Chinese quantum computing milestone 121 quantum bits, breaking the error correction break-even point

A big day for quantum computing in China! Today was a big day for quantum computing in China, as Dongling Deng of Tsinghua University, Haohua Wang of Zhejiang University and co-workers[1] demonstrated a 121-bit superconducting quantum processor with non-Abelian exchange statistics, setting a new record for the number of quantum bits in China, the most ever before being 66 quantum bits in Zuchongzhi II.

In another study, Yu Dapeng's team at the Southern University of Science and Technology collaborated with researchers from Tsinghua University, Fuzhou University and the Chinese University of Science and Technology[2] to "break the break-even point with logical quantum bits encoded in discrete variables", a milestone in quantum error correction.

01 121-bit superconducting quantum processor

This work by ZJU and Tsinghua has two important implications. Firstly, the 121-bit superconducting quantum processor sets a new domestic record and closes the gap with the international community; secondly, the team reports the observation of non-abelian arbitrators with up to 68 programmable superconducting transmon quantum bits, which is significant for topological quantum computing, which can be achieved by weaving and fusing non-abelian arbitrators to achieve this.

121-bit superconducting quantum processor at Zhejiang University

Researchers have observed non-abelian exchange statistics on two different quantum processors, called version I (above) and version II, both of which were fabricated using a flip-chip formulation described elsewhere [36]. Processor I (II) has an 11 × 11 (6 × 6) frequency-tunable array of transmon quantum bits with tunable couplers between adjacent quantum bits. On both processors, the maximum resonant frequencies of the quantum bits and couplers are approximately 4.8 GHz and 9.0 GHz, respectively. the effective coupling strength between two adjacent quantum bits can be dynamically tuned to -25 MHz. each quantum bit capacitively couples to its own readout resonator, designed at around 6.5 GHz, for quantum bit state measurements. In the experiments, 68 (20) of the 110 (36) functional quantum bits on processor I (II) were used, and their measured relaxation times and Hahn echo decoherence times at idle frequencies are shown below, with median values of T1 = 109.8 (137.5) μs and T2 = 17.9 (16.4) μs, respectively.

675cac284d25e96ce2185356f2e01b25

23d2c221ea5cfde0940db5ca7a08561c

Coherence times T1 and T2, Blue Line Processor I, Orange Line Processor II

Quantum manipulation on this superconducting quantum processor is physically achieved by applying analogue signals with continuous control parameters (e.g. amplitude and phase) to each quantum bit/coupler. The calibration procedure is a collection of experiments to learn and optimise these control parameters to enable full control of the processor. A systematic, automated and scalable calibration procedure is essential to achieve high-fidelity quantum bit operation across the entire device. They have divided the calibration process into two phases: a single-quantum-bit calibration phase for individually starting all quantum bits/couplers from scratch and collecting basic device and control parameters, and a multi-quantum-bit calibration phase for calibrating the processor at the system level and achieving high-fidelity single- and double-quantum-bit gates.

The fidelity of the single-quantum-bit gate errors, |0> and |1> are shown below, with median values of single-quantum-bit gate fidelity of 99.87% and 99.91% for Processor I and 36-quantum-bit Processor II, respectively, for 121 quantum bits.

b33c1f8479dc6cb6ff66cdf3051e98e4

e47c1bd2ac9596cd9ca7be0d37687058

Single quantum bit gate and readout fidelity cases, blue line processor I, orange line processor II

The double quantum bit CZ gate error is shown below, and when converted to a fidelity representation, the median fidelity values for processors I and II are 99.33% and 99.44% respectively.

077e0f4b66f61035797265317a371cab

Double quantum bit CZ gate bubble errors on processors I and II with sample sizes of 70 and 20, respectively, blue line I, orange line II

02Logical quantum bits break the break-even point

In the work of Academician Yu Dapeng's team, the break-even point for quantum error correction was broken. Specifically, in most quantum error correction (QEC) codes, a logical quantum bit is encoded in some discrete variable, such as the number of photons. Repeated QEC demonstrations have been reported on various platforms based on this discrete variable encoding, but the lifetime of the encoded logical quantum bit is still shorter than the lifetime of the best available physical quantum bit in the overall system, representing the break-even point at which any QEC code needs to be surpassed to be practically usable. Here, the team achieves a lifetime enhancement of about 16% beyond the break-even point, illustrating the potential of hardware-efficient discrete variable QEC codes for reliable quantum information processors.

The team demonstrated exceeding the QEC break-even point by performing real-time feedback corrections to discrete-variable photonic quantum bits in a microwave cavity whose codewords (codewords) remain mutually orthogonal and can be clearly distinguished. The main error of the logical quantum bit, i.e. the single photon loss, is mapped to the state of a non-linear oscillator based on a Josephson junction that is decentrally coupled to the cavity and used as an auxiliary quantum bit that is achieved by continuous pulses containing a cleverly tailored frequency component comb. As the driving frequency is targeted at the error space where the photon loss event occurs, perturbations on the logical quantum bit are highly suppressed while the logical quantum bit remains in the encoded logic space. Another intrinsic advantage of this error syndrome detection is that the continuous drive protects the system from the de-phase noise of the auxiliary device. They demonstrated this process with a lowest-order binomial code and extended the lifetime of stored quantum information by 16% over the best physical quantum bits.

A more important feature associated with this error detection process is that neither the logic space nor the error space needs to have deterministic parity, which allows the implementation of QEC codes that can tolerate more than one photon loss.

03d713281bb409ac2cf8813b7be1beed