Sequence Reconstruction on Mage result question.

Topics: Rawr.Mage
Nov 24, 2011 at 1:28 PM
Edited Nov 24, 2011 at 1:29 PM

I get some very short time gaps before moonwell chalice is popped and the rest cooldowns are popped as well. Is there a rationale or should I just ignore them as minor noise?

It's not so simple btw, e.g.:


I got Moonwell Chalice and Heroism slightly before all the rest are popped, then on the second, Arcane Power and Moonwell Chalice together but slightly after them, it's mana gem.


On genetic solver.

A second run gave identical results.


Simple stacking is off.


edit: the first one:

Nov 25, 2011 at 10:19 PM

In the context of Rawr model that is quite likely the true optimum when balancing mana adept. You have to take into account though that Rawr model is continuous, that is it assumes that all cycles are infinitely divisible. In practice we're dealing with discrete events so such small offsets won't have any real impact. The problem of converting these continuous solutions to discrete ones is not at all easy and it's not always the case that a simple discretization will result in the best discrete solution. If this kind of optimizations interest you a lot I would suggest you try to model something like this in Simc and see how such offsets affect performance in discrete models. The only problem is that it can be very hard in most cases to model such offsets in Simc, but that is the closest you'll get to an actual answer.