What is the QVM?
The Quantum Virtual Machine (QVM) is a novel quantum computing architecture designed to bridge the gap between quantum and classical computing. Unlike traditional quantum simulators that simply emulate qubit behavior, the QVM is a full-stack computation engine built for production workloads.
Design Philosophy
We built the QVM with three core principles:
- Accessibility — Make quantum computing available to developers who don't have a PhD in quantum physics
- Performance — Deliver real computational advantages, not just academic demonstrations
- Portability — Run the same code on simulators, Google Willow, IBM Quantum, or TPU-accelerated backends
Architecture Layers
1. State Simulation Layer
The foundation of the QVM is an optimized state vector simulator that handles qubit superposition, entanglement, and decoherence. We use sparse matrix techniques and TPU acceleration to simulate circuits that would be impractical on standard hardware.
2. Gate Compilation Engine
High-level quantum circuits are compiled into optimized gate sequences tailored to each target backend. For Google Willow, we compile to its native gate set (H, CZ, Rx, Ry). For IBM processors, we use their native CX-based gate decomposition.
3. Hybrid Execution Runtime
The QVM seamlessly splits workloads between quantum and classical processors. Variational algorithms, for example, run their quantum circuits on real hardware while the classical optimization loop runs on TPU clusters.
4. Error Correction Module
Built-in surface code error correction ensures fault-tolerant computation on real quantum hardware. The module automatically selects the appropriate correction strategy based on the target processor's error characteristics.
Performance
The QVM achieves 762,601 quantum operations per second when running on our TPU-accelerated backend. For comparison, this is orders of magnitude faster than most quantum simulators available today.
Open Source
The QVM is fully open source under our dual MIT + Brion Proprietary Addendum license. We believe that making this technology accessible accelerates the entire field.
Get started at github.com/Brionengine.