May 5, 2025 \u2014 Qoro and Galicia Supercomputing Center (CESGA) recently collaborated to explore the potential of scalable, distributed quantum circuit simulations using
high-performance computing.<\/p>\n
Dr. Andr\u00e9s G\u00f3mez, Applications & Projects Dept. Manager, lead his team focussed on application support to CESGA\u2019s supercomputing users and the promotion and management of R&D&I projects.<\/p>\n
This two-week pilot project focused on deploying Qoro Quantum\u2019s parallelized quantum algorithm software package, scheduler, and orchestration platform to
execute a parallelized version of the Variational Quantum Eigensolver (VQE) and the Quantum Approximate Optimization Algorithm (QAOA) across 10 computing nodes in CESGA\u2019s QMIO infrastructure using the distributed QPU emulator platform CUNQA.<\/p>\n
One of the key takeaways from this pilot was how seamlessly CESGA and Qoro Quantum\u2019s platforms integrated using common interfaces, allowing for smooth
communication between Qoro\u2019s application and scheduling system and CESGA\u2019s QPU emulator CUNQA. By leveraging Qoro Quantum\u2019s application software, Divi, for
automated algorithm parallelization, and cloud infrastructure for scheduling and orchestration, we were able to automatically generate large-scale quantum workloads and distribute them efficiently across CESGA\u2019s HPC resources. This demonstrated an important step in structuring quantum workloads for scalable execution in distributed quantum computing environments in the near term.<\/p>\n
Dr. Stephen DiAdamo, Co-Founder & CTO, Qoro, commented \u201cIt was a very smooth collaboration, our systems integrated very well together and the end-to-end
functionality worked exactly as expected. In one day of setup, we were able to run meaningful simulations on a complex distributed system. It opens up many new
opportunities for exploring further developments for developing scalable and effective middleware for quantum computing.\u201d<\/p>\n
HPC systems tackle some of the world\u2019s most computationally intensive problems. As quantum computing matures, the expectation is that HPC and quantum systems will work together in hybrid architectures, where classical and quantum resources are orchestrated to solve problems more efficiently than either could alone. Currently, quantum computing remains in its early stages, with most applications running on either simulated environments or small-scale physical quantum processors. This pilot project represents a crucial first step toward integrating quantum computing into large-scale HPC workflows by demonstrating how quantum circuits can be efficiently scheduled and executed across a distributed computing environment.<\/p>\n
CESGA\u2019s CUNQA framework plays a crucial role in this process, acting as an interface that allows the emulation of a distributed quantum infrastructure composed of
several quantum nodes. This provides researchers and engineers with a testbed for developing distributed hybrid classical-quantum algorithms at scale, ensuring that
as real QPUs become more powerful, they can be seamlessly integrated into existing HPC workflows.<\/p>\n
By successfully integrating Qoro Quantum\u2019s orchestration platform with CESGA\u2019s HPC resources, this project has demonstrated how quantum workloads can be structured
to eventually transition from simulation to execution on real hybrid quantum-HPC systems. As quantum computing hardware continues to advance, this type of integration will be key to unlocking its full potential.<\/p>\n
The Variational Quantum Eigensolver (VQE) is a hybrid quantum-classical algorithm used to estimate the ground-state energy of quantum systems\u2014a fundamental
problem in quantum chemistry and materials science. VQE is well-suited to near-term quantum devices, using a parameterized quantum circuit (ansatz) to
prepare quantum states and a classical optimizer to iteratively minimize the expected energy of the system.<\/p>\n
In this use case, we simulated the hydrogen molecule using two different ans\u00e4tze, UCCSD and Hartree-Fock, over 20 bond lengths, in parallel across 10 nodes in
CESGA\u2019s HPC cluster. Divi was used to automate the parallelization of the problem, generating batched VQE circuits based on a range of bond lengths and ansatz
parameters. Monte Carlo Optimization was applied to explore the parameter space efficiently, with Divi producing 6,000 circuits for evaluation. These were distributed
automatically across the nodes and scheduled using Qoro\u2019s orchestration software.<\/p>\n
The circuits were executed using CESGA\u2019s CUNQA simulation platform, which emulates quantum processing across the cluster. Upon completion, results were
returned to Divi for aggregation and analysis. The full workload was simulated in just 0.51 seconds, demonstrating how distributed execution can accelerate VQE
experiments at scale. Using only 15 lines of code from Divi, we enabled high-throughput comparison of quantum ans\u00e4tze across multiple bond lengths\u2014highlighting the potential of this approach for rapid exploration in quantum chemistry research.<\/p>\n
The Quantum Approximate Optimization Algorithm (QAOA) is a powerful hybrid quantum-classical approach for tackling combinatorial optimization problems, such
as Max-Cut. In Max-Cut, the objective is to divide the nodes of a graph into two groups while maximizing the number of edges between them \u2014 a problem with
real-world relevance in areas like logistics, circuit design, and clustering. QAOA approximates solutions by applying alternating layers of parameterized quantum
gates and classical optimization, making it suitable for current quantum hardware and efficient when run at scale.<\/p>\n
In our collaboration, we tested Max-Cut with QAOA on a 150-node graph partitioned into 15 clusters using Divi. Divi took in the problem structure and generated batches
of parameterized circuits using Monte Carlo Optimization. These batches were distributed across 10 nodes in CESGA\u2019s computing infrastructure, where CUNQA simulated the quantum circuits in parallel. Divi then collected the results and performed the final aggregation and analysis, enabling seamless end-to-end orchestration of a distributed QAOA workflow. Again, with less than 20 lines of code, we could generate these complex optimization problems.<\/p>\n