Currently, every grid that the user defines creates an independent domain decomposition, which will be filled with data by the esdm_write_grid() and esdm_read_grid() calls. Even when the grid matches an existing grid exactly.
Somehow deduplicate the grids when they enter fixed axes state. Either an entire grid with the exact same axes, or share references to the stored fragments (hash table keyed on fragment shapes? or detection based on axis comparisons?).
Currently, every grid that the user defines creates an independent domain decomposition, which will be filled with data by the
esdm_write_grid()andesdm_read_grid()calls. Even when the grid matches an existing grid exactly.Somehow deduplicate the grids when they enter fixed axes state. Either an entire grid with the exact same axes, or share references to the stored fragments (hash table keyed on fragment shapes? or detection based on axis comparisons?).