What's Happening?
Researchers from IBM Quantum, in collaboration with Mitsubishi Chemical Corporation and Keio University, have successfully demonstrated a generative quantum eigensolver (GQE) trained with quantum hardware at a 70-qubit scale. This achievement involved
training policies to generate circuits for sample-based quantum diagonalization, initially using simulations and then integrating the IBM Heron processor ibm_kobe into the training loop. The method produced circuits for organotin hydroxides, which are models for materials used in extreme-ultraviolet lithography. These circuits achieved accuracy within 15 milli-Hartree (mHa) of a density-matrix renormalization group reference in 27 out of 28 runs. Notably, at 56-70 qubits, these circuits required significantly fewer two-qubit gates (34 to 72 times less) compared to alternative methods, highlighting a substantial reduction in computational cost.
Why It's Important?
This breakthrough signifies a critical step forward in quantum computing, particularly for complex chemical simulations. The ability to train quantum algorithms directly on hardware at a 70-qubit scale, rather than relying solely on classical simulations, pushes the boundaries of what is currently achievable. This is important because classical simulations become computationally intractable for larger quantum systems, limiting the scope of scientific discovery. By reducing the number of two-qubit gates, the researchers have also addressed a major challenge in quantum computing: gate errors. Fewer gates mean less accumulation of errors, leading to more accurate and reliable results on noisy intermediate-scale quantum (NISQ) devices. This advancement could accelerate the development of new materials, drug discovery, and a deeper understanding of molecular properties, with significant implications for industries ranging from pharmaceuticals to advanced manufacturing.
What's Next?
The successful training of a 70-qubit generative quantum eigensolver on hardware opens new avenues for using quantum hardware's unique feedback capabilities beyond classical simulation limitations. The researchers plan to continue refining these training policies and explore their application to an even broader range of chemical systems. The focus will likely be on improving the robustness and accuracy of these quantum algorithms, as well as scaling them to even larger qubit counts. This work also sets a precedent for integrating real quantum devices into algorithm development, moving away from purely simulated environments. Future research will aim to further optimize circuit generation and explore the potential for transfer learning across different chemical systems, ultimately paving the way for more practical and widespread applications of quantum computing in scientific research and industrial innovation.
Beyond the Headlines
The development of hardware-trained quantum algorithms at this scale has profound implications for the future of scientific research and technological innovation. It represents a shift towards a more empirical approach in quantum algorithm design, where the nuances of real quantum hardware directly inform and refine the algorithms. This could lead to the discovery of entirely new computational paradigms that are uniquely suited to quantum mechanics, potentially unlocking solutions to problems currently deemed intractable. Furthermore, the ability to accurately model complex molecular systems could revolutionize fields like materials science, enabling the design of materials with unprecedented properties. Ethically, as quantum computing capabilities advance, discussions around data security and the potential for quantum supremacy will become increasingly relevant, necessitating careful consideration of responsible development and deployment.
