Izhar Medalsy

AI Summary

→ WHAT IT COVERS Izhar Medalsy, CEO of Quantum Elements, explains how his company builds large-scale digital twins of quantum hardware — simulating up to 100 noisy qubits on classical supercomputers — and uses AI to identify noise sources, optimize error suppression, and push algorithm accuracy from 80% to 99% on IBM's platform using Shor's algorithm. → KEY INSIGHTS - **Digital Twin Validation:** Quantum Elements demonstrated digital twin accuracy by running Shor's algorithm on IBM hardware, achieving 80% baseline accuracy, then using crosstalk analysis from thousands of simulated qubit configurations to identify an error suppression remedy. Implementing that remedy on the physical IBM system raised accuracy to 99% — surpassing the previous literature high of 70% and providing a concrete closed-loop validation method. - **Noise Modeling at Scale:** Effective quantum simulation requires modeling five specific noise channels: T1/T2/T-star decoherence rates, 1/f noise, crosstalk between qubits, and leakage into higher energy states. Crosstalk is particularly problematic at scale — calibrating one qubit destabilizes neighbors, creating a whack-a-mole effect. Beyond roughly 100 qubits, manual calibration becomes unmanageable, making AI-driven optimization the only practical path forward. - **Density Matrix Advantage:** Classical quantum simulators require 10,000+ repeated "shots" to build a statistically valid result. Quantum Elements' digital twin produces the full density matrix — the complete probability distribution of all outcomes — in a single run. This capability lets developers pause simulation mid-circuit to inspect system state, something physically impossible on real quantum hardware, enabling precise noise attribution and labeled dataset generation. - **AI Training Data Generation:** Quantum hardware is too scarce, expensive, and unstable to serve as an AI training source — calibration parameters drift daily and hardware generations turn over rapidly. Digital twins solve this by generating massive labeled datasets at low cost, including representations of future hardware states. This enables training AI decoders for quantum error correction that remain robust across hardware generations, not just current machines. - **Logical Qubit Milestone:** Using digital twin optimization, Quantum Elements achieved logical qubit behavior on IBM hardware — previously demonstrated only by Google on superconducting systems. Separately, the platform pushed Rigetti hardware to 99.9% single-qubit gate fidelity. Three metrics to track quantum progress: two-qubit gate fidelity approaching 99.99%, physical qubit counts toward IBM's 10,000-qubit target, and reducing the physical-to-logical qubit ratio below 10-to-1. - **Hardware-Aware Stack Strategy:** Unlike classical computing where a zero is universally a zero, every layer of quantum software — pulse calibration, circuit execution, error mitigation, error correction, and application algorithms — must be tuned to specific hardware. Quantum Elements serves customers across this entire stack by providing a configurable canvas where users input their own hardware parameters without disclosing proprietary designs, preserving IP while still benefiting from simulation-scale optimization. → NOTABLE MOMENT Medalsy revealed that Quantum Elements simulated over 100 noisy qubits — surpassing quantum error correction distance-seven, a threshold widely cited as the boundary beyond which classical devices cannot meaningfully simulate quantum hardware. The team continues pushing that boundary further, consistently exceeding their own expectations for how far classical simulation can reach. 💼 SPONSORS None detected 🏷️ Quantum Computing, Digital Twins, Quantum Error Correction, AI Training Data, Qubit Noise Modeling, Quantum Hardware Optimization