Qubit Pharmaceuticals and Sorbonne University achieve a major scientific breakthrough by simulating quantum calculations at more than 40 qubits on conventional computers

Qubit Pharmaceuticals, a deeptech company specializing in the discovery of new drug candidates through molecular simulation and modeling accelerated by hybrid HPC and quantum computing, announces a major scientific breakthrough after achieving quantum computations simulating 40 qubits with its new Hyperion-1 emulator.  

"This is an exact simulation of 40 logic qubits carried out at very high velocity, which is an unprecedented achievement in the application of quantum computation, in particular to quantum chemistry," confirms Jean-Philip Piquemal, Professor at Sorbonne University and Director of the Theoretical Chemistry Laboratory (Sorbonne University/CNRS), co-founder and Scientific Director of Qubit Pharmaceuticals, and head of the team that developed Hyperion-1.

Such a level of performance places Qubit Pharmaceuticals among the world's leading actors in quantum computing, all the more so as it was achieved without approximation and with the highest level of fidelity, i.e. without error (or "noise", to use the prevailing expression in quantum physics) and in a very short time, close to what one would expect from a true quantum computer. This performance was achieved in partnership with Sorbonne University's Theoretical Chemistry Laboratory, and the calculations were carried out in just a few hours on GENCI’s Jean Zay HPC/IA converged supercomputer on 16 computing nodes (128 GPUs¹ A100 NVIDIA) hosted and operated by IDRIS computing center (CNRS), on which the Hyperion-1 emulator was developed.

The ultimate objective: select a drug candidate in half the time

This achievement reinforces Qubit Pharmaceuticals' ambition to become the industry reference in molecular modeling-based drug discovery. The result of academic research carried out by internationally renowned scientists² in France and the United States, Qubit Pharmaceuticals models molecules and simulates their interactions to identify more effective and safer drug candidates. The aim is to halve the time needed to select and optimize a candidate of interest, and more than 10-fold the investment required. This process requires immense computing capacities, available today with supercomputers and multiplied tomorrow with quantum computers.

Key benefits of Hyperion-1, the new emulator from Qubit Pharmaceuticals

• Revolutionary simulations: Hyperion-1 is capable of achieving 40 qubit (and more) converged simulations in production without error or noise, setting new standards of precision in quantum research.
• Massively parallel processing capability: thanks to its emulation of a quantum computer on a classical architecture accelerated by GPUs(1), Hyperion-1 offers exceptional precision and massively parallel processing capability capable of concentrating classical computing power.
• Exceptional performance: Hyperion-1 has demonstrated outstanding performance in high-speed simulations of complex quantum circuits of up to 40 qubits, using moderate computing resources, i.e. 16 Jean Zay nodes each accelerated by 8 NVIDIA A100 80GB GPUs. This increased speed makes it possible to converge quantum chemistry simulations requiring hundreds of algorithm iterations.
• Advanced technology: Featuring optimized internal code based on a mathematical library that is proprietary and agnostic, i.e. able to use any type of supercomputer, Hyperion-1 uses parallel acceleration via MPI and takes full advantage of the new generation of NVIDIA DGX-A100 accelerated nodes for enhanced performance.

The Hyperion-1 quantum emulation opens up new perspectives on the technology of tomorrow

In the ever-evolving landscape of quantum computing, a critical gap persists between machines with a limited number of perfect qubits and those with a large number of qubits but laden with error (noise and instability). Hyperion-1, with the velocity and accuracy of its calculations, is a testament to the immense possibilities that lie ahead. Its capabilities demonstrate what will be possible in the wider landscape of quantum emulation and quantum computing. Qubit Pharmaceuticals is proud of Hyperion-1's potential, not only as a proprietary tool, but also as a symbol of perfect emulation, fostering a new era of technological innovation with far-reaching implications for sectors such as pharmaceuticals, finance, encryption, and many others.

Robert Marino, PDG de Qubit Pharmaceuticals, states: “These quantum chemistry calculations on 40 exact qubits far exceed the performance achieved to date in Europe, and place Qubit Pharmaceuticals on the same footing as some of the top American tech giants. This breakthrough enables us to carry out in a few hours calculations that traditionally take several months.”

Jean-Philip Piquemal, Professor at Sorbonne University and Director of the Theoretical Chemistry Laboratory (Sorbonne University/CNRS), co-founder and Chief Scientific Officer at Qubit Pharmaceuticals, states: “Hyperion-1 allows quantum state simulation while benefiting from the stability of classical computers, thus avoiding the errors inherent in today's quantum computers. Thanks to the GPUs in our machines and GENCI's infrastructure, we are able to develop and validate new quantum algorithms applied to drug discovery - a field of research with real public utility.”

Élisabeth Angel-Perez, Vice President of Research and Innovation at Sorbonne University: "Sorbonne University is a community of talent, and it's also a commitment: a commitment to supporting innovation stemming from French research. And it's because we've given ourselves the necessary resources to develop an ecosystem for the transfer of expertise and innovation that we've been able to accompany and support genuine nuggets such as Qubit Pharmaceuticals. Science must be at the service of society and its well-being. That's the dynamic we're supporting alongside the researchers who make Sorbonne University so rich.”

⁽¹⁾ GPU= Graphics Processing Unit
⁽²⁾ Louis Lagardère (Sorbonne Université et CNRS), Matthieu Montes (CNAM), Jean-Philip Piquemal (Sorbonne Université et CNRS), Jay Ponder (Washington University in St Louis), Pengyu Ren (University of Texas at Austin).