Skip to content

Experimental progress of quantum machine learning based on spin systems

MetadataDetails
Publication Date2021-01-01
JournalActa Physica Sinica
AuthorsYu Tian, Zidong Lin, Xiangyu Wang, Liangyu Che, Dawei Lu
InstitutionsSouthern University of Science and Technology
Citations1
AnalysisFull AI Review Included

Technical Documentation: Quantum Machine Learning on Spin Systems

Section titled “Technical Documentation: Quantum Machine Learning on Spin Systems”

Source Paper: Tian Yu, Lin Zi-Dong, Wang Xiang-Yu, et al. Experimental progress of quantum machine learning based on spin systems. Acta Physica Sinica, 70, 140305 (2021).


This review highlights the critical role of spin systems, particularly Nitrogen-Vacancy (NV) centers in diamond, as robust platforms for advancing Quantum Machine Learning (QML) algorithms.

  • Core Achievement: Successful experimental verification of complex QML algorithms (HHL, QSVM, qPCA, QAE) using solid-state spin systems (NMR and NV centers).
  • Material Focus: High-purity Single Crystal Diamond (SCD) hosting NV centers is confirmed as a leading platform due to its high control precision and exceptionally long room-temperature coherence times ($T_{2}$).
  • Performance Metrics: Demonstrated high fidelity (>96%) for the Quantum Linear Systems Algorithm (HHL) and high accuracy (>99%) for Quantum Support Vector Machines (QSVM) in character recognition.
  • Coherence Enhancement: Quantum Autoencoders (QAE) were used to protect entanglement, extending the coherence lifetime of the spin system from 2 ”s to 3 ms (a >1000x improvement).
  • Algorithm Acceleration: QML algorithms implemented on these quantum platforms show potential for exponential or quadratic speedup compared to classical counterparts, addressing the computational limits of deep learning.
  • 6CCVD Value Proposition: The successful replication and scaling of this research critically depends on access to ultra-high purity, low-strain MPCVD Single Crystal Diamond (SCD) substrates, a core specialization of 6CCVD.

The following hard data points summarize the physical parameters and performance metrics achieved using spin systems, primarily NV centers in diamond, for QML experiments.

ParameterValueUnitContext
Core Spin SystemsNMR, NV CentersN/APlatforms for QML experiments
NV Center Qubit Count (Demonstrated)4QubitsAchieved using electron spin and coupled nuclear spins ($^{14}$N or $^{13}$C)
NV Ground State Splitting ($D_{gs}$)2.87GHzZero-field splitting of the NV electron spin
NV Excitation Wavelength532nmLaser used for spin initialization and readout
NV Emission Wavelength637-750nmFluorescence range used for spin state discrimination
QSVM Recognition Accuracy> 99%Achieved in hand-written character recognition
qPCA Eigenvalue Fidelity99%Achieved in human face recognition dataset processing
Coherence Time Extension (QAE)2 ”s to 3 msN/A>1000x improvement using Quantum Autoencoders
NV Coherence Time ($T_{2}$) (Room Temp)> 1.8msLongest reported single-spin $T_{2}$ in solid diamond
HHL Algorithm Fidelity> 96%Achieved on a 4-qubit NMR system

The experimental progress reviewed relies on advanced quantum control and hybrid classical-quantum optimization techniques applied to solid-state spin systems.

  1. Material Selection: Utilization of high-purity Single Crystal Diamond (SCD) to host isolated Nitrogen-Vacancy (NV) centers, ensuring minimal decoherence from environmental noise.
  2. Spin Initialization and Readout: Optical pumping using a 532 nm laser to initialize the NV electron spin to the $m_{s} = 0$ state, and subsequent fluorescence measurement (637-750 nm) for spin state readout.
  3. Quantum Control: Application of resonant microwave (MW) and radiofrequency (RF) pulses to precisely manipulate the NV electron spin ($S=1$) and coupled nuclear spins ($I=1/2$), enabling the construction of universal quantum logic gates.
  4. Quantum Algorithms Implemented:
    • HHL (Harrow-Hassidim-Lloyd): Demonstrated using phase estimation and auxiliary qubit rotation for solving linear equations.
    • AQC (Adiabatic Quantum Computing): Used for solving linear equations by slowly evolving the system Hamiltonian from a simple initial state ($H_{0}$) to a complex target state ($H_{p}$).
    • QSVM/qPCA: Implemented using hybrid classical-quantum control approaches (HQCA) where the quantum processor handles complex calculations (e.g., density matrix encoding) and the classical computer performs iterative optimization (gradient descent).
  5. Coherence Protection: Implementation of Quantum Autoencoders (QAE) to encode fragile entanglement information onto longer-lived nuclear spins, dramatically extending the effective coherence time.

The research reviewed demonstrates that high-quality diamond is the foundational material enabling scalable, room-temperature quantum computing and sensing based on NV centers. 6CCVD is uniquely positioned to supply the necessary materials and engineering support to replicate and extend this cutting-edge QML research.

To achieve the long coherence times ($T_{2} > 1.8$ ms) and high fidelity demonstrated in the paper, researchers require ultra-low nitrogen concentration and low-strain diamond.

  • Optical Grade Single Crystal Diamond (SCD): Essential for replicating NV center QML experiments. 6CCVD provides high-purity SCD grown via MPCVD, specifically engineered for quantum applications with extremely low native nitrogen content (P1 centers < 1 ppb).
  • Custom SCD Thickness: We offer SCD plates ranging from 0.1 ”m (ideal for surface-sensitive NV applications) up to 500 ”m, allowing researchers to select the optimal thickness for specific NV creation depths and optical integration.
  • Boron-Doped Diamond (BDD): For researchers exploring quantum sensing or electrochemistry applications related to QML, 6CCVD supplies custom BDD films (SCD or PCD) with controlled doping levels.

Scaling QML circuits requires precise material engineering beyond standard wafers. 6CCVD’s in-house capabilities directly address these needs:

Research Requirement6CCVD Custom CapabilityBenefit to Researcher
Integrated Control LinesCustom Metalization Services (Au, Pt, Pd, Ti, W, Cu)Enables direct fabrication of microwave/RF control lines and superconducting circuits onto the diamond surface for precise spin manipulation.
Non-Standard GeometriesPrecision Laser Cutting and ShapingProvides plates/wafers in custom dimensions up to 125 mm (PCD) or specific shapes required for optical setups and microwave resonators.
Surface QualityUltra-Low Roughness PolishingAchieves surface roughness $R_{a} < 1$ nm (SCD) and $R_{a} < 5$ nm (Inch-size PCD), critical for minimizing surface defects and maximizing optical coupling efficiency.
Large-Scale IntegrationLarge Area Polycrystalline Diamond (PCD)Offers PCD plates up to 125 mm in diameter, providing a pathway for scaling up QML architectures and sensor arrays.

6CCVD’s in-house PhD team specializes in MPCVD growth and material optimization for quantum technologies.

  • Material Consultation: We assist researchers in selecting the optimal diamond substrate specifications (purity, orientation, thickness) required to maximize NV center $T_{2}$ coherence times for similar Quantum Machine Learning projects.
  • Defect Engineering Guidance: We provide technical guidance on post-processing techniques (e.g., high-energy electron irradiation and annealing) necessary to create high-density, stable NV centers within our high-purity SCD substrates.

For custom specifications or material consultation, visit 6ccvd.com or contact our engineering team directly.

View Original Abstract

Machine learning is widely applied in various areas due to its advantages in pattern recognition, but it is severely restricted by the computing power of classic computers. In recent years, with the rapid development of quantum technology, quantum machine learning has been verified experimentally verified in many quantum systems, and exhibited great advantages over classical algorithms for certain specific problems. In the present review, we mainly introduce two typical spin systems, nuclear magnetic resonance and nitrogen-vacancy centers in diamond, and review some representative experiments in the field of quantum machine learning, which were carried out in recent years.