Universal predictability of large avalanches in the Manna sandpile model

Area of Science:

• Complex Systems Science
• Statistical Physics
• Dynamical Systems Theory

Background:

• Self-organized criticality (SOC) describes systems exhibiting power-law dynamics without external tuning.
• Sandpile models, like Manna and Bak-Tang-Wiesenfeld, are fundamental to understanding SOC.
• Predicting extreme events in SOC systems has been a significant challenge.

Purpose of the Study:

• To investigate the relationship between system volume and the predictability of large events in Manna and Bak-Tang-Wiesenfeld sandpile models.
• To establish a scaling-based approach for forecasting extreme events in SOC systems.

Main Methods:

• Analysis of event size-frequency distributions in Manna and Bak-Tang-Wiesenfeld sandpile models.
• Development of a scaling-based framework to quantify prediction efficiency.
• Relating predictability to system volume and event size.

Main Results:

• Identified predictable extreme events in the Manna model, occurring beyond the power-law segment.
• Quantified prediction efficiency via a universal linear dependence on event size.
• Demonstrated that prediction efficiency scales with a power-law function of lattice volume.

Conclusions:

• The predictability of large events in the Manna sandpile model is linked to system volume.
• A scaling-based approach offers a method for forecasting extremes in SOC systems.
• Findings may inform predictions in other SOC phenomena, such as earthquakes and neural networks.