There was an error while loading. Please reload this page.
greedy_alloc rewritten to the implemented algorithm, new needy_alloc page greedy_alloc.md described an algorithm that did not exist and listed shadow_prices among the results, which the operator never produces. Both pages now describe what 20.12.0 implements: one static ranking of the land units, a minimum-claims sweep and a maximum-claims sweep, no shadow prices and no reallocation, with bid_price therefore being the plain suitability of the allocated type. greedy_alloc ranks by best suitability, needy_alloc by regret (best minus next best). Neither regime primes the other and neither uses discrete_alloc as a preparation step. Both pages state plainly that they are heuristics: maximising total suitability over the land-unit/claim matroid intersection is what discrete_alloc solves exactly, and a single-pass greedy can fall short of it. Allocation-functions.md lists both families and says what separates them. See https://github.com/ObjectVision/GeoDMS/issues/1171 Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>