A hybrid Density Functional Theory–Quantum Embedding (QDFT) framework integrates DFT with a quantum electronic-structure solver, enabling treatment of chemically relevant parts of a system using the Variational Quantum Eigensolver (VQE) while the rest is handled by DFT, improving accuracy and retaining the scalability of High-Performance Computing (HPC). Namrata Manglani at C-DAC, Pune, and colleagues have created a new computational method that merges conventional supercomputing with the potential of quantum computers.

This hybrid approach, termed DFT-Quantum Embedding, tackles limitations in current molecular simulations, particularly when dealing with complex electron interactions. By dividing calculations strategically, the team improved the accuracy of modelling electronic structures while maintaining the processing power of existing high-performance computing systems.

Namrata Manglani and colleagues at C-DAC, Pune, have pioneered a computational framework that blends the strengths of traditional supercomputing and emerging quantum computing technologies. Density Functional Theory, a key tool of computational chemistry, can be likened to creating a rough sketch of a molecule’s energy; it’s efficient but sacrifices detail. DFT approximates the many-body Schrödinger equation, offering a computationally tractable method for calculating the electronic structure of materials, but these approximations can lead to inaccuracies, especially in systems where electron correlation is strong.

To enhance accuracy, the team focuses computational effort on the most chemically important parts of a molecule, the ‘active space’, similar to highlighting the key players in a complex narrative. This active space, typically encompassing only a few atoms and valence electrons, often between 5 and 15, is then solved using the Variational Quantum Eigensolver, a quantum algorithm designed to find the lowest energy state, while the remainder of the molecule is handled by conventional methods. This partitioning allows for a more accurate description of the critical electronic interactions without incurring the immense computational cost of applying quantum methods to the entire system.

Hybrid quantum-classical computation addresses limitations in modelling strongly correlated systems

Scientists at C-DAC, Pune and Anchor Kutchhi Engineering College have developed a hybrid Density Functional Theory-Quantum Embedding (QDFT) framework that improves selected electronic-structure properties while maintaining the scalability of classical High-Performance Computing. Previous methods struggled to accurately model strongly correlated systems, where electrons strongly influence each other. These systems, prevalent in materials science and catalysis, exhibit behaviour that is poorly described by standard DFT approximations.

The new approach partitions molecular systems, enabling treatment of chemically relevant ‘active spaces’ with the Variational Quantum Eigensolver, a quantum algorithm, and utilising conventional Density Functional Theory for the remainder of the molecule. The Variational Quantum Eigensolver operates by preparing a trial quantum state, parametrised by a set of variational parameters, and then optimising these parameters to minimise the energy expectation value. This process relies on repeated measurements on the quantum computer to estimate the energy, making it particularly suited for NISQ devices.

CPU-based quantum simulation was identified as a computational bottleneck through detailed profiling, prompting the development of a runtime-estimation methodology for assessing execution on actual quantum hardware. This establishes a foundation for scalable HPC-quantum hybrid simulations, paving the way for practical applications as quantum technology advances. The framework incorporates six key steps: active-space selection, embedded Hamiltonian construction, symmetry preservation, operator mapping, self-consistent density updating, and classical-quantum coupling, all designed to improve the accuracy of electronic-structure properties.

Active-space selection is crucial, determining which parts of the molecule receive the more accurate, but computationally expensive, quantum treatment. The embedded Hamiltonian construction accurately represents the interactions within the active space, while accounting for the influence of the surrounding environment. Symmetry preservation ensures that the quantum calculations respect the symmetries of the molecule, reducing computational cost and improving accuracy. Operator mapping translates the electronic structure problem into a form suitable for execution on the quantum computer.

Self-consistent density updating ensures that the DFT and quantum calculations are mutually consistent. Classical-quantum coupling facilitates the exchange of information between the classical and quantum processors. Despite the need for advancements in quantum hardware and mitigation of noise inherent in real quantum processors, the runtime-estimation methodology demonstrates a commitment to future implementation, anticipating the challenges of qubit limitations and instability. Profiling also revealed that CPU-based quantum simulation currently limits overall performance, highlighting an area for optimisation as quantum technology matures.

Hybrid algorithms enhance electronic property prediction for strongly correlated materials

Current research drives efforts to model molecular interactions with greater precision and efficiency. Density Functional Theory excels at large-scale simulations, but its accuracy falters when dealing with complex electron behaviour in ‘strongly correlated’ systems. These systems, found in materials exhibiting phenomena like high-temperature superconductivity and magnetism, require more sophisticated theoretical approaches.

Researchers at C-DAC, Pune and Anchor Kutchhi Engineering College acknowledge that simply scaling up quantum computing power isn’t a viable short-term solution, given the current limitations of qubit numbers and quantum state stability. Current NISQ devices typically have fewer than 100 qubits, and these qubits are prone to errors due to decoherence and other noise sources.

By focusing quantum resources on only the most critical parts of a molecule, this hybrid approach offers a valuable stepping stone towards more accurate molecular modelling, reducing the demands on unstable quantum processors. The team overcame limitations of both traditional DFT and current quantum hardware, achieving improved accuracy without sacrificing computational scalability. The ability to accurately predict electronic properties is crucial for designing new materials with desired functionalities. Future development will unlock truly scalable molecular simulations, building on the successful integration of classical and quantum computing techniques to improve the accuracy of electronic property predictions for complex molecules. This methodology is vital for assessing the feasibility of running increasingly complex simulations on emerging quantum hardware and for optimising the partitioning between classical and quantum resources to minimise execution time and cost. The QDFT framework represents a step towards harnessing the power of quantum computing for practical materials science applications.

The research demonstrated a hybrid classical-quantum framework, integrating Density Functional Theory with a Variational Quantum Eigensolver to model complex molecular systems. This approach allows researchers to focus quantum computing resources on chemically relevant portions of a molecule, improving the accuracy of electronic property predictions beyond what traditional methods achieve. By partitioning large systems, the framework addresses limitations of both current quantum hardware and conventional simulations. The authors developed a methodology to estimate runtime on quantum hardware, which is important for assessing the feasibility of future simulations.

👉 More information
🗞 Hybrid HPC-Quantum Simulations: DFT-Quantum Embedding for Molecular Systems
✍️ Namrata Manglani, Samrit Maity, Shashank Sharma, Tejjan Arora, Soham Phulare, Shreyas Kadam and Sanjay Wandhekar
🧠 ArXiv: https://arxiv.org/abs/2608.12884

More like this