Qiskit Fall Fest 2026: MalaysiaControl, connection and evidence

A team spanning IBM Quantum, Rutgers University and several US research institutions reported in Nature Physics that adaptive monitored circuits can steer a quantum version of the chaotic Bernoulli map. On an IBM superconducting processor, local mid-circuit measurements, resets and conditional feedback competed with information scrambling to produce a transition between quantum and classical dynamics. Experiments reached 100 qubits and used nearly 5,000 entangling gates plus 5,000 non-unitary mid-circuit operations; the extracted universal critical behaviour agreed with noisy simulations and theoretical models.

This is a peer-reviewed control experiment and a substantial adaptive-circuit workload, not a practical-computation advantage. The task is a purpose-built model of chaotic dynamics, and it does not demonstrate an error-corrected general-purpose algorithm. Its value lies in showing that measurement and feedforward—also central to quantum error correction—can remain scientifically useful at a demanding scale.

Cornell University researchers Saswata Roy, Owen C. Wetherbee and Valla Fatemi introduced a dynamic-encoding method for controlling a logical qubit stored in an oscillator. Their peer-reviewed Nature Communications paper reports an error-semitransparent X, H and T gate set: when a photon-loss event occurred, the new controls reduced the resulting gate infidelity fivefold, extended the lifetime of information during active manipulation with error correction, and enabled an eight-gate composite non-Clifford operation.

The work addresses a key gap between preserving a logical qubit while it idles and operating on it without surrendering that protection. “Semitransparent” is intentionally narrower than fully error-transparent or fault-tolerant: the method handles most single-photon-loss errors during these gates, while ancilla faults and system-scale integration still require additional protection. It is an experimental logical-control result, not a complete fault-tolerant processor.

An IonQ-led collaboration experimentally entangled a barium-138 ion qubit with a silicon-vacancy centre in a diamond nanophotonic cavity. A converted photon from the ion interacted with the solid-state memory, and a single detection event heralded an ion–memory Bell pair. The 7 October preprint reports an average rate of 1.03(1) kilohertz and Bell-state fidelity of 87.9(7)%, roughly four times the earlier trapped-ion photonic-link rate.

Fast heralded links could connect specialised quantum modules without waiting far longer than a processor cycle, and mixing an atomic qubit with an efficient solid-state memory is an important systems idea. The evidence is nevertheless a preprint and a laboratory two-node demonstration. It does not yet show a distributed algorithm, fault-tolerant networking, useful distance or multi-node scaling; those are the tests that must follow the rate milestone.

Quantinuum and HQS Quantum Simulations reported an end-to-end nuclear magnetic resonance workflow on Quantinuum’s H2 trapped-ion system. After reducing a 22-spin molecular model to an effective 21-spin Hamiltonian, the team used 21 system qubits and 21 ancillas for leakage detection. Seventy Trotter steps and circuits containing more than 1,400 two-qubit gates reproduced a subtle double-peak feature in the simulated spectrum of 1,2-di-tert-butyl-diphosphane.

The hardware experiment is described in a preprint first posted on 15 September; the detailed company and Quantum Computing Report coverage appeared this week. It demonstrates deeper, chemically meaningful digital simulation, but not quantum advantage. The authors note that current classical tools remain more efficient for routine liquid-state NMR and that the approximately 0.07-parts-per-million spectral resolution is insufficient for practising spectroscopists. Longer simulated times and much deeper circuits are still needed.

DARPA advanced Atom Computing, Diraq, IBM and IonQ to Stage C of its Quantum Benchmarking Initiative after a year examining their development plans and risks. The four approaches—neutral atoms, silicon CMOS spin qubits, modular superconducting processors and trapped ions—join Microsoft and PsiQuantum in the programme’s final phase. Government evaluators will work alongside the companies to test whether proposed systems can be built and operated as their models and roadmaps claim.

QBI asks whether any architecture could reach “utility scale” by 2033, defined by DARPA as computational value exceeding total cost. Stage C is therefore a verification process, not evidence that the target has been reached. DARPA explicitly says the programme is not a contest that selects a winner; multiple, one or no approaches may ultimately satisfy its criteria. The significance is the move from roadmap review toward independent hardware and systems evidence.

This week’s industry coverage followed an Indian government announcement of a field trial connecting BISAG-N and IIT Gandhinagar. On the night of 27–28 September, QNu Labs’ pointing, acquisition and tracking system maintained a 5.56-kilometre, 1,550-nanometre free-space channel. The DPS-decoy protocol held the reported quantum-bit error rate below 5% and produced secure keys at 230–260 bits per second. Those keys were passed to BISAG-N’s Vedic Kavach platform, which combines post-quantum cryptography and quantum random-number generation.

The test matters as an end-to-end integration exercise and as a step toward optical links where fibre is impractical. It remains a point-to-point nighttime trial, however—not a continuously operated network or satellite link. Weather tolerance, availability, security analysis and scaling to multiple nodes will determine whether this architecture progresses beyond demonstration.