Category: Physics

Summary: Testing whether adaptive dissipation suppresses extreme sandpile avalanches at low control delay but becomes counterproductive when observations are delayed and noisy.

Self-organized critical sandpile models generate bursts of activity across a wide range of sizes, making them a standard setting for studying cascade control. This experiment asks whether feedback that adjusts dissipation in response to recent avalanche activity can reduce the risk of large events when control is timely, and whether the same strategy overshoots once sensing becomes delayed or noisy.

The simulation runs repeated sandpile trials on a two-dimensional lattice while comparing fixed dissipation with adaptive dissipation. It varies control delay, sensor noise, and correlation in where grains are added, then measures tail probabilities for large avalanches together with average avalanche size and control variability.

That design targets a regime inversion rather than a universal win for adaptivity. The scientific question is whether delayed feedback can turn a stabilizing policy into a destabilizing one in a critical system, especially when the external driving is itself temporally correlated.

Method: Repeated two-dimensional sandpile simulations comparing fixed and adaptive dissipation across control delays, sensor-noise levels, and correlated driving conditions.

What is measured: Avalanche-size tail probabilities, mean avalanche size, maximum avalanche, mean dissipation probability, control variance, and delay-driven adaptive-versus-fixed performance differences.