Meeting 2026 10 19

Meeting - October 19, 2026

• Date: Monday, October 19, 2026
• Time: (Slot #2) 7 pm - 8 pm US Eastern Daylight Time (View Meeting Time in Your Timezone)
• Location: Virtual Meeting.
• Meeting Link: openqse.org/captcha

Upcoming meetings

• Next: Monday, October 26, 2026 @ (Slot #1) 12 pm - 1 pm US Eastern Daylight Time (View Meeting Time in Your Timezone)

Agenda

• Kalyan Gottiparthi (ORNL)

Attendance

Person	Institution
Amir Shehata	ORNL
Kalyan Gottiparthi	ORNL
Ryan Landfield	ORNL
Nick Wright	NVIDIA
Neal Erickson	Quantinuum
Alain Roy	IonQ
Michael Ferguson	HPE
Claudio Siqueira de Carvalho	IBM
Jonathan Skone	NERSC
Joaquin Chung	ANL
Miwako Tsuji	RIKEN
Masoud Mohseni	HPE
Namit Anand	HPE
Doug Oucharek	ORNL
Eun Kyung Lee	IBM
Jeff Heckey	AWS
Kevin Kissell	Alice & Bob
Martin Schulz	TUM

Notes

• NOTICE: Notes generated using AI; please verify accuracy.

1. Opening

   • Amir Shehata introduced Kalyan Gottiparthi from Oak Ridge National Laboratory and invited him to present the group's work on reactive flows using quantum computing.
   • Gottiparthi framed the talk around reactive-flow simulations, quantum linear solvers, and possible HPC–QC integration workflows.

2. Presentation Overview (Kalyan Gottiparthi)

   2.1. Reactive-flow simulation challenges

   • Reactive-flow CFD combines Navier–Stokes equations with species equations and reaction source terms. Nonlinear convection and chemical reaction rates add computational complexity.
   • Large simulations may track many species and reaction steps. The presentation used an example requiring more than five billion grid points and noted that chemical kinetics can introduce time scales down to nanoseconds.
   • Stiff chemical kinetics require resolving short time scales; solvers may perform multiple chemistry sub-iterations for each fluid-flow time step.
   • Direct numerical simulation and large-eddy simulation address different ranges of resolved scales, with the latter requiring closure models for filtered terms, including reaction rates.

   2.2. Reducing the problem to quantum linear solves

   • The work studies simplified species and temperature balance equations to understand methods before applying them to the full three-dimensional system.
   • The reaction rates and fluid equations contain nonlinear terms. The group uses Kalman linearization and Taylor expansion to approximate nonlinearities; Gottiparthi reported reasonably good results in tests.
   • After linearization, the system can be expressed as a linear solve suitable for quantum linear-solver algorithms.

   2.3. Solver approaches and workflow structure

   • The presentation discussed HHL, variants of variational quantum linear solvers (VQLS), and SQLS, which uses singular-value decomposition as preprocessing. A truncated SVD variant was described as a way to reduce memory use.
   • A standalone workflow may include circuit synthesis, a QPU run, and HPC post-processing; SQLS adds SVD preprocessing.
   • Reactive-flow calculations repeat these steps as the simulated system advances, making the HPC–QC workflow iterative.
   • The talk described the current application results as reliable under the methods used, while explicitly stating that it did not demonstrate quantum advantage.

   2.4. HPC–QC integration needs

   • State preparation and circuit synthesis/transpilation can dominate runtime as problem size grows, while the quantum circuit simulation itself may be comparatively fast.
   • GPU acceleration was discussed as a way to make state preparation and synthesis faster and potentially accommodate larger problems. Gottiparthi said faster CPU-based methods could also meet the goal.
   • The presented results used state-vector simulation. No quantum hardware backend was used; the HHL circuit was formed with Qiskit.

3. Discussion & Q&A

   3.1. Scope of the results

   • Gottiparthi clarified that the presentation was not claiming quantum advantage, but sharing application work and discussing patterns for HPC–QC integration.
   • He noted that reported calculations were state-vector simulations, not shot-based simulations or hardware runs.

   3.2. Acceleration of synthesis and transpilation

   • Michael Ferguson asked whether GPU acceleration was needed specifically to reduce runtime or to avoid moving data away from GPUs already used for preconditioning.
   • Gottiparthi said the primary goals were faster state preparation and synthesis, and fitting larger problems; a substantially faster CPU method would also be useful.
   • Ferguson observed that synthesis and transpilation could otherwise take far longer than the quantum portion of the workflow. Gottiparthi agreed that state preparation becomes a bottleneck as problems scale.

   3.3. Meaning of transpilation and hardware execution

   • Amir Shehata asked whether the cited transpilation reached hardware-level pulse descriptions.
   • Gottiparthi clarified that the work had not run on a backend; Qiskit was used to form the HHL circuit for the supplied matrix system. Additional steps would be needed to run on real hardware.

4. Key Takeaways

   • Reactive-flow simulations combine nonlinear fluid dynamics, many species, and stiff chemical kinetics, creating substantial computational demands.
   • Linearization provides a route to applying quantum linear-solver methods to reduced species and temperature equations.
   • In the discussed workflows, state preparation and circuit synthesis/transpilation are important integration costs and may outweigh circuit execution time.
   • The reported work used simulation and did not establish quantum advantage or demonstrate execution on quantum hardware.

5. Action Items

   • No explicit action items captured.

6. Closing

   • Shehata thanked Gottiparthi for the presentation and discussion. Gottiparthi thanked attendees and expressed hope for further discussion and progress on the problem.

• Download Transcript

Files