Feature - Implemented Let maximizer stop when it is not finding a better score

[roippi]

I don't own a single piece of brimstone or cloathing so practically I can't really be the one to fix this. The fix is also rather complex, so don't get your hopes up about it coming terribly soon.

Just use the already-existing preference to limit the number of combinations checked if you're having issues.

 

[Darzil]

It took me 3.5 hours to do the fix for Fluxx, several days to add in Crown/Bjorn/Card Sleeve. This is a bigger change. Each time you make a major change to maximizer you have to do loads of testing too, as you need to ensure you haven't broken a tool that many people rely on. I don't know when I'd expect to have time to work on this.

 

[jasonharper]

If you want to maximize faster, then stop using +equip - except in the case of using it with +outfit, in which case there's no other way to express that interaction. Instead, equip those items manually, then exclude their slots in the maximization.

This should perhaps be automatic, but there are some problem cases with doing that - in particular, if you +equip a 1-handed weapon and have DFSS, the weapon might have to go in the main hand (for use with an actual offhand item), or might have to go in the offhand (for use with a mainhand-only weapon).

 

[Fluxxdog]

If you want to maximize faster, then stop using +equip

I have to ask because it seems counter-intuitive: Why would using +equip make things take longer? It would seem that using +equip pantsgiving would eliminate all the pants-wearing combos. DFSS seems more an edge case than the norm.

Update: I adjust my wardrobe script so instead of using numerous equips, it equips the item in a predetermined slot and add a -slot and HOLY CRAP that's a hell of a lot faster! I would never have thought it.

 

[Crowther]

My recent work around has been to do "outfit birthday suit; maximize item, equip pantsgiving" or similar. Since I switched to getting naked before maximizing, I've yet to hit one that wasn't quick. I got the idea form the above discussion.

 

[Darzil]

I have to ask because it seems counter-intuitive: Why would using +equip make things take longer? It would seem that using +equip pantsgiving would eliminate all the pants-wearing combos. DFSS seems more an edge case than the norm.

Because what +equip does is not only consider X, but always consider X, so X will always get through to the combining items section, adding to the potential number of combinations. It then forces it afterwards. As Jason says, maybe it should only consider X, but when it can be in more than one slot that is problematic. Perhaps if it can only go in one slot then it should only consider X.

If you want to use an item, just equip it and -slot.

 

[lostcalpolydude]

I think I will add that to the documentation when I'm at a computer where I can do that.

 

[Fluxxdog]

Because what +equip does is not only consider X, but always consider X, so X will always get through to the combining items section, adding to the potential number of combinations. It then forces it afterwards.

Ah, I see. So if we tell it what pants to wear, it still calculates each pair of pants rather than just limiting the pool to that one pair. Thanks!

 

[Darzil]

Ah, I see. So if we tell it what pants to wear, it still calculates each pair of pants rather than just limiting the pool to that one pair. Thanks!

Yup, which makes sense for weapons that can go in either hand, accessories, and things that the familiar might be able to wear, but less so otherwise.

 

[ereinion]

Thanks a lot for your replies and suggestions, and thanks a lot for all the work that has already been put into the maximizer (and mafia). I am very grateful for that and hope I don't come across as very demanding

 

[PeKaJe]

My recent work around has been to do "outfit birthday suit; maximize item, equip pantsgiving" or similar. Since I switched to getting naked before maximizing, I've yet to hit one that wasn't quick. I got the idea form the above discussion.

Wow, this really makes a massive difference. I've not seen any that weren't significantly faster, and everything I've tried optimized to the same result. This makes me wonder, why *is* the currently equipped stuff added to the shortlist in the first place? I can see how it might shorten the optimization a bit, if you're already wearing the near-best gear and don't have a lot of synergy equipment. But this is cutting down tested combinations from millions and millions to a couple hundred thousand, for me. I'd prefer to not actually have to unequip stuff every time I'm optimizing, because that would likely cost me MP when my max is reduced. Maybe add an option/checkbox to not always consider current gear? Or just remove that functionality entirely, if no practical reason for its existence can be argued.

 

[lostcalpolydude]

Current equipment needs to be considered when -tie is used, to prefer it when there's nothing useful for the slot. Perhaps it could be removed as soon as something better is found for the slot though (I really don't know how practical that is).

 

[jasonharper]

This makes me wonder, why *is* the currently equipped stuff added to the shortlist in the first place?

If it didn't work that way, and you didn't have a useful item for every slot, there would be cases where the maximizer removes an item you had equipped, and replaces it with something completely useless (rather than leaving the slot empty). I recall some people getting really bent out of shape over that.

 

[ereinion]

How about if it first ignores all current equipment, and then considers if there are any slots which aren't improved by what's found? Or would you get in trouble with synergy bonuses again going that route?

 

[PeKaJe]

Couldn't that be done by looking at the final results when -tie is used, and keeping equipment in the slots with 0 score? Actually, I thought it already did so, but I haven't tried any really complex optimizations.

 

[jasonharper]

Couldn't that be done by looking at the final results when -tie is used, and keeping equipment in the slots with 0 score?

Due to outfit bonuses and other synergies, it can't really tell how much any one slot contributed to the overall score.

 

[PeKaJe]

I suspected that was the case. Still if the slot really ended up with a 0 score, in what situation could it actually have contributed? Maybe a stopgap would be to not specifically consider current equipment when -tie is not used? For the moment I've made a quick-and-dirty patch to add a checkbox on the modifier panel to selectively ignore equipped stuff. I'll keep an eye out for inconsistent results, but so far it's made me love the optimizer again. The difference between e.g. 430k and 9k combos on a simple "mox, ml", starting from +item gear, can really be felt.

 

[Theraze]

The other problem with that is that with the way the maximizer currently works with goals, if you weight something above a requirement, you can fail to meet the requirement. However, if the requirement starts out met by your current equipment, it won't break the requirement and will only consider equipment that leaves the requirement met. It's why getting proper spooky resistance for clue 3 of the Dreadscroll can be tricky with a single maximization. I always just "maximize hp" and follow that up with something like "maximize hp, 600 min, 900 max, 1000 spooky res" to put every point that isn't in health into spooky resistance... but if the current slot is ignored and the 'best' slot is always picked, the current workaround of 2 maximizations to actually meet your requirement would fail.

 

[roippi]

Still if the slot really ended up with a 0 score, in what situation could it actually have contributed?

outfit bonuses and other synergies

 

[xKiv]

How about, after (current) maximization is done, something like

for each slot {
  if original equipment can still be equipped in the slot and isn't already included in current best result (in that slot) {
    if score(current best)=score(current best with the original equipment placed back in that slot) {
      change current best to use the original equipment in that slot
    }
  }
}

That should keep the original equipment if it would get overwritten with useless equipment or emptiness, but not when it got overwritten with a synergetically useful thing.
And it should only cost fewer than 100 score calculations, instead of tens of millions.