and is open source.
the New library was named TensorFlow Quantum (TFQ). It complements the widely known tool TensorFlow designed to work with artificial intelligence on classical computers.
the Project was implemented by the Google Quantum AI team together with the students of the University of Waterloo, and companies Alphabet X and Volkswagen.
“News.Science” (nauka.vesti.ru) told in detail what a quantum computer the qubits. Recall that a qubit, unlike a classical bit, can exist not only in the state “0” or “1” but also in quantum superposition (“mixture”) of these States. Potentially it provides quantum computers the tremendous computing power. But at the moment the system of qubits is highly vulnerable to interference and errors.
TFQ provides tools for working with the basic components of quantum computing: qubits, the quantum measurement procedure and so forth. The system can be used for programming real quantum computers and their simulations in classic cars. To address the latter issue, Google has also released a simulator of quantum computing circuits qsim.
As the experts of the company, the world is entering an era of noisy quantum processors medium-scale (Noisy Intermediate-Scale Quantum or NISQ). Such devices will have 50-100 qubits (this is a “medium scale”). It is expected that they will be able to something that is inaccessible to classical computers. At least with the latter is already quite difficult to model a system with this many qubits.
on the other hand, the qubits are still running quite unstable (and therefore processors called “noisy”). While currently proposed algorithms of error correction will be effective when the computing power of millions, not tens of qubits.
Thus, the possibility of quantum computers of the era NISQ is limited to their vulnerability to errors. For greater efficiency, they should work in one system with classical processors. And TensorFlow Quantum provides all possibilities for this, because it is a “child” TensorFlow well adapted to the integration of processors with different instruction sets.
in addition, TFQ integrated with the platform Cirq. It is also the development of the Google Quantum AI that is designed to work with quantum computers.
the New platform will help in the development of programs for quantum computers.Illustration Pixabay.
Artificial intelligence in the service of quantum computing
Recall that the system of the era NISQ vulnerable to noise and computational errors. How to protect yourself from this evil?
Google Engineers propose a solution. In fact, they say, we have before us a common goal: to filter data from random noise and to extract useful information. Such problems often arise, for example, when recognition, and are solved using artificial intelligence.
“Conduct.Science” told of how studying classical neural network. About the same works, and quantum, although, of course, it has its own specifics.
Quantum neural network quantum data reads by the same measuring procedure. The result of this procedure are quite common sets of numbers. Handling them to provide a classic (non-quantum) neural network with deep learning.
According to developers, in a pair of classical and quantum neural network will extract the maximum information from the “tainted” quantum data.
Details for professionals fromsuggested in the Preprint scientific articles published on the website arXiv.org.
By the way, before “News.Science” (nauka.vesti.ru) wrote about the promised by the company Honeywell the world’s most powerful quantum computer.
This is a syndicated post. Read the original post at Source link .