How Supercomputers Are Unlocking Chemical Mysteries
The precise behavior of electrons determines everything from the color of a rose to the effectiveness of a life-saving drug, yet modeling these interactions with perfect accuracy has remained one of science's most elusive goals.
Imagine trying to predict the exact properties of a new material or drug molecule without ever stepping into a laboratory. For decades, computational chemists have pursued this dream, yet they've faced a fundamental barrier: the exponential complexity of quantum mechanical equations. The very laws that govern behavior at the atomic scale become computationally prohibitive for all but the simplest systems.
The amount of information needed to describe a quantum system grows exponentially with its size.
We remain in the Noisy Intermediate-Scale Quantum (NISQ) era, where quantum resources are too scarce and error-prone for large-scale chemical simulations1 .
This challenge arises because the amount of information needed to describe a quantum system grows exponentially with its size. As renowned physicist Richard Feynman observed, simulating quantum systems efficiently requires a quantum computer itself—a concept he proposed in 19822 . While quantum hardware has advanced significantly, we remain in the Noisy Intermediate-Scale Quantum (NISQ) era, where quantum resources are too scarce and error-prone to support large-scale chemical simulations1 .
In a fascinating twist, scientists are now turning this challenge on its head by using classical supercomputers to emulate quantum computers—essentially creating quantum simulators that can model chemical systems with unprecedented accuracy.
This approach doesn't replace quantum computing but provides a crucial bridge toward its practical application while allowing researchers to develop and test quantum algorithms for the chemistry problems of tomorrow.
To understand why quantum computing holds such promise for chemistry, we need to consider the fundamental challenge of simulating quantum systems. When chemists want to predict how molecules will behave, they need to solve the Schrödinger equation—a mathematical formula that describes how quantum particles like electrons interact.
7.3 Trillion
A complete active space problem of just 24 electrons and 24 orbitals corresponds to a diagonalization problem of this size1 .
The problem is sheer computational scale. A complete active space problem of just 24 electrons and 24 orbitals corresponds to a diagonalization problem of size 7.3 trillion1 . Even with today's most powerful supercomputers, exact solutions are impossible for all but the smallest molecular systems. As a result, chemists must rely on approximations that sacrifice accuracy for feasibility.
Quantum computational chemistry emerges from the recognition that quantum computers could potentially simulate quantum systems much more efficiently than classical computers9 . The field exploits quantum algorithms specifically designed to extract electronic structure information from molecular systems. Methods like the Variational Quantum Eigensolver (VQE) and Quantum Phase Estimation (QPE) have shown particular promise for calculating molecular energies and properties9 .
"The pursuit for chemical accuracy in numerical simulations of quantum many-body systems is a longstanding problem since the computational complexity grows exponentially with the system size," researchers noted in a recent breakthrough paper1 .
This exponential wall has forced the field to seek alternative pathways—including emulating quantum computers on classical supercomputers.
In 2023, a team of researchers achieved what many considered a landmark feat in quantum emulation: they successfully simulated a quantum computer with up to 1000 qubits for chemical calculations, reaching a performance of 216.9 PFLOP/s on the cutting-edge Sunway supercomputer1 . This accomplishment set a new state-of-the-art for quantum computing emulation in quantum chemistry and demonstrated the potential of massively parallel classical computing to bridge the quantum resource gap.
At the heart of their approach lies a mathematical framework called Matrix Product States (MPS), which provides an efficient way to represent quantum states without the prohibitive memory requirements of exact simulation1 . MPS is particularly effective for simulating quantum systems with limited entanglement, making it well-suited for many chemical applications.
The researchers combined MPS with Density Matrix Embedding Theory (DMET), which breaks large molecular systems into smaller, manageable fragments1 . This hybrid approach allowed them to study practical chemical problems, including the torsional barrier of ethane and protein-ligand interactions—key challenges in drug discovery and materials science.
The team faced a fundamental constraint: the exponential memory requirements of quantum simulation. To overcome this, they developed sophisticated optimization strategies for the most computationally intensive operations—tensor contractions and singular value decomposition (SVD)1 .
On the Sunway supercomputer's unique architecture, featuring 390 cores per processor, they implemented specialized techniques including:
To streamline tensor operations
To reduce memory access overhead
These optimizations allowed their simulator to achieve remarkable efficiency, with their customized SVD implementation becoming more than 60 times faster than non-optimized versions for typical matrix sizes1 .
| System/Platform | Qubit Count | Key Achievement | Application |
|---|---|---|---|
| Sunway (MPS-VQE) | 1000 | Largest quantum chemistry emulation | Protein-ligand interactions |
| Sunway (fully converged VQE) | 92 | Complete VQE convergence | Torsional barriers |
| Previous record (iridium complexes) | 72 | Largest VQE emulation at time | Excited states of complexes |
| DMET-VQE hybrid | 16 | Fragment-based approach | C18 molecule |
The breakthrough in quantum emulation rests on several sophisticated technologies working in concert. Understanding these components helps illuminate how classical computers can mimic quantum behavior so effectively.
The VQE algorithm is a hybrid quantum-classical approach that uses both quantum and classical computing resources to find the minimum eigenvalue of a molecule's Hamiltonian—essentially calculating its ground state energy9 . This method is particularly valuable in the NISQ era because it's more resilient to noise than alternatives like Quantum Phase Estimation.
MPS provides a compact representation of quantum states by decomposing them into a network of tensors1 . This approach dramatically reduces memory requirements for simulating quantum systems, particularly one-dimensional arrangements or systems with limited entanglement.
DMET is an embedding technique that divides large molecular systems into smaller fragments, solves each fragment individually, then reassembles the solution1 . This "divide and conquer" strategy makes otherwise intractable problems manageable.
The Sunway supercomputer's architecture, with its 390 cores per processor and specialized memory hierarchy, provided the necessary computational power1 . The researchers' optimized algorithms for this system were crucial to achieving their record-breaking performance.
| Technology | Function | Advantage |
|---|---|---|
| Matrix Product States (MPS) | Compact representation of quantum states | Reduces exponential memory requirements |
| Density Matrix Embedding Theory (DMET) | Fragments large systems into smaller parts | Makes large molecules tractable |
| Variational Quantum Eigensolver (VQE) | Hybrid quantum-classical algorithm | Noise-resilient for current hardware |
| One-sided Jacobi SVD | Optimized matrix decomposition | 60x faster than non-optimized versions |
| Tensor Contractions | Efficient multi-dimensional operations | Leverages many-core processor architecture |
The successful emulation of 1000-qubit quantum chemistry calculations represents more than just a technical achievement—it opens new possibilities for scientific discovery. By providing a platform to study molecular systems at unprecedented scale and accuracy, this approach accelerates research in drug design, materials science, and catalyst development.
These emulation capabilities provide a crucial testing ground for quantum algorithms1 . As researchers develop new approaches to quantum simulation, they can use classical emulators to validate their methods before deploying them on actual quantum hardware.
The field is moving toward increasingly hybrid approaches that combine classical high-performance computing, quantum emulation, and actual quantum processing units7 . Companies like Quantinuum are already developing full-stack quantum solutions that integrate hardware, software, and applications7 .
"Achieving quantum advantage in chemistry will require more than just quantum hardware; it will require a synergistic approach that combines quantum computing workflows with classical supercomputing and AI"7 .
| Platform/Method | Maximum Qubits (Chemistry) | Key Strength | Limitation |
|---|---|---|---|
| MPS-VQE (Sunway) | 1000 | Largest scale achieved | Optimized for lower entanglement |
| Trapped Ions | 51 | High-fidelity control | Challenging to scale |
| Superconducting Qubits | Varies | Fast gate operations | Noise and decoherence |
| Tensor Network Simulators | Varies | Efficient for 1D systems | Struggles with high entanglement |
The path toward practical quantum computing for chemistry resembles a relay race rather than a sprint. Classical emulation techniques like the 1000-qubit MPS-VQE simulator represent a critical first leg—pushing the boundaries of what's possible today while developing the algorithms and methodologies that will run on tomorrow's fault-tolerant quantum computers.
As hardware continues to improve, with companies like Quantinuum demonstrating the first scalable error-corrected computational chemistry workflows7 , the line between classical emulation and quantum computation will increasingly blur. The ultimate goal remains unchanged: to solve chemical problems that are currently impossible to address with any classical approach alone.
What makes this journey particularly exciting is its collaborative nature. Rather than positioning classical and quantum approaches as competitors, researchers are increasingly viewing them as complementary technologies that, when combined, can accelerate scientific discovery. As we stand at this intersection of classical and quantum computing, one thing becomes clear: the future of computational chemistry will be neither purely classical nor purely quantum, but a sophisticated integration of both.
"The three-dimensional world of ordinary experience—the universe filled with galaxies, stars, planets, houses, boulders, and people—is a hologram, an image of reality coded on a distant two-dimensional surface," observed physicist Leonard Susskind7 . In much the same way, the quantum simulations of tomorrow may emerge from the intricate interplay between classical and quantum processors working in concert—each performing the tasks to which it's best suited in a harmonious symphony of computational power.