Hi there,
It appears when I try to check the bis for hunters is always take the hit scope over the dmg scope.
When I however lock the dmg scope and search for bis then it gives a higher dmg output than the hit scope. Could you fix this that it takes the highest dmg output as actual bis?
Since the gear is pretty differant.
Can you post a snapshot id (using the help link) of a case where the chosen scope is reducing DPS?
f46d4fd08ef64352b29b17c390a717c7
Unlock the scope. It will go hit and change some pieces lowering dps
Then I noticed something else:
I locked the 7dmg scope and I got this:
264c6c76d9a248338875713d7d842f1b
It removed some of the T2 set. = +11,13% dps
Put in other set pieces T2 that he took out. Lock them and dps will be +13,09% dps.
I can take a look… the first case is extremely minor: the theoretical difference is well within the margin of error accepted by the optimization algorithm. The optimizer will almost always get the highest-scoring solution, but it is not a guarantee. The guarantee is that it will get very close.
(If you like, I can go into more detail on why the optimizer works like this – the short version is that it would be very slow and very expensive to do it otherwise, and you as a player would gain nothing from all that wasted time and money.)
The second case is probably an issue with it not checking for full 8-piece set bonuses aggressively enough… I’ll see if we can tweak that. In that case it’s coming out further from the theoretical optimal solution than desired.
scope differance is minor, but it’s still an upgrade and it should prioritize that. And 7 dmg scope is hell of a lot cheaper than the hit scope.
Right, but it’s not just the scope – it’s a completely different combination of items and enchants. There are many trillions of combinations of items+enchants available, thus it is impossible to inspect every single one.
Optimization algorithms use a combination of heuristics to search a subset of solutions with very high probability of being optimal or near optimal. If you start giving different information to the optimizer (like locking in certain items or enchants) then it changes the subset of solutions that will be searched, and you might end up with a slightly different result.
It would be nice if the optimizer always found the exact same solution, and it was guaranteed to be the one best of the many trillions available… but it just isn’t feasible. It would be very slow and very expensive, and you would gain almost nothing from it. Two sets of gear that are estimated by a theoretical model to be within 0.5% of each other are for all practical purposes interchangeable – use whichever you prefer.