Soft $0$-$\pi$ gates and pure-loss capacity (09/16/26)

TL;DR The authors give an error-bounded fixed-point recipe for flux-line predistortion filters—digital corrections that undo cable and bias-tee waveform warping before it reaches the qubit—and implement the FIR/IIR cascade (finite/infinite impulse response) on QubiC (an open-source FPGA control stack) at a 500 MHz fabric clock via a super-sample-rate datapath with scattered look-ahead pipelining (a restructuring that removes the IIR feedback loop from the critical path without changing the target transfer function). A representative cascade for 1 GS/s flux lines uses 88 DSP slices and adds 162 ns of latency, and they validate it by correcting characteristic bias-tee droop—an intermediate step toward full cryoscope-based calibration (qubit spectroscopy of the flux step response).

• Problem: Uncompensated flux-line distortion leaves residual detuning, conditional-phase error, and leakage on flux-activated two-qubit gates; commercial cryoscope calibrations usually hide the FPGA fixed-point formats that custom stacks must choose.
• Main result: Coefficient word lengths follow from a step-response error tolerance via bounds on quantization-induced pole/zero displacement over each distortion’s physical parameter range, and the accumulator format follows from the smallest input change that must remain resolvable at the output.
• Stability: The same analysis supplies conditions under which coefficient quantization preserves filter stability.
• Hardware: On QubiC, an SSR FIR/IIR cascade with scattered look-ahead pipelining meets timing at 500 MHz; a typical integrator + second-order section + 20-tap FIR costs 88 DSP slices and 162 ns latency.
• Validation: Hardware correction of bias-tee high-pass droop confirms the fixed-point realization as a step toward full cryoscope-based flux calibration.

TL;DR This work proposes the first unprotected microwave-activated CZ gate (controlled-Z: a conditional $\pi$ phase on $|11⟩$) for capacitively coupled soft $0$-$\pi$ qubits (a partially protected superconducting qubit with nearly degenerate ground states at phase $0$ and $\pi$). Under negligible disorder and a phenomenological noise model, simulations reach $\sim$99.9% fidelity in $\sim$160 ns, filling the two-qubit-gate gap left after Gyenis et al.'s soft-$0$-$\pi$ single-qubit demonstration.

• Main result: An unprotected CZ for two capacitively coupled soft $0$-$\pi$ qubits is driven by a microwave that couples a computational state to a higher non-computational level, then returns population to finish a conditional $\pi$ phase.
• Device setting: Capacitive coupling between soft $0$-$\pi$ circuits supplies the two-qubit interaction; the protocol does not rely on the full topological protection of an ideal $0$-$\pi$ qubit.
• Performance: With negligible device disorder and a phenomenological decoherence model, the optimized simulated CZ reaches about 99.9% fidelity at a gate time of about 160 ns.
• cQED implication: Soft $0$-$\pi$ hardware now has a concrete two-qubit primitive beside Gyenis et al.'s unprotected single-qubit gate, so multi-qubit soft-$0$-$\pi$ processors become designable rather than single-qubit only.

TL;DR Numerically, the conditional spectrum of coupled transmons—the interaction-split drive frequencies that fire only for chosen computational configurations—compiles a local recovery rule into a fixed open-loop pulse cycle, with no syndrome measurement or classical feedback. On a four-qubit repetition-code ring this input-independent control slows logical-coherence decay under Pauli-$X$ noise and holds logical-$|1⟩$ population against data relaxation better than uncorrected references, offering a hardware-native, bounded-degree path for measurement-free recovery in cQED.

• Main result: In interacting-transmon numerics, a local recovery predicate compiles into a fixed open-loop control cycle by addressing the conditional spectrum—drive tones that act only when neighbor states place the target on a selected transition.
• Hardware setup: A four-site repetition-code ring with nearest-neighbor couplings runs a two-color, bounded-degree recovery schedule so each site sees only a fixed local drive pattern, independent of the input state.
• Pauli-X channel: Relative to uncorrected references, the compiled cycle slows the decay of logical coherence under repeated Pauli-$X$ noise on the data qubits.
• Relaxation channel: Under data $T_1$ relaxation the same open-loop cycle stabilizes logical-$|1⟩$ population versus the unprotected ring.
• cQED implication: Recovery becomes a local microwave waveform rather than a measure–decode–feedback loop, so scaling maps onto bounded-degree pulse scheduling instead of wider classical syndrome wiring.

TL;DR This theory proposal replaces optically pumped microwave–optical transducers with a pump-free source of heralded microwave–optical Bell pairs—entangled pairs whose readiness is announced by a detection click—for teleportation-based quantum transduction (state transfer via entanglement rather than direct frequency conversion). A spin or atomic system resonantly coupled to microwave and optical cavities yields loss-robust pairs at kilohertz-scale heralding rates with high fidelity across color centers, atomic ensembles, and phonon-mediated platforms, easing the link from superconducting processors to optical networks.

• Main result: A resonant spin/atom–cavity assembly generates heralded microwave–optical Bell pairs without optical pumping, targeting teleportation-based transduction instead of coherent pump-driven conversion.
• Hardware implication: Removing the optical pump sidesteps pump-induced heating, noise, and cryogenic/optical-integration complications that currently limit microwave–optical interfaces for superconducting qubits.
• Mechanism: Resonant coupling of a spin or atomic system to both a microwave cavity and an optical cavity produces loss-robust entangled pairs whose success is heralded by photon detection.
• Platforms: The same architecture is analyzed for color centers, atomic ensembles, and phonon-mediated systems, with projected kilohertz-range heralding rates and high fidelity.
• cQED takeaway: For circuit QED hardware, the practical change is a Bell-pair resource for microwave–optical teleportation rather than an in situ pumped transducer on the dilution-fridge stack.

TL;DR This paper proves a strong converse (above the rate threshold, fidelity of entanglement generation cannot stay near one) for the unconstrained quantum capacity of the pure-loss bosonic channel (a bosonic mode mixed with vacuum on a beam splitter of transmissivity $\eta$). At every fixed rate above capacity, every code's entanglement-generation fidelity is $O(1/n)$ in the number of channel uses $n$, with no energy constraint and for arbitrary inputs and joint decoders.

• Main result: For the unconstrained pure-loss bosonic channel, any coding rate strictly above the quantum capacity forces the entanglement-generation fidelity of every code to decay at least as fast as a constant times $1/n$, the reciprocal of the number of uses.
• Scope: The bound needs no photon-energy constraint and covers arbitrary encoded states—including inputs correlated across all modes—and arbitrary joint measurements at the decoder.
• Balanced case: For transmissivity $\eta =1/2$, a uniform relative-entropy-variance bound (control on information fluctuations) follows by rewriting the beam splitter in bright and dark modes: the dark modes sit in vacuum, and any orthogonal component carries at least one dark photon.
• General transmissivity: Dilating the degrading attenuator (an extra loss map that weakens the channel to a degradable form) reduces the decoder test at general transmissivity $\eta$ to the balanced setting and inserts exactly the factor that recovers the known quantum-capacity threshold.
• Method: The argument combines quantum Chebyshev and hockey-stick testing inequalities (operator concentration and one-sided hypothesis tests on quantum states) with that variance bound to close the strong converse.