They reveal methods that double speed and precision in correction of quantum errors with GPU.
NVIDIA quantum advances, do Bitcoin security put in check?
Nvidia, one of the world’s largest technological companies, published on September 30 a report in which she details how she has been exploring quantum computing.
With this step, it adds to other entities that already experience with quantum infrastructuresuch as Ionq, IBM, the HSBC Bank or even the US stock and values commission. UU. (SEC).
From Nvidia they pointed out that “quantum computing promises to transform industries”, but also assured that this development It depends on solving three key problems: “Error correction, simulations of cubits designs and optimization of circuit compilation tasks.”
Although there is still no quantum computer with sufficient capacity to break the cryptography that protects Bitcoin and cybersecurity in general, the advances of NVIDIA in quantum computing They offer clues about how that scenario could evolve.
Your error correction research, circuit optimization and quantum systems accelerate the development of more stable and scalable processors, potentially reducing times and costs to reach more powerful machines.
This does not mean that Bitcoin is vulnerable today, but technologies such as those developing Nvidia could shorten the distance to computers capable of challenging current cryptographic systems.
Correction of quantum errors, the first target of Nvidia
The Cubits They are the basic units of quantum computing and, unlike classic bits, they can be in several states at the same time. This property allows mass parallel calculations, but also makes them very sensitive to noise and errors.
It is as if the quantum cubits were a choir of voices that sing in harmony to solve complex problems in unison, but the slightest whisper of the environment can disregard and cause errors.
Precisely to counteract those “whispers” of the environment, the technique of correction of quantum errors (QECfor its acronym in English) uses a set of maneuvers that the Nvidia document details as essential to handle noise:
The QEC is the way in which researchers distill thousands of noisy physical cubits in a handful of logical qubits without noise, decoding data in real time, detecting and correcting errors as they arise.
NVIDIA quantum report.
Not only does it need a lot And the system ceases to be useful.
To optimize this task of correcting quantum computing errors, Nvidia collaborated with the quantum software laboratory of the University of Edinburgh (United Kingdom).
Together, they developed a new decoding method called Autodec, through the nominated bookstore “CUDA-Q QEC” of Nvidia.
Decoding methods in this area are techniques designed to interpret and correct the errors that arise in the cubits due to noiseensuring that quantum calculations are precise and reliable.
Thus, according to the report, the Autodec method He managed to “duplicate the speed and accuracy” of the decoding of quantum errors, Thanks to the parallel decoding functionality in GPU (graphic processing unit) that distributes the task in many processing centers at the same time.
At the same time, Nvidia collaborated with the company Quera to develop a decoder based on artificial intelligence (AI).
These AI models achieved “an increase of 50 times in the decoding rate, along with better precision,” according to the report.
The logic behind is previously training the AI models with a lot of heavy computing and, once trained, Execute much faster inferences in real timesomething essential for the highest distance codes that is expected to use in future quantum computers, they say from Nvidia.
Second Focus of Nvida: Improve quantum circuits
Another Nvidia work front was the optimization of the quantum circuit compilationthat is, the Translation of a quantum algorithm to available physical hardware.
This process implies assigning logical cubits (a type of cubit) to physical positions in the chip, a very complex problem that resembles organizing pieces of a puzzle with millions of possible combinations.
In collaboration with the Q-CTRL and Oxford Quantum Circuit (OCQ) company, NVIDIA created an accelerated GPU mechanism that offered “Up to 600 times more speed” In compilation tasks, according to the report.
The last quantum link studied by Nvidia
Finally, Nvidia guided her efforts to the Numerical simulation of quantum systemsa key step to understand and predict how quantum processors and thus design more stable cubits.
In collaboration with the University of Sherbrooke and Amazon Web Services (AWS), «the researchers saw up to 4,000 times more performance » When studying a Cubit Transon Coupled to a resonator, a type of critical experiment to improve the quality of the cubits.
A transcon is a type of superconductor cubit designed to be less sensitive to electrical noiseand the resonator is a circuit that acts as a “resonance box” to control and measure the cubit.
By accelerating error correction, circuit compilation and device simulation, the company seeks to reduce bottlenecks Key in the “race” to build practical quantum computers.