Quantum-inspired simulation platform
Today’s heavy-duty sims put you in a bind: run toy problems on your machine or sit in line for pricey clusters. Too much time goes into plumbing instead of trying out ideas.
We built QEngine to change that—one stack that runs on your laptop, in the cloud, or on future quantum systems. Huge problem sizes without blowing up RAM, a visual builder, and AI that checks your setup so you move from concept to real runs with fewer crashes.
Solver checked on reference shapes and multiple resolutions. Below: velocity magnitude |u| from a postprocessed run and an inclined-wing case.
Run a simulation and generate a GIF to see it here.
Run a simulation and generate a GIF to see it here.
We’re starting in aerospace. The solver has been validated on canonical geometries and a range of mesh sizes—the numbers speak. In regimes where conventional codes don’t scale, we do. We’re on par with HPC for typical jobs and pull ahead when legacy solvers hit the wall. A quantum-ready core means we’re set to benefit as new hardware arrives.
Circle, square, wing (NACA-like), pixel masks, and custom obstacles. Edit from the mesh view or config.
Central, upwind, MUSCL, adaptive upwind. First-order, RK2, or adaptive time-stepping based on flow diagnostics.
Lift/drag (Cl/Cd), Strouhal number, bond-dimension and timing summaries. Offline post-processing from MPS checkpoints.
Save only MPS checkpoints during long runs; compute plots, forces, and GIFs offline. Resume from any checkpoint with correct history.
AI-assisted simulator, live mesh view, and run controls in the portal. Backend API for automation and integration.
JSON configs, preset examples, and export from the UI. Cotengra and bond-dimension tuning for performance.
Full 2D incompressible Navier–Stokes: many geometries, adaptive stepping, visual UI, and AI that helps catch config mistakes—so you get from idea to working runs with fewer failures.
Velocity and pressure as matrix product states; operators as MPOs. High resolution stays tractable on a single node.
Pressure Poisson solved with DMRG at each step. Adaptive bond dimension and cotengra-optimized contractions for speed.
Same MPS/MPO formulation that runs on classical HPCs maps to future quantum and hybrid backends. One codebase, multiple futures.
The problem is hard; the team is built for it. Sanjay (CEO) studied at CMU and holds Quantum AI patents from IBM and PsiQuantum. Tushar (CTO) has a math PhD from Texas A&M, a B.Tech from IIT Kanpur, and experience in quantum algorithms at Oak Ridge plus deep work in generative AI. Together we cover quantum, large-scale numerics, and AI.
CEO · CMU · Quantum AI patents (IBM, PsiQuantum)
CTO · PhD Math (Texas A&M) · IIT Kanpur · Oak Ridge, GenAI
Visual builder plus AI that double-checks your setup. Ask in plain language: stable or turbulent? Transient or converged? Get guidance and live analysis so you go from idea to real runs with fewer dead ends.
Ask what the output means: stable or turbulent? Transient or converged? The assistant explains flow regimes and diagnostics in context.
Describe your goal; get suggestions on geometry, Re, schemes, and settings to get the behavior you need.
Chat about the run while it's going. Analyze plots, forces, and diagnostics as they update—no need to wait for post-processing.
Reproducible configs, Cl/Cd, Strouhal, vorticity. From laptop to HPC to quantum-ready formulation.
One interface, one formulation. Use it on your machine, in the cloud, or on quantum hardware when it’s ready—same workflow everywhere.
Single machine. Full control, no queue.
Clusters and supercomputers. Scale resolution and steps.
Quantum-ready formulation. The future of simulation.
We’re building the simulation platform we wanted but never had: QEngine brings together AI, scalable cloud, and a quantum-ready path so high-fidelity simulation is within reach for more teams.
Open Simulator →