TL;DR
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
An IBM patent application filed May 5, 2025, and published as approved on Nov. 5, 2026, describes software for scheduling compatible users’ quantum jobs on the same processor while reusing stored noise information. It outlines a proposed approach, not a demonstrated system: the filing reports no measured gains in throughput, wait times or costs.
IBM is seeking patent protection for software that would schedule compatible users’ quantum programs together and reuse stored information about processor noise, according to a newly published application titled “Quantum Queue Optimization for Error Mitigation.” The proposed method aims to reduce repeated preparation for error mitigation and make more of a quantum processor available, but the filing reports no measured improvement in throughput, waiting times or costs.
The application was filed May 5, 2025, and, according to The Quantum Insider’s report, was approved on Nov. 5, 2026. It describes software that examines jobs waiting to run, identifies programs that might share a processor, and checks whether available qubits, connections, execution times and user preferences allow them to run together.
A central feature is reuse of noise models: stored descriptions of how errors affect particular operations on a processor. Some error-mitigation methods use extra experiments to measure those effects. If compatible jobs can use a model already collected for their combined operations, the scheduler could avoid repeating some of that work, the application says. The filing also describes taking new measurements or deriving needed information from existing models when a direct match is unavailable.
The application’s example uses a 127-qubit processor, with one program assigned 52 qubits and another assigned 30. Four unused qubits separate the groups to help limit crosstalk—the possibility that activity in one area affects another. The software would also check whether the required connections between qubits are available, account for users’ requested separation from other activity, and potentially place another job into a portion of the processor that becomes free while a longer program continues elsewhere.
Quantum processors are limited resources, and a program that needs only some of a device’s qubits can leave other areas unused if it runs alone. Running compatible jobs concurrently could improve utilization, while sharing relevant noise measurements could reduce the extra preparation some error-mitigation techniques require. The proposal addresses both issues in the same scheduling process.
That combination matters because simultaneous operations can also change how errors behave. A noise model measured for one arrangement of qubit connections may not apply to another. The application’s scheduler would have to weigh potential gains against interference, model accuracy and job priorities, rather than treating every free qubit as interchangeable. The filing identifies these as practical challenges, not solved outcomes.
For users, the proposed system could offer more opportunities to run jobs and, according to the application, billing rebates for those who permit sharing. Any benefit would depend on the processor, job mix, noise conditions and scheduling choices. The application does not provide experimental results showing whether users would wait less or pay less.
Products Worth Considering
Apple MacBook Air 13-inch Laptop with M5 Chip, 16GB Memory, 512GB SSD, 12MP Center Stage Camera, and Wi-Fi 7, Sky Blue
Bose Lifestyle Ultra Soundbar and Ultra Subwoofer Bundle
Amazon Kindle Paperwhite Signature Edition (12th Generation, 2024), 32 GB, Metallic Raspberry
As an affiliate, we earn on qualifying purchases.
How the Proposed Scheduler Works
In quantum computing, qubits and the operations performed on them are susceptible to noise. Error-mitigation methods can use additional measurements to characterize errors and help account for them in computational results. Those measurements take resources, and the relevant noise can vary with the operations and connections involved.
IBM’s filing proposes keeping models of measured noise and checking whether waiting programs can run together using information already available. Because processor noise changes over time, the software would update frequently used models and discard outdated ones. It could also move a job forward in a queue when doing so would let it use a model before that information becomes stale.
The filing describes different levels of flexibility: schedule jobs only when the required models already exist; collect additional measurements to create models; or use existing models to derive information for another operational pattern. It also discusses using machine learning to help decide where programs should run. The Quantum Insider has separately reported on Columbia Engineering’s HyperQ, which was tested on IBM processors and uses quantum virtual machines to assign parts of a processor to concurrent programs. The IBM application focuses on coordinating shared use with error-mitigation measurements.
“The application describes identifying programs that can run together and scheduling them to reuse previously collected information about processor noise.”
— IBM patent application, “Quantum Queue Optimization for Error Mitigation”
Performance and Deployment Remain Unknown
The application describes a method for scheduling and error-model reuse; it does not establish that IBM has implemented the system or that it improves performance on operating hardware. No measured throughput, wait-time or cost results are reported in the source account. The extent to which the approach would reduce repeated measurements in real workloads is therefore unknown.
It is also unclear how well the proposed scheduling would handle changing noise, interference between simultaneous jobs, and differences in users’ privacy or separation requirements. The filing itself identifies interference, possible information leakage and coordination across programs as challenges. The source report identifies inventors by reference to Patentlyze but provides only a partial name, so their full identities cannot be confirmed from the supplied material.
Evidence Needed Beyond the Filing
The next meaningful evidence would be a technical demonstration or deployment report describing how the scheduler performs on real processors and workloads. That would allow comparisons of processor utilization, error-mitigation overhead, queue times and costs against existing scheduling methods. The patent application alone does not give a timeline for such testing or indicate whether IBM plans to commercialize the approach.
Further details would also be needed on how the system keeps noise models current, protects information between users, and decides when the possible gains from sharing justify additional measurements. Until IBM or another source reports those results, the proposal should be understood as a patent-described scheduling approach, not a confirmed performance improvement.
Key Questions
What does IBM’s patent application propose?
It describes software that would group compatible quantum programs for simultaneous execution and reuse stored noise models where appropriate, potentially reducing some repeated error-mitigation measurements.
Has IBM shown that the system improves quantum computer performance?
Not in the information reported here. The application provides no measured results for throughput, queue times or costs, and the source material does not confirm a working deployment.
It would check available qubits and their connections, potential interference, execution timing, existing noise information, job priorities and users’ preferences about sharing or separation.
Why does reusing a noise model matter?
Some error-mitigation methods require extra measurements to characterize processor noise. Reusing a model that applies to the combined operations could avoid repeating some of that preparation work, though the application reports no measured savings.
What risks does the application identify?
The filing identifies possible interference between programs, the risk of information leaking between users, and the difficulty of coordinating different jobs. It does not establish how those challenges would be resolved in a deployed system.
Source: thequantuminsider.com
