RNG Must Use Pseudo Random Distribution

Here is the Utilities section of the forum.
http://www.grimdawn.com/forums/forumdisplay.php?f=32

GD Stash and Item Assistant are the primary ones you would want to consider using. I don’t hoard currently so I haven’t really needed to use them but as far as I know you can use them while the game is running when needed, just tab out and so on.

Much easier than messing with mules.

Sent from my SCH-I605 using Tapatalk

they have not been eliminated completely, only made less likely.

Something like

  • determine unique item
  • if same item already dropped in session
  • determine unique item again (no check for duplicate drop after this)
  • determine unique item
  • if same item already dropped in session
  • determine unique item again (no check for duplicate drop after this)

Ah, that’s identical to what I proposed initially without realizing such a mechanic already existed. If such is the case and enough people still feel that duplicates are being encountered even within the same session, perhaps increasing the iterations from 2 to 3 might do the trick.

That is, instead of one additional chance to choose a non-duplicate, have two additional chances.

Basically:


- determine unique item
- if same item already dropped in session:
    - determine unique item for the second time
    - if same item already dropped in session:
         - determine unique item for the third time

Doesn’t help my personal case since I’m constantly playing across sessions with different characters and accumulating a lot of duplicates across them, but mitigates at least per-session duplicates.

So basically, it is the ^2 of the chance. if 0.05 then it is 0.0025, yes it is fast and quick but that “session” resets after “exiting main menu” or “quit to desktop”?.. my guess is main menu which is not enough if you are farming a boss, or mats etc…

The math hurts my head and requires a variable in there since the probabilities change as more items are marked as already dropped (the chance of a duplicate increases in that case).

Assuming half the eligible items to be dropped have already been dropped, then the probability of a duplicate dropping with two iterations using that method is 33% (two dupes chosen in a row), while the chance of a non-duplicate dropping is 66% (non-dupe on the first try, or non-dupe on the second). 1/3 of the time, you will still get a dupe in that scenario.

With three iterations in such a scenario it would be 25% for a duplicate to drop and 75% for a non-duplicate to drop I think (doesn’t help me do math when drinking all night).

With four iterations it’d be 20% (1/5) for duplicate to drop, with five it’d be 16.7% (1/6), and so on.

It’s kind of a quick and dirty way to do it and hard to control to a precise level, but a very straightforward constant-time technique that’s cheap to compute. A way to guarantee against dupes in a single session is just remove the item from an eligible list of candidates once dropped. That would make it impossible for a duplicate to drop in the same session (but not sure impossible is really desirable here) and requires a variable-sized data structure in the mix to keep it constant-time which makes it more difficult to multithread efficiently (as well as safely) since atomicity has to apply at the data structure level instead of item level and so on and so forth (yada yada programming nerd talk).

Yes, as we progress in the game and acquired lots of epic+, the chance for non dropped items will increase if we are to first roll the same one we have dropped before… OTHERWISE ITS PROBABILITY INCREASE AMOUNT WILL BE ATOMIC LEVEL. So actually, it just increases all other item drop chance by (x^2/amount of epic+ pool) which is sad :frowning: