Here, There, Everywhere: NVIDIA Platform Accelerates Quantum Circuit Simulation Ecosystem


Quantum computing promises scientific leaps — simulating molecules of atoms for drug discovery, for occasion — in the around foreseeable future.

Handling exponentially additional information and facts than today’s computers, quantum personal computers harness the physics that govern subatomic particles to make parallel calculations. Teams worldwide in academia, market and nationwide labs are exploring quantum personal computers and algorithms. Several operate quantum circuit simulations to accelerate their investigation timelines.

NVIDIA introduced at GTC 2021 the cuQuantum program advancement package to pace quantum circuit simulations managing on GPUs. Early do the job suggests that cuQuantum provides orders of magnitude speedups for circuit simulations, paving the way for Nobel Prize-profitable breakthroughs of tomorrow.

The anticipated arrival of quantum supremacy — when a quantum personal computer solves a challenge a classical personal computer simply cannot in a acceptable time — nonetheless, stays an open up debate. Nonetheless to be solved is decoherence, or falling out of quantum states, as a restricting factor that corrupts the functionality of quantum circuits.

Also, quantum computing depends on quantum bits, or qubits, that can be , one or each — and quite a few much more qubits are required to mistake accurate for decoherence.

The cuQuantum SDK accelerates quantum circuit simulators to assist scientists layout better quantum desktops and confirm success, design hybrid-classical systems, and learn much more best quantum algorithms.

It also provides instruments for developers to use to the solutions of their preference, supporting different ways this kind of as the state vector strategy or the tensor network method.

Condition Vector Strategy

Researchers at the Jülich Supercomputing Centre have harnessed the state vector technique to simulate actual physical realizations of quantum personal computers on GPUs. Their benchmark checks, reviewed at GTC, confirmed a 25x speedup on GPU clusters in contrast with CPU-based techniques.

A non-public corporation, QCWare, has been publishing papers across the simulation and application of quantum computing domains. Working together, NVIDIA and QCWare have demonstrated powerful proof that for the quantum approximate optimization algorithm, at 20 qubits, the effectiveness variation is important.

QCWare and NVIDIA have shown orders of magnitude additional general performance jogging quantum simulations on NVIDIA GPUs as opposed to CPUs alone.

Caption: QCWare and NVIDIA have demonstrated orders of magnitude much more functionality operating quantum simulations on NVIDIA GPUs in contrast to CPUs by yourself

A solitary NVIDIA DGX A100 with eight NVIDIA A100 80GB Tensor Core GPUs is able of simulating up to 36 qubits, providing orders of magnitude speedup more than a twin-socket CPU server on primary state vector simulations.

Apart from Jülich and QCWare, corporations that use condition vector simulators operating on NVIDIA GPUs include IBM, Oxford Nanopore, Amazon Net Products and services and the NVIDIA AI Know-how Center.

Tensor Community Technique

Tensor community simulations are a more recent approach that works by using a lot less memory but a lot more computation than point out vector techniques.

Tapping into tensor network techniques, NVIDIA and Caltech accelerated a major quantum circuit simulator with cuQuantum working on NVIDIA A100 Tensor Core GPUs. This set up produced a sample in 9.three minutes on the NVIDIA Selene supercomputer from a complete-circuit simulation of the Google Sycamore circuit. This feat was only just lately predicted to consider times on millions of CPU cores.

In addition to Caltech, all those using tensor network simulators involve Alibaba, Amazon Internet Expert services, Argonne National Lab and Oak Ridge National Lab.

Density Matrix Sims

Scientists from the Pacific Northwest National Laboratory, Lehigh University and Washington Condition College have made a new multi-GPU programming methodology, called MG-BSP. They used it to build a density matrix quantum simulator.

The investigation group shown the simulation of 1 million standard gates in 94 minutes on an NVIDIA DGX-two, far further circuits than have earlier been revealed, according to the group.

The benefits uncovered that their density matrix simulator is additional than 10x quicker than condition vector quantum simulators on GPUs and other platforms, according to their paper.

Join our early curiosity checklist to continue to be informed about the most up-to-date updates on cuQuantum. 

Inside of IBM Quantum computing procedure pictured. Impression credit history: IBM

Leave a comment

Your email address will not be published.