Output Tracking of Periodically Time-Varying Boolean Networks: State-Flipped Control and Q-Learning Approaches.

Area of Science:

• Control Theory
• Computational Biology
• Discrete Dynamical Systems

Background:

• Periodically time-varying Boolean networks (PTVBNs) model complex biological rhythms and cyclic systems.
• Output tracking in PTVBNs is challenging due to periodic changes and partial state manipulation.
• Existing methods struggle with controller synthesis for PTVBNs under partial control.

Purpose of the Study:

• Investigate the output tracking problem for PTVBNs.
• Develop a state-flipped control strategy for manipulating network states.
• Establish criteria for trackability and synthesize controllers with reduced model dependence.

Main Methods:

• Formalized output tracking and state-flipped control using matrix-based representations.
• Developed an algebraic approach to analyze PTVBN trackability.
• Introduced a model-free reinforcement learning (Q-learning) scheme for controller synthesis.

Main Results:

• Established a comprehensive criterion for output trackability in PTVBNs.
• Identified the minimal state-flipping set required for successful output tracking.
• Validated the approach on biological models like repressilators and cell cycle networks.

Conclusions:

• The proposed state-flipped control strategy effectively addresses output tracking in PTVBNs.
• The combination of algebraic analysis and reinforcement learning offers a robust approach for controller synthesis.
• The findings have implications for understanding and controlling rhythmic biological processes.