Comparing prediction efficiency in the BTW and Manna sandpiles

Area of Science:

• Complex Systems
• Statistical Physics
• Computational Physics

Background:

• Self-organized criticality (SOC) theory posits that certain systems naturally evolve to a critical state.
• Extreme events in SOC systems are often preceded by a period of reduced activity, suggesting potential predictability.
• The Bak-Tang-Wiesenfeld (BTW) and Manna models are canonical examples used to study SOC phenomena.

Purpose of the Study:

• To investigate the predictability of extreme events in the BTW and Manna models.
• To analyze how prediction efficiency scales with lattice length in these SOC models.
• To determine if inactivity-based prediction is universally applicable across different SOC universality classes.

Main Methods:

• Simulated the BTW and Manna models on square lattices of varying lengths.
• Employed an algorithm to forecast large events following a decrease in system activity.
• Quantified prediction efficiency by relating event size to lattice length using power-law functions.

Main Results:

• Prediction efficiency for both models scales universally with event size and lattice length.
• The Manna model exhibits a power-law exponent of 2.75, consistent with known scaling behavior.
• The BTW model shows a maximum power-law exponent of 3, indicating different scaling properties.
• In the thermodynamic limit, the Manna model demonstrates predictability based on inactivity, while the BTW model does not.

Conclusions:

• The predictability of extreme events in SOC models depends on their universality class.
• The Manna model's scaling suggests reliable prediction is possible, whereas the BTW model's scaling indicates limitations.
• Differences in universality classes explain the varying predictive capabilities based on preceding inactivity.