We propose and demonstrate an architecture for fluxonium-fluxonium two-qubit gates mediated by transmon couplers (FTF, for fluxonium-transmon-fluxonium). Relative to architectures thatexclusively rely on a direct coupling between fluxonium qubits, FTF enables stronger couplings for gates using non-computational states while simultaneously suppressing the static controlled-phase entangling rate (ZZ) down to kHz levels, all without requiring strict parameter matching. Here we implement FTF with a flux-tunable transmon coupler and demonstrate a microwave-activated controlled-Z (CZ) gate whose operation frequency can be tuned over a 2 GHz range, adding frequency allocation freedom for FTF’s in larger systems. Across this range, state-of-the-art CZ gate fidelities were observed over many bias points and reproduced across the two devices characterized in this work. After optimizing both the operation frequency and the gate duration, we achieved peak CZ fidelities in the 99.85-99.9\% range. Finally, we implemented model-free reinforcement learning of the pulse parameters to boost the mean gate fidelity up to 99.922±0.009%, averaged over roughly an hour between scheduled training runs. Beyond the microwave-activated CZ gate we present here, FTF can be applied to a variety of other fluxonium gate schemes to improve gate fidelities and passively reduce unwanted ZZ interactions.

Superconducting quantum processors comprising flux-tunable data and coupler qubits are a promising platform for quantum computation. However, magnetic flux crosstalk between the flux-controllines and the constituent qubits impedes precision control of qubit frequencies, presenting a challenge to scaling this platform. In order to implement high-fidelity digital and analog quantum operations, one must characterize the flux crosstalk and compensate for it. In this work, we introduce a learning-based calibration protocol and demonstrate its experimental performance by calibrating an array of 16 flux-tunable transmon qubits. To demonstrate the extensibility of our protocol, we simulate the crosstalk matrix learning procedure for larger arrays of transmon qubits. We observe an empirically linear scaling with system size, while maintaining a median qubit frequency error below 300 kHz.

The microscopic origin of 1/f magnetic flux noise in superconducting circuits has remained an open question for several decades despite extensive experimental and theoretical investigation.Recent progress in superconducting devices for quantum information has highlighted the need to mitigate sources of qubit decoherence, driving a renewed interest in understanding the underlying noise mechanism(s). Though a consensus has emerged attributing flux noise to surface spins, their identity and interaction mechanisms remain unclear, prompting further study. Here we apply weak in-plane magnetic fields to a capacitively-shunted flux qubit (where the Zeeman splitting of surface spins lies below the device temperature) and study the flux-noise-limited qubit dephasing, revealing previously unexplored trends that may shed light on the dynamics behind the emergent 1/f noise. Notably, we observe an enhancement (suppression) of the spin-echo (Ramsey) pure dephasing time in fields up to B=100 G. With direct noise spectroscopy, we further observe a transition from a 1/f to approximately Lorentzian frequency dependence below 10 Hz and a reduction of the noise above 1 MHz with increasing magnetic field. We suggest that these trends are qualitatively consistent with an increase of spin cluster sizes with magnetic field. These results should help to inform a complete microscopic theory of 1/f flux noise in superconducting circuits.

Routing quantum information between non-local computational nodes is a foundation for extensible networks of quantum processors. Quantum information can be transferred between arbitrarynodes by photons that propagate between them, or by resonantly coupling nearby nodes. Notably, conventional approaches involving propagating photons have limited fidelity due to photon loss and are often unidirectional, whereas architectures that use direct resonant coupling are bidirectional in principle, but can generally accommodate only a few local nodes. Here, we demonstrate high-fidelity, on-demand, bidirectional photon emission using an artificial molecule comprising two superconducting qubits strongly coupled to a waveguide. Quantum interference between the photon emission pathways from the molecule generate single photons that selectively propagate in a chosen direction. This architecture is capable of both photon emission and capture, and can be tiled in series to form an extensible network of quantum processors with all-to-all connectivity.

We consider mediated interactions in an array of floating transmons, where each qubit capacitor consists of two superconducting pads galvanically isolated from ground. Each such paircontributes two quantum degrees of freedom, one of which is used as a qubit, while the other remains fixed. However, these extraneous modes can generate coupling between the qubit modes that extends beyond the nearest neighbor. We present a general formalism describing the formation of this coupling and calculate it for a one-dimensional chain of transmons. We show that the strength of coupling and its range (that is, the exponential falloff) can be tuned independently via circuit design to realize a continuum from nearest-neighbor-only interactions to interactions that extend across the length of the chain. We present designs with capacitance and microwave simulations showing that various interaction configurations can be achieved in realistic circuits. Such coupling could be used in analog simulation of different quantum regimes or to increase connectivity in digital quantum systems. Thus mechanism must also be taken into account in other types of qubits with extraneous modes.

Dielectrics with low loss at microwave frequencies are imperative for high-coherence solid-state quantum computing platforms. We study the dielectric loss of hexagonal boron nitride(hBN) thin films in the microwave regime by measuring the quality factor of parallel-plate capacitors (PPCs) made of NbSe2-hBN-NbSe2 heterostructures integrated into superconducting circuits. The extracted microwave loss tangent of hBN is bounded to be at most in the mid-10-6 range in the low temperature, single-photon regime. We integrate hBN PPCs with aluminum Josephson junctions to realize transmon qubits with coherence times reaching 25 μs, consistent with the hBN loss tangent inferred from resonator measurements. The hBN PPC reduces the qubit feature size by approximately two-orders of magnitude compared to conventional all-aluminum coplanar transmons. Our results establish hBN as a promising dielectric for building high-coherence quantum circuits with substantially reduced footprint and, with a high energy participation that helps to reduce unwanted qubit cross-talk.

Particle transport and localization phenomena in condensed-matter systems can be modeled using a tight-binding lattice Hamiltonian. The ideal experimental emulation of such a modelutilizes simultaneous, high-fidelity control and readout of each lattice site in a highly coherent quantum system. Here, we experimentally study quantum transport in one-dimensional and two-dimensional tight-binding lattices, emulated by a fully controllable 3×3 array of superconducting qubits. We probe the propagation of entanglement throughout the lattice and extract the degree of localization in the Anderson and Wannier-Stark regimes in the presence of site-tunable disorder strengths and gradients. Our results are in quantitative agreement with numerical simulations and match theoretical predictions based on the tight-binding model. The demonstrated level of experimental control and accuracy in extracting the system observables of interest will enable the exploration of larger, interacting lattices where numerical simulations become intractable.

As progress is made towards the first generation of error-corrected quantum computers, careful characterization of a processor’s noise environment will be crucial to designingtailored, low-overhead error correction protocols. While standard coherence metrics and characterization protocols such as T1 and T2, process tomography, and randomized benchmarking are now ubiquitous, these techniques provide only partial information about the dynamic multi-qubit loss channels responsible for processor errors, which can be described more fully by a Lindblad operator in the master equation formalism. Here, we introduce and experimentally demonstrate Lindblad Tomography, a hardware-agnostic characterization protocol for tomographically reconstructing the Hamiltonian and Lindblad operators of a quantum channel from an ensemble of time-domain measurements. Performing Lindblad Tomography on a small superconducting quantum processor, we show that this technique characterizes and accounts for state-preparation and measurement (SPAM) errors and allows one to place strong bounds on the degree of non-Markovianity in the channels of interest. Comparing the results of single- and two-qubit measurements on a superconducting quantum processor, we demonstrate that Lindblad Tomography can also be used to identify and quantify sources of crosstalk on quantum processors, such as the presence of always-on qubit-qubit interactions.

Demonstrating the quantum computational advantage will require high-fidelity control and readout of multi-qubit systems. As system size increases, multiplexed qubit readout becomesa practical necessity to limit the growth of resource overhead. Many contemporary qubit-state discriminators presume single-qubit operating conditions or require considerable computational effort, limiting their potential extensibility. Here, we present multi-qubit readout using neural networks as state discriminators. We compare our approach to contemporary methods employed on a quantum device with five superconducting qubits and frequency-multiplexed readout. We find that fully-connected feedforward neural networks increase the qubit-state-assignment fidelity for our system. Relative to contemporary discriminators, the assignment error rate is reduced by up to 25 % due to the compensation of system-dependent nonidealities such as readout crosstalk which is reduced by up to one order of magnitude. Our work demonstrates a potentially extensible building block for high-fidelity readout relevant to both near-term devices and future fault-tolerant systems.

Interacting many-body quantum systems show a rich array of physical phenomena and dynamical properties, but are notoriously difficult to study: they are challenging analytically andexponentially difficult to simulate on classical computers. Small-scale quantum information processors hold the promise to efficiently emulate these systems, but characterizing their dynamics is experimentally challenging, requiring probes beyond simple correlation functions and multi-body tomographic methods. Here, we demonstrate the measurement of out-of-time-ordered correlators (OTOCs), one of the most effective tools for studying quantum system evolution and processes like quantum thermalization. We implement a 3×3 two-dimensional hard-core Bose-Hubbard lattice with a superconducting circuit, study its time-reversibility by performing a Loschmidt echo, and measure OTOCs that enable us to observe the propagation of quantum information. A central requirement for our experiments is the ability to coherently reverse time evolution, which we achieve with a digital-analog simulation scheme. In the presence of frequency disorder, we observe that localization can partially be overcome with more particles present, a possible signature of many-body localization in two dimensions.