You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
A fundamental issue with the current implementation of heuristic mode is that it samples by randomizing each gear slot individually. This means that most of its samples cluster around "an even mixture of all of the affix options" and it essentially never tests, for example, "full ritualist," which will prevent it from being right about builds like quickbrand.
There are various ways to fix this. The way I've been thinking about for a long time involves running heuristics using percentage splits of full sets of gear (test 100% berserker / 99% berserker and 1% assassin / 98% berserker, 1% assassin, and 1% dragon / etc), but that would take some rewriting. A simpler way might just be to tweak the randomizer in a way that results in an even distribution (time to google some combinatorics terms).
The text was updated successfully, but these errors were encountered:
A fundamental issue with the current implementation of heuristic mode is that it samples by randomizing each gear slot individually. This means that most of its samples cluster around "an even mixture of all of the affix options" and it essentially never tests, for example, "full ritualist," which will prevent it from being right about builds like quickbrand.
There are various ways to fix this. The way I've been thinking about for a long time involves running heuristics using percentage splits of full sets of gear (test 100% berserker / 99% berserker and 1% assassin / 98% berserker, 1% assassin, and 1% dragon / etc), but that would take some rewriting. A simpler way might just be to tweak the randomizer in a way that results in an even distribution (time to google some combinatorics terms).
The text was updated successfully, but these errors were encountered: