Shallow entangled circuits for quantum time series prediction on IBM devices

Area of Science:

• Quantum Computing
• Machine Learning
• Time Series Analysis

Background:

• Forecasting temporal dynamics is crucial across science and engineering.
• Classical methods like neural networks are computationally expensive and require large datasets.
• Existing quantum approaches often involve deep circuits and high parameter counts.

Purpose of the Study:

• To investigate quantum entanglement as a resource for temporal pattern learning.
• To develop a Quantum Time Series (QTS) framework using shallow quantum circuits.
• To explore hardware-efficient encoding and entanglement schemes for time series analysis.

Main Methods:

• Proposed a Quantum Time Series (QTS) framework encoding sequential data into single-qubit rotations.
• Utilized forward and cross-entanglement layers to capture temporal correlations.
• Implemented phase encoding-based sparse entanglement for hardware efficiency.
• Conducted experiments on synthetic and geophysical datasets using IBM quantum processors.

Main Results:

• Shallow QTS circuits successfully reproduced complex temporal patterns from limited data.
• Phase encoding-based sparse entanglement demonstrated linear circuit depth and efficient two-qubit complexity.
• The QTS framework showed robustness and scalability up to 100 qubits.
• Outperformed deep variational quantum circuits in terms of parameters and depth.

Conclusions:

• Structured quantum entanglement can serve as a resource for temporal pattern learning.
• The QTS framework offers a scalable route for near-term quantum applications in time series analysis.
• Quantum entanglement may provide a short-term memory effect for efficient time-series forecasting.