Welcome to the Quantum Realm. 

Enjoy today’s breakdown of news, research, & events within quantum.

🖥️ QuEra team, alongside Harvard and JILA, demonstrated the largest QML experiment to date with up to 108 qubits and effectively eliminated computational cost related to gradient optimization while demonstrating comparative advantage. Plus, IQM unveils 3 9s for CZ gate fidelity and an analysis of recent executive leadership changes in the industry.

🗓️ UPCOMING

Sunday, July 21 | QTM-X Quantum Education Series 6 of 10: Quantum Hardware

📰 QUANTUM QUICK BYTES

✨ IQM the latest to achieve record low error rates: It’s time for IQM Quantum Computers to shine with their announcement of a new record in superconducting quantum computing — a record low error rate for two-qubit operations with a CZ gate fidelity of 99.91% and qubit relaxation time (T1) of 0.964 milliseconds and dephasing time (T2 echo) of 1.155 milliseconds. These improvements were validated by interleaved randomized benchmarking and speak to the maturation of IQM's fabrication technology and its readiness to support next-generation quantum processors.

💰️ Planqc secures €50 million in latest funding round, which is a good thing with 3 computers on the horizon: Planqc, a spin-off from the Max Planck Institute of Quantum Optics and Munich Quantum Valley, recently secured €50 million in a Series A financing round, with €10 million coming from the German government's Deep Tech and Climate Fund. The company is focused on addressing talent shortages, particularly in electrical and mechanical engineering, to achieve its ambitious goals. Planqc is developing three major quantum computers: one for the German Aerospace Centre, a 1,000-qubit system for the Leibniz Supercomputing Centre, and an in-house quantum computer to offer cloud-based computing capacity to industrial clients in sectors such as chemicals, pharmaceuticals, automotive, and air conditioning technologies.

🔐 QuSecure and NVIDIA announce cupPQC: QuSecure has partnered with NVIDIA to support the newly launched cuPQC library, which uses NVIDIA GPUs to strengthen post-quantum cryptography and secure communications against quantum threats. This collaboration is in an earnest effort to protect sectors like telecommunications, finance, and critical infrastructure from quantum-induced vulnerabilities. QuSecure's involvement with NVIDIA under the Linux Foundation emphasizes the importance of developing robust PQC solutions, with cuPQC demonstrating significant acceleration of quantum-resistant algorithms on NVIDIA GPUs.

♻️ Leadership changes in quantum startups receipts of industry evolution: Several quantum computing companies, including Atom Computing, Infleqtion, Oxford Quantum Circuits, and Quantum Circuits Inc., have recently replaced their CEOs, with Pasqal and IQM shifting to co-CEOs. This trend, common among high-profile deep-tech startups, can be driven by various financial, personal, and strategic challenges as the industry progresses. New leadership may often be necessary to guide companies through different stages of development, secure funding, and drive innovation. While quantum computing faces challenges, such as competition for investment from AI, these leadership changes are part of the natural growth and maturation of the industry.

☕️ FRESHLY BREWED RESEARCH

LARGE-SCALE QUANTUM RESERVOIR LEARNING WITH AN ANALOG QUANTUM COMPUTER

QUICK BYTE: Researchers from QuEra Computing, Harvard University, and JILA propose a quantum reservoir computing approach using neutral atoms for quantum machine learning, demonstrating effective learning with up to 108 qubits. This method, combining quantum reservoirs with classical processing, addresses key challenges like barren plateaus and computational costs in QML. The study also shows comparative quantum kernel advantage and identifies a universal parameter regime.

PRE-REQS: 

• Reservoir computing is a type of machine learning approach that uses a dynamic reservoir to transform inputs into a higher-dimensional space. For this research, the concept is adapted to quantum systems, using the quantum dynamics of neutral atoms (Aquila) as the reservoir to process data within the need for gradient-based optimization.

• Gradient, relevant to machine learning, is a representation of the direction and rate of change of a loss function relative to the model’s parameters. Gradient-based optimization on quantum hardware is computationally expensive and relates to issues with barren plateaus.

• Barren-plateau is a phenomenon in quantum neural networks where the gradient of the cost function becomes exponentially small as the number of qubits increases, which makes training difficult.

• Kernel geometry (difference): In ML, a kernel defines the similarity between data points. Kernel geometry refers to the structure of data relationships. In this research, the concept of kernel geometry difference is used to quantify the difference between quantum kernel transformations and classical kernel transformations. This difference allows them to create synthetic datasets that illustrate the comparative quantum kernel advantage.

SIGNIFICANCE: Quantum machine learning offers potential advantages over classical machine learning due to the expansive Hilbert space available for computation. However, current QML approaches face significant challenges, including noise, barren plateaus, and the computational costs associated with variational parameter optimization and gradient estimation on quantum hardware. To address these issues, the researchers propose a paradigm shift inspired by classical reservoir computing, which avoids gradient optimization on quantum hardware. Their approach uses a quantum reservoir for feature encoding combined with classical pre- and post-processing to reduce computational demands. This methodology resembles variational quantum circuits in machine learning applications but without the overhead of parameterized circuit optimization.

The researchers demonstrate that certain datasets exhibit comparative quantum kernel advantage, suggesting the potential for quantum-enhanced learning in specific scenarios. This prompts further investigation into identifying suitable datasets and applications, including purely quantum tasks like learning Hamiltonian phase diagrams and classical problems such as drug candidate activity prediction.

RESULTS: 

• Effective learning with up to 108 qubits was experimentally demonstrated, making it the largest QML experiment to date

• Comparative quantum kernel advantage was observed experimentally, with performance improvements evident even with as few as 40 measurement shots per data point. This advantage, based on kernel geometry differences, suggests that quantum reservoir computing kernels closely resemble their noiseless counterparts even in the presence of experimental noise.

• The researchers identified a universal parameter regime that contributes to the versatility and noise resistance of their approach by eliminating the need for extensive parameter optimization on quantum hardware.

• The QRC algorithm shows robustness and scalability across various machine learning tasks, including binary and multi-class classification, as well as time series prediction.

• An important note is that depending on the data type, feature dimension reduction may be needed to fit the problem to hardware-tractable sizes.

HONORABLE RESEARCH MENTIONS: 

An experimental implementation of a two-client verifiable blind quantum computing protocol is conducted in a distributed architecture. Using a linear quantum network (Qline), this protocol allows clients to delegate private computations to remote quantum servers while ensuring the server's honesty through verification, even with untrusted quantum sources. Their results show successful verification of computations, supporting the development of a secure, scalable quantum cloud infrastructure. —> link to paper

A noise mitigation mechanism proposed for quantum teleportation is applicable to both discrete and continuous variable systems. By investigating non-Markovian decoherence dynamics, they demonstrate that forming a bound state in the energy spectrum can significantly suppress noise-induced decoherence, maintaining high fidelity in quantum teleportation. This approach shows potential for practical, noise-tolerant quantum teleportation by leveraging the constructive interplay between non-Markovian effects and bound state formation. —> link to paper

The researchers investigate the problem of optimizing entanglement distribution in a linear quantum network with a source, repeater, and destination. They introduce two polynomial algorithms, Purify-then-Swap (PtS) and Swap-then-Purify (StP), to manage entanglement purification and swapping, aiming to maximize fidelity and utility. Numerical results indicate that PtS outperforms StP, but the Swap-only approach, which omits purification, provides the best overall performance. —> link to paper

Eight-dimensional qudits, or qu8its, are explored for quantum simulations of the dynamics of 1+1D SU(3) lattice quantum chromodynamics (QCD). By reorganizing the Hamiltonian and leveraging recent advances in quantum hardware, they demonstrate that qu8its require significantly fewer two-qudit entangling gates than qubits, leading to higher fidelity and more efficient time evolution. This approach aims to enhance the capabilities of quantum simulations for non-Abelian lattice gauge theories. —> link to paper

UNTIL TOMORROW.