{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreifqtu4xyolw3nsru54dq56a54rfz76fesi5szm4bs2y3ibaiolz54",
    "uri": "at://did:plc:qzjwstutqk2cy7df7jbzd2hx/app.bsky.feed.post/3mh6uih2brox2"
  },
  "coverImage": {
    "$type": "blob",
    "ref": {
      "$link": "bafkreide7su5wu23j3gptmme7z7nkcxe62sj6gwbts2ieo6xwoxngwlx5e"
    },
    "mimeType": "image/jpeg",
    "size": 4065024
  },
  "path": "/article/4145726/quantum-elements-cuts-quantum-error-rates-using-ai-powered-digital-twin.html",
  "publishedAt": "2026-03-16T15:02:32.000Z",
  "site": "https://www.networkworld.com",
  "tags": [
    "Artificial Intelligence, Data Center, High-Performance Computing",
    "it developed and validated the approach",
    "published in Nature Communications",
    "Izhar Medalsy",
    "Arian Vezvaee",
    "Bob Sutor",
    "simulate the physics of its quantum processor",
    "simulate 50 qubits",
    "released this past October",
    "research report"
  ],
  "textContent": "Quantum Elements, a Los Angeles startup, has demonstrated a new technique for suppressing errors in logical qubits that shows the highest fidelity of entangled, logical qubits on a superconducting quantum computer ever achieved.\n\nThe company says it developed and validated the approach using its own AI-powered quantum digital twin platform, which simulates not just how a quantum circuit is supposed to work but also how it actually behaves on real hardware, noise and all. The peer-reviewed paper was published in Nature Communications in late February.\n\nThe new technique is called logical dynamical decoupling and involves shifting qubits between different states, says Izhar Medalsy, cofounder and CEO at Quantum Elements.\n\n“We combined this with error detection, and that’s how we showed that you can create qubits with 95% fidelity, which is way beyond what you can do with code alone, which was 43%,” says Arian Vezvaee, the company’s quantum research scientist.\n\nThe logical dynamical decoupling is not a standalone method, Vezvaee adds. “It really is a complementary thing,” he says. “When combined, the combination performs better than each individual part.”\n\nBob Sutor, founder and CEO at Sutor Group Intelligence and Advisory, says that a qubit can be thought of as akin to a spinning basketball. Over time, the spinning can slowly veer in a particular direction. When the basketball is flipped upside down, it will continue to veer—but now in the opposite direction, correcting the initial error.\n\n“That’s pretty clever, actually,” Sutor says. “It’s a little microwave pulse. That fixes some of the errors.”\n\nThe Quantum Elements paper specifically addressed quantum error correction in IBM’s 127-qubit superconducting processor. But these techniques might also be able to be generalized to other types of quantum computers, Sutor says.\n\nAnd any improvement in error correction will bring usable quantum computers that much closer. So will the other aspect of this announcement—the fact that the new error-correction technique was developed using Quantum Element’s AI-powered, digital-twin-style quantum computer simulator, Constellation.\n\nMost quantum computer simulators allow people developing quantum applications to test them in ideal environments. But real quantum computers have errors and noise. Quantum Elements’ simulator models that noise, allowing developers to test in near-real-world conditions.\n\nThere are also other simulation platforms, including IBM’s Qiskit Aer and Quantinuum’s H-Series Emulator.\n\nAccording to Medalsy, the simulators from IBM and Quantinuum use simplified models that don’t reproduce all the noise. “Quantum Elements’ digital twin is aimed at hardware-faithful simulation at experiment scale,” he says. “It is designed to preserve the full noise signature, both coherent and incoherent.”\n\nAnd the simulator can scale to experimentally relevant systems of around 100 qubits. Other platforms max out at 20 to 30 qubits, he says.\n\nThe third major quantum simulation platform is Nvidia’s Cuda-Q, which also has a noisy simulator. Google used it in 2024 to simulate the physics of its quantum processor with 40 qubits. But this simulation required the use of an Nvidia Eos supercomputer and 1,024 Nvidia H100 GPUs.\n\nSimilarly, in late 2025, a research team at the Jülich Supercomputing Center, was able to simulate 50 qubits, but only with the help of an exascale supercomputer.\n\nQuantum Element’s Constellation digital twin platform was released this past October, and the company has announced partnerships with IBM, Amazon, Rigetti, and Quantum Machines, as well as academic partners USC and UCLA.\n\nThe new error correction technique will be available in the coming month, says Medalsy.\n\n“We just came off a conversation with a company that is developing a new architecture for their superconducting qubits,” he says. “We are helping them to virtualize this technique and then giving them optimization tools to make the qubits perform with lower noise levels and higher fidelity.”\n\nQuantum Element’s new error correction technique and its AI-powered quantum digital twin platform will help companies that are designing quantum chips, the physicists researching the technology, and the computer scientists working on error correction codes, says Sutor.\n\nWith multiple vendors setting new error-correction performance records, the timeline to practical quantum computers has shifted, research firm Forrester said in a research report released last week. Quantum utility is now feasible within the next five years, the firm predicts.",
  "title": "Quantum Elements cuts quantum error rates using AI-powered digital twin"
}