Coherent defects, phonon jumps, and logical mitigation (10/09/26)

TL;DR A flux-pulse sequence that parks the qubit during the idle (no continuous flux during the energy-exchange wait) pushes SWAP spectroscopy—resonant qubit–TLS exchange scans—to coherent Rabi chevrons (oscillatory swap fringes vs detuning and wait time) at swap rates down to $0.1\mathrm{MHz}$. Defects once classed as incoherent show a median TLS energy-relaxation time of $10\mu \mathrm{s}$, and the dressed gap (hybridized qubit–TLS splitting) protects the swap from flux $1/f$ noise that limits the bare qubit $T_2^{\ast }$.

• Protocol: Fast flux is used only for state preparation and readout, then switched off during the energy-relaxation delay, enabling finer frequency resolution and longer coherent interaction times than continuous-flux SWAP sweeps.
• Sensitivity: Rabi chevrons appear at swap rates as low as $0.1\mathrm{MHz}$, about an order of magnitude below prior coherent-TLS reports.
• TLS lifetime: Fitting the coherent swap oscillations in the dressed-state picture yields a median TLS energy-relaxation time $T_1$ of $10\mu \mathrm{s}$ at coupling strengths previously associated with incoherent defects.
• Noise protection: Extracted swap coherence times exceed the qubit $T_2^{\ast }$ set by flux noise, consistent with the dressed gap shielding the hybridized exchange from low-frequency $1/f$ flux noise.
• Hardware stake: Sub-MHz coherent TLS spectroscopy reframes weak-coupling loss channels in cQED devices as coherent partners that can be timed and characterized, not only as incoherent sinks.

TL;DR A flux-tunable transmon with a $\mathrm{Au}$-on-$\mathrm{Nb}$ ground plane and island, plus strongly gap-engineered $\mathrm{Al}$ junctions (engineered superconducting gaps that push quasiparticles away from the junction), survives more than $10\times$ higher near-infrared optical power before excess relaxation than an otherwise matched all-$\mathrm{Al}$ device, with no resolvable quasiparticle-induced frequency shift. Extending gap engineering beyond the junction therefore hardens cQED hardware against optical and radiation-generated quasiparticles, even though dephasing and SPAM errors still rise with optical power.

• Main result: Relative to an all-$\mathrm{Al}$ transmon with nominally identical junctions, the $\mathrm{Nb}$ device withstands more than an order of magnitude higher near-infrared optical power before excess energy relaxation appears.
• Frequency stability: Over that same optical-power range, the $\mathrm{Nb}$ transmon shows no resolvable quasiparticle-induced qubit-frequency shift.
• Device approach: Gap engineering is extended beyond the junction by combining a $\mathrm{Au}$-on-$\mathrm{Nb}$ ground plane and island with strongly gap-engineered $\mathrm{Al}$ junctions that keep quasiparticles away from the junction leads.
• Remaining errors: Dephasing and state-preparation-and-measurement (SPAM) infidelity still increase with optical power while relaxation remains protected, so phase and readout channels are not fully shielded.
• Hardware implication: Engineering the quasiparticle energy landscape beyond the junction can raise resilience to nearby optical fields and radiation-induced error bursts in microwave–optical cQED platforms.

TL;DR A superconducting qubit dispersively coupled to a nanomechanical resonator enables repeated quantum nondemolition (QND; measurements that leave the measured phonon number intact) readout of mechanical energy. Aligned transfer-print integration yields a mechanical energy lifetime $T_1=2.1$ ms and a per-phonon qubit frequency shift $2\chi /2\pi =328$ kHz, enough to herald single-phonon states at 85% fidelity and resolve discontinuous jumps between $|1⟩$ and $|0⟩$ in a massive vibrating object.

• Main result: Repeated QND phonon-number measurements resolve quantum jumps between the nanomechanical resonator’s first excited state and ground state.
• Device metrics: Transfer-print integration of the qubit and resonator achieves mechanical lifetime $T_1=2.1$ ms and dispersive shift $2\chi /2\pi =328$ kHz per phonon.
• State preparation: Single-phonon states are heralded with 85% fidelity.
• Hardware implication: Energy (not displacement) coupling via a nonlinear qubit turns a long-lived nanomechanical resonator into a readable Fock-state sensor for cQED-style phonon control.

TL;DR A chain of fluxonium-like nodes—superconducting circuit elements whose continuous, unbounded phase variables encode a scalar field—forms a nonlinear transmission line that analog-simulates $(1+1)$-dimensional ${\varphi }^4$ theory without truncating the field Hilbert space. Tuning between single- and double-well on-site potentials opens real-time particle wave-packet scattering and topological kink–antikink collisions on cQED hardware.

• Hardware mapping: Continuous Josephson phases on fluxonium-like sites encode the scalar field $\varphi$, while anharmonic inductive potentials supply the ${\varphi }^4$ interaction and inductive coupling supplies kinetic terms—avoiding the usual bosonic-mode cutoff of digital encodings.
• Single-well regime: Circuit parameters that yield a single-well potential emulate unbroken ${\varphi }^4$ theory. Lattice Monte Carlo spectra of the circuit and of lattice ${\varphi }^4$ match at low energy, validating the correspondence for inelastic counter-propagating wave-packet scattering.
• Broken-symmetry regime: A double-well parameter set hosts degenerate vacua whose interfaces are topological kinks (solitons interpolating between the two vacua), enabling kink–antikink collision simulations.
• Protocols: The paper specifies state preparation and spatially resolved measurement sequences for both particle and soliton collision experiments, turning the line into a real-time interacting QFT lab on superconducting hardware.

TL;DR Virtual error cancellation (VEC)—bias-free logical error mitigation done entirely in classical post-processing—cancels residual logical bias using only syndrome records, with no noise learning and no circuit changes. In distance-7 rotated-surface-code simulations, a dual-decoder stack (fast online decoder plus a heavier offline decoder) drives the mitigated error down with sample count, exceeding three orders of magnitude of suppression relative to the decoded logical error in an experimentally accessible regime.

• Main result: VEC turns syndrome records into post-execution weights that cancel residual logical bias on universal logical circuits, without prior noise characterization or modified experiments.
• Method: A dual-decoder scheme keeps a low-latency decoder in the live loop and runs a high-complexity decoder offline to build mitigation weights from the same syndrome data.
• Beyond postselection: End-to-end protocols reduce sampling overhead past fundamental limits of syndrome-aware techniques that only discard shots, including syndrome-based postselection.
• Numerical evidence: On a distance-7 rotated surface code, the full-stack dual-decoder estimator's error shrinks as the number of samples grows and can exceed $10^3\times$ suppression versus the decoded logical error rate.
• Implication: Syndrome streams become a computational resource for correcting universal logical estimates after the quantum run finishes.