Geometric memory in incomplete phase transitions across dimensions

The study models a solid-state phase transition as plates (2D squares, 3D cubes, or square-faced lamellae) that grow in a self-similar way until they either hit a size limit or are stopped by other already-formed plates. To represent an incomplete reverse transformation, the model preferentially removes smaller plates while keeping larger ones, so the system does not fully “reset.” A later forward transformation then yields a changed plate-size distribution, which the authors treat as a memory effect. Extending an earlier 2D model, the authors compare how memory depends on dimensionality by running simulations for 2D, 3D, and 3DL cases. They introduce a quantitative memory descriptor (size mass ratio) and use additional analyses including growth snapshots, arrest-regrowth cycles, size distributions, and Shannon size-entropy, finding memory is robust across geometries but stronger in 2D.