JoS QUANTUM

Everyone asks how many qubits it takes to break RSA or ECC. It's the wrong question. The number that matters isn't qubit count — it's how efficiently Shor's algorithm compiles into fault-tolerant gates, and how much overhead error correction adds on top. And that overhead has been falling fast: → RSA-2048: from 20 million noisy qubits (2019) to under 1 million (2025) — same hardware assumptions. → ECC-256: from ~9 million physical qubits (2023) to under 500,000, in about nine minutes (2026). Hardware is improving steadily. But the part most people underestimate is that algorithmic and error-correction requirements can drop by orders of magnitude — which is exactly why the timeline keeps compressing even when the chips stand still. No cryptographically relevant quantum computer exists yet. What's changed is the shape of the target. We wrote up where the real bottleneck sits — the T-gate and magic-state distillation — and what the collapsing estimates mean for the post-quantum migration. Link in comments. #QuantumComputing #PostQuantum #Cryptography #QEC #QuantumSecurity

How much does quantum actually buy you in risk analysis? We built an app to show — not tell. Five years ago we published A Quantum Algorithm for the Sensitivity Analysis of Business Risks (arXiv:2103.05475). The theory said quantum amplitude estimation converges as O(1/M) where classical Monte Carlo crawls at O(1/√M). The natural question from every risk manager since: when quantum will overtake classical MC for practical usecases? So we built risQ — an application where you can model business risk, evaluate VaR/CVaR. You draw a business-risk network as a probabilistic graph (noisy-OR cascades — cyber attack → system outage → regulatory fine), and it evaluates risk classically and quantum: 🎲 Classical Monte Carlo — VaR, CVaR, tornado + Sobol sensitivity, scaling to hundreds of nodes. Large scale sensitivity analysis is computationally infeasible. ⚛️ Quantum Amplitude Estimation — Estimation of VaR and CVaR via QAE including quantum hardware resource estimations (# oracle calls, # qubits, # CNOT). 🌪️ Sensitivity analysis — QAE inside Grover provides quartic speedups. A fourth view fits the convergence of each and locates the precision crossover where quantum overtakes classical. It also stays honest about the hardware: hit the 20-qubit simulation ceiling and the app shows you why — a 24-node enterprise model needs a gate controlled on all 24 qubits, decomposing into thousands of elementary operations. That's the concrete, quantified reason it's classical-only today. Quantum risk analysis isn't a slide anymore. It's something you can open in a browser and test it. More to come soon including estimations with #quantumerrorcorrection schemes and connection to real hardware. Quantinuum IQM Quantum Computers IBM Quantum IonQ #QuantumComputing #RiskManagement #QuantumFinance #QuantumMonteCarlo

Exciting news! 🚀 We are thrilled to share JoS QUANTUM’s latest research: "Pauli Cloners for Pauli Channels." While the "No-Cloning Theorem" tells us we can't perfectly copy an unknown quantum state, our team has developed a new class of quantum circuits—Pauli Cloners—that push the boundaries of what’s possible in noisy environments. 💡 Key Contributions: 👉 Scalable Quantum Circuits: We’ve designed circuits that implement cloning machines for registers of N qubits. 👉 Noise-Adaptive Performance: Unlike standard approaches, these cloners can be tailored to specific Pauli noise profiles. 👉 Outperforming the Standard: Our results show that Pauli Cloners generally outperform well-known benchmarks, including the Universal Quantum - Cloning Machine (UQCM). Theoretical Deep Dive: We explore the fascinating interplay between Pauli errors, Mutually Unbiased Bases (MUBs), and cloning fidelity. Understanding how to optimally "copy" information in the presence of noise is an important step toward more robust and secure quantum communication. Congrats to the team at JoS QUANTUM Marcelin Gallezot Thomas Decker Sven Kerstan Markus Braun Niklas Hegemann! 👇 You find the link to the paper in the comments below. #QuantumComputing #QuantumInformation #DeepTech #JoSQUANTUM #QuantumResearch #Innovation

From Theory to Therapy: We’re bridging the gap between Deep Learning and rigorous mathematics for skin cancer detection! Excited to share our work on a hybrid AI model that fuses a U-Net with the classical Isoperimetric Inequality. While modern AI is powerful, its geometric shortcomings can be critical in diagnosis. Our mathematical refinement step (using an HSV color space adaptation) doesn't just clean up edges—it actively recovers faint biological structures missed by the neural network. The result? ~98% Precision and a +1.46% increase in Recall over the raw U-Net. This is a testament to the power of Explainable AI (XAI) and geometric priors in making medical technology more accurate and trustworthy. It's not about bigger models; it's about better definitions. Read the full article for the deep dive! #MedicalImaging #ComputerVision #DeepLearning #Mathematics #HybridAI #Dermatology #xAI

When will #quantumcomputers be able to break encryption like RSA? You need to watch 2 parameter: 1️⃣ Hardware improvements: Number of qubits AND Error rates (success rate of a gate operation, current best hardware from IonQ, Quantinuum IQM Quantum Computers, Google range around 99.9-99.99%) 2️⃣ Algorithmic requirements (more efficient algorithms need less qubits etc.) Key takeaways: 👉 Latest estimates from Craig Gidney do not rely on new, exotic architectures. Assumes a standard grid of nearest-neighbor-connected qubits. 👉 1M qubit target is still ~9,500x larger (13.2 doublings) than today's best physical devices. 👉 Expect lower boundaries around 1024 logical qubits to factor RSA-2048 with surface code distance of at least 10 (or 242 physical qubits for each logical qubit) --> ~242,000 qubits Very nice webpage and objective information from Samuel Jaques -- many thanks!

Why we build our own quantum workflow software: 1️⃣ Geqo enables our clients to build quantum applications much faster. 2️⃣ Geqo provides the basis for full-scale quantum-classical workflows. 3️⃣ Geqo can be used for HPC orchestration. Contact Markus Braun or Niklas Hegemann for a first demo

Introduction of the quantum-centric framework Geqo to build and test quantum programs. Here our colleague Lucas Euler presents building a risk model as quantum program being evaluated using the Monte Carlo equivalent on quantum computers: the (low depth) Quantum Amplitude Estimation. #quantumtechnology #quantumcomputing #quantumready