PDA

View Full Version : Lightning Mace Abuse Question



Forged Fury
2010-12-22, 07:21 PM
I'm sure this has been addressed somewhere, but I cannot seem to find it in my searches and I am considering a Lightning Maces build. Honestly, I'm not going all out with Disciple of Dispater, Aptitude Weapons, etc. Just a regular build, but I did have a question.

The premise of the Lightning Mace abuse is to use aptitude weapons with higher threat ranges and feats/classes/etc that jack up your threat range to a point where just about any roll you make is within your critical threat range, which purportedly creates an extra attack regardless of whether you hit.

How does that square with the SRD discussion of Increased Threat Range, specifically the bolded part.
Increased Threat Range
Sometimes your threat range is greater than 20. That is, you can score a threat on a lower number. In such cases, a roll of lower than 20 is not an automatic hit. Any attack roll that doesn’t result in a hit is not a threat.
Can someone explain if I'm missing something? Maybe in Disciple of Dispater? Thanks!

Nanoblack
2010-12-22, 07:32 PM
I believe the bolded statement is saying that if you can only hit your target on a natural 20, and the crit range for you weapon is 19-20 or greater, the attack won't count for the purposes of lightning mace unless you actually hit.

AslanCross
2010-12-22, 07:35 PM
Yep. Only a natural 20 is an autohit. Even if your weapon calls for a critical hit at a natural 18, if the result of your attack still actually doesn't hit (Say, AB = 20, target's AC = 39), it still doesn't hit.

Forged Fury
2010-12-22, 07:39 PM
My reading agrees with what both of you are saying. Descriptions I saw in earlier Lightning Maces related-posts seemed to indicate you received a free extra attack even if you missed, as long as the die came up within the threat range. That seemed a little too strong to me...

Skorj
2010-12-22, 07:40 PM
This one is very simple by RAW - if you miss, you miss. There's no critical threat on a miss to roll to confirm, instead you just miss. People will stretch anything to make the more comical charop chalanges work, but that shouldn't be taken seriously.

FMArthur
2010-12-22, 07:41 PM
OP, I think you are misinterpreting a discussion you saw somewhere else that I foggily remember being about the fact that you only need to score a threat to gain the extra attack, which is easier than on a crit because it doesn't need to confirm and it works on things immune to crits. But scoring the threat at all requires that you hit your target on that roll, which is something that was never in dispute.

MeeposFire
2010-12-22, 07:41 PM
And any op board worth their dice will bring up the actual rules as well.

Glimbur
2010-12-22, 07:45 PM
Hitting generally isn't very difficult. One effective step is to put a wand chamber in one of your weapons and use a wand of wraithstrike. Touch attacks are easier.

Godskook
2010-12-22, 08:51 PM
Iirc, there are ways to auto-confirm critical hits.

If I don't, there's *DEFINITELY* ways to boost the confirmation roll higher than your attack roll.

AtomicKitKat
2010-12-22, 09:24 PM
Iirc, there are ways to auto-confirm critical hits.

If I don't, there's *DEFINITELY* ways to boost the confirmation roll higher than your attack roll.

But Lightning Mace cares not about Confirmation Rolls. As long as you're connecting your hits, and the rolls are in the critical threat range of your weapon, the extra attacks trigger.

The main conflict for most comes with the wording for Aptitude...

tyckspoon
2010-12-22, 09:28 PM
Iirc, there are ways to auto-confirm critical hits.

If I don't, there's *DEFINITELY* ways to boost the confirmation roll higher than your attack roll.

There are, although they generally involve spells that are a bit of a sidetrack for the 'canon' Lightning Mace build; Bless Weapon auto-confirms against evil foes, and Dolorous Blow both grants the same effect as Keen and auto-confirms everything. But it's a 3rd-level, Standard action, relatively short duration spell, which makes it fairly inefficient to use.

Especially since as mentioned up-thread Lightning Maces doesn't actually need you to confirm. It's nice if you do, because it means you can start feeding Blood In The Water, but you'll still get your extra attacks even if you don't.

JaronK
2010-12-22, 09:35 PM
Heh, I think I made this build.

The reason you (nearly) always hit with all critical threats is because you were using Blood in the Water with your attacks. You critically hit a ton, so your bonus to hit goes up extremely fast. As such, after the first round or so you should be hitting on a 9 anyway. Also, you were a Warblade so you get Int to confirm critical hits, raising the chances even more.

But yes, if a 10 would have missed then you just miss, even if you have a 9-20 crit threat range. You have to actually hit. It's blood in the water that makes sure you hit.

JaronK

Keld Denar
2010-12-22, 09:35 PM
The problem is that the aptitude weapon was PROBABLY intended to work like the Warblade's Aptitude ability in reverse. Instead of changing feats that apply to one weapon onto another, such as changing Weapon Focus(Dagger) into Weapon Focus(Longsword), the Aptitude weapon allows you to use those feats with it. So if you had Weapon Focus(Dagger), and you were wielding a +1 Aptitude Longsword, you would get to add your +1 to hit from Weapon Focus(Dagger) to it on top of all of your other bonuses. If you were a fighter with Focus, Spec, Greater Focus, Greater Spec, and Weapon Supremecy with a Greatsword, it might behoove you to carry a +1 Aptitude Morningstar as a backup weapon to handle DR issues without all of those feats being more of a waste than they already are.

Unfortunately, when they wrote it, they did a bad job, and nobody in editing caught it (was there even an editing department? Did they even run a spellcheck half of the time?). The way they wrote it COULD be interpreted in an abusive way, and thus it is touted about. This is one of those action economy things you should probably never let happen in your game. If you can get your threat range high enough, you can set off a near infinite number of attacks. Its never truely infinite because if you take the limit as attacks -> infinity, the chance of rolling enough natural 1s to stall out becomes a real possibility. VERY unlikely, but possible.

I wrote MatLab code to prove it [/geekcred]

JaronK
2010-12-22, 10:10 PM
If you can get your threat range high enough, you can set off a near infinite number of attacks. Its never truely infinite because if you take the limit as attacks -> infinity, the chance of rolling enough natural 1s to stall out becomes a real possibility. VERY unlikely, but possible.

I wrote MatLab code to prove it [/geekcred]

Wait, you did? Are you sure about this? I mean, it depends on the precise numbers... a character with a 9-20 critical hit chance, Lightning Mace, and Roundabout Kick with Aptitude Kukris is right on the edge, but any more attacks than that (8-20, for example) should go infinite every time. What numbers exactly did you use? Did you actually see it stall?

JaronK

Godskook
2010-12-22, 10:19 PM
Wait, you did? Are you sure about this? I mean, it depends on the precise numbers... a character with a 9-20 critical hit chance, Lightning Mace, and Roundabout Kick with Aptitude Kukris is right on the edge, but any more attacks than that (8-20, for example) should go infinite every time. What numbers exactly did you use? Did you actually see it stall?

JaronK

It only depends on one thing:

"Is there a case, such that the attack will fail to produce an additional attack"

If there is, than a stall is theoretically possible as one approaches infinite. If not, it goes infinite.

Keld Denar
2010-12-22, 10:44 PM
Dude, I posted the code in the thread you built it in using your own numbers. It's just an iterative while loop that records data in a 1 x n array and then sums it up when it stalls. I haven't actually run it, since I don't have a MatLab license anymore. I think I still have the code though...I'll look when I get home from Xmas vacation.

Ernir
2010-12-22, 11:03 PM
Wait, you did? Are you sure about this? I mean, it depends on the precise numbers... a character with a 9-20 critical hit chance, Lightning Mace, and Roundabout Kick with Aptitude Kukris is right on the edge, but any more attacks than that (8-20, for example) should go infinite every time. What numbers exactly did you use? Did you actually see it stall?

JaronK

Same thing happened when I did it, but I wasn't going full TO, the crit range was still very close to 10-20. Stalled in the hundreds of hits in all the test runs, if I recall.

The ironic thing is that I wrote the program to see whether a particular character build I had in mind would actually work. With a 15-20ish threat range, I was getting a damage count of ~200/round once the relevant parts of the build were completed. Imagine my surprise when I added in aptitude weapons, and suddenly the damage calculation went over 9000...

The funny thing is that I wrote my script in Matlab too. :smallbiggrin:


EDIT: Keld, I found your function (http://www.giantitp.com/forums/showpost.php?p=6825921&postcount=30).

Keld Denar
2010-12-22, 11:07 PM
Eh, MatLab is really good at what it does...math. unfortunately, its not good for anything else! LOL!

The Glyphstone
2010-12-22, 11:08 PM
At least it's better than Maple. Gaaah, Maple.

Coidzor
2010-12-22, 11:09 PM
I'm confused. With a crit range of, say 10-20, even if you hit on a roll between 2-9 (and have a way to always reroll natural 1s that come up), due to them being non-threats (and thus non-crits as well for blood in the water) they count as ending that loop and proceeding to the next iterative attack's loop.

What's in play that would change this? :smallconfused: ...Or does blood in the water make the roll count as higher than it really is for the purpose of threatening crits?


Eh, MatLab is really good at what it does...math. unfortunately, its not good for anything else! LOL!

So... It's good for reality as long as you know all of the necessary maths.

Ernir
2010-12-22, 11:10 PM
Eh, MatLab is really good at what it does...math. unfortunately, its not good for anything else! LOL!

Pff. Don't make excuses. I <3 Matlab forever.

Popertop
2010-12-22, 11:12 PM
OP, I think you are misinterpreting a discussion you saw somewhere else that I foggily remember being about the fact that you only need to score a threat to gain the extra attack, which is easier than on a crit because it doesn't need to confirm and it works on things immune to crits. But scoring the threat at all requires that you hit your target on that roll, which is something that was never in dispute.

so you can critically threaten something you can't critically hit?

Keld Denar
2010-12-22, 11:15 PM
Hey, its about the easiest programming language to learn. It's no C++, but its a thousand times easier to read and write. Lemme see if I can dig up the code...

Forged Fury
2010-12-23, 12:17 AM
OP, I think you are misinterpreting a discussion you saw somewhere else that I foggily remember being about the fact that you only need to score a threat to gain the extra attack, which is easier than on a crit because it doesn't need to confirm and it works on things immune to crits. But scoring the threat at all requires that you hit your target on that roll, which is something that was never in dispute.
I wasn't misinterpreting. But, I was probably putting a little too much importance on the fact no one corrected the error made by the posters. See, this thread (http://www.giantitp.com/forums/showthread.php?p=6825921), posts #9 & #25 to be precise. I thought it may have been wrong, that was why I requested clarification.

Thanks!

Thespianus
2010-12-23, 02:54 AM
Did you actually see it stall?
An interesting fact about the pseudo random number generators normally used in computers is that they very rarely produce the kind of "deserts" that would be required to make this simulation towards infinity ever stall out.

I'm not saying that the random number generator in Matlab is bad, it definitely isn't bad, just saying that it's not a truly random function and might have an inclination to not produce the long stretches of 1s needed to stall out the build.

But still, it's an amazing build and math is fun. Good job! :)

ShneekeyTheLost
2010-12-23, 08:55 AM
so you can critically threaten something you can't critically hit?

You have to critically threaten (rolling within your crit range) before you can confirm it as a critical hit (rolling again and backing it up).

You cannot, however, critically threaten if you miss

If you are asking about how this pairs with things like Fortification, then yes, you still threaten a critical, even if your opponent is immune to them.

Coidzor
2010-12-23, 11:46 AM
But why does one need a long string of ones? And how long of a string of ones are we talking here per iterative loop?

Or am I misinterpreting the scenario and the extra attacks get added on after the normal iteratives rather than being taken immediately after being triggered?

Because as far as I've understood the situation, one only needs a single natural 1 to get through per attack the character would have without lightning maces. So what am I not getting?

true_shinken
2010-12-23, 01:37 PM
So what am I not getting?
Let's say you have that 9-20 threat range. With Lighting Maces and Roundabout Kick and someone being completely crazy about Weapon Aptitude, that's +2 attacks for each roll on that range. Let's say you have BAB 16+, full TWF line, Snap Kick and haste on. That's 9 attacks. Each roll from 9-20 generates two extra attacks. Each of those extra attacks could generate more attacks if they roll 9-20 as well. Your number of attacks progresses very fast.

You probably missed Roundabout Kick in there, I guess.



The problem is that the aptitude weapon was PROBABLY intended to work like the Warblade's Aptitude ability in reverse. Instead of changing feats that apply to one weapon onto another, such as changing Weapon Focus(Dagger) into Weapon Focus(Longsword), the Aptitude weapon allows you to use those feats with it. So if you had Weapon Focus(Dagger), and you were wielding a +1 Aptitude Longsword, you would get to add your +1 to hit from Weapon Focus(Dagger) to it on top of all of your other bonuses. If you were a fighter with Focus, Spec, Greater Focus, Greater Spec, and Weapon Supremecy with a Greatsword, it might behoove you to carry a +1 Aptitude Morningstar as a backup weapon to handle DR issues without all of those feats being more of a waste than they already are.

Thanks for saying that. It's refreshing to see someone else doing this. :smallbiggrin:

Zeful
2010-12-23, 01:45 PM
Let's say you have that 9-20 threat range. With Lighting Maces and Roundabout Kick and someone being completely crazy about Weapon Aptitude, that's +2 attacks for each roll on that range. Let's say you have BAB 16+, full TWF line, Snap Kick and haste on. That's 9 attacks. Each roll from 9-20 generates two extra attacks. Each of those extra attacks could generate more attacks if they roll 9-20 as well. Your number of attacks progresses very fast.

You probably missed Roundabout Kick in there, I guess.

Only if the attack would hit on a 9. If you needed an 11 to hit the opponent in question only rolls from 11-20 generate extra attacks, because in order to critically threat the roll still needs to be able to hit the enemy, except if you roll a 20, which is automatically a hit.

ShneekeyTheLost
2010-12-23, 01:55 PM
Only if the attack would hit on a 9. If you needed an 11 to hit the opponent in question only rolls from 11-20 generate extra attacks, because in order to critically threat the roll still needs to be able to hit the enemy, except if you roll a 20, which is automatically a hit.

Hence Blood In The Water to give you a stupid attack bonus piling on with every crit.

You are right, it isn't truly infinite, because there is an infinitesimal chance that you will run into the string of non-threats which will eventually end the progression. However, as the number of free attacks you receive increases, the number of consecutive non-threats increases as well, in order to stall the attack progression. And therefore, as the iterations progress, it becomes steadily less and less likely for it to ever stop.

Not that it is really necessary, of course, since long before the action-chain stabilizes, any and all opponents in reach are already a fine red mist.

Thespianus
2010-12-23, 01:56 PM
Only if the attack would hit on a 9. If you needed an 11 to hit the opponent in question only rolls from 11-20 generate extra attacks, because in order to critically threat the roll still needs to be able to hit the enemy, except if you roll a 20, which is automatically a hit.

Yes, but once you get started with the chain reaction, Blood in the Water will give you a fantastic boost to hit, so all you need to do is to get the combination started, and it will self perpetuate, until you roll a signifikant number of 1s (or your opponent is dead, which WILL come before the desert of 1s)

AtomicKitKat
2010-12-23, 01:58 PM
Coidzor: As Shinken pointed out above, Roundabout Kick. The thing was only nigh-infinite before, due to basically letting you "tap the ball" over and over again(like Quagmire during the golf game on Family Guy), so that your previous attack, while active, gets replayed. After Roundabout Kick is thrown in(with Aptitude Boots/Catsuit or whatever), every hit that threatens a critical now produces 2 more chances to hit. It is now a chain reaction not unlike that of a nuclear bomb. Except that a nuclear bomb will eventually run out of material. :smallbiggrin:

*sez the nuclear feline*

Edit: Regarding the "1 desert", you need a desert of one "1" per attack that can generate the bonus attacks. After the first multiplication, you now need X+1 "One"s
Second time you get the bonus attacks, you now need X+2, and so on. Given that each 1 is a 1/20 chance (5%), you need say, 1/20^7(for the basic 7 attacks to all critically miss), or 1/1,280,000,000 (0.78125/1,000,000,000, or 78.125 of a MILLIARDTH of a %) chance to outright fail. Everytime you succeed, you just multiplied the difficulty by 20. For all practical purposes, it's as good as infinite.

Forged Fury
2010-12-23, 02:01 PM
Only if the attack would hit on a 9. If you needed an 11 to hit the opponent in question only rolls from 11-20 generate extra attacks, because in order to critically threat the roll still needs to be able to hit the enemy, except if you roll a 20, which is automatically a hit.
Now that I understand this more fully, I think I see where it's coming from. The biggest trick is to land that first hit, which should be at your full BAB and should have a pretty good chance of landing. High ACs are generally there to prevent iterative attacks from landing, not to prevent that first blow. If that first blow connects and falls within the unseemly threat ranges cited, it generates a number of extra attacks at the same attack modifier. With Blood in the Water, it probably takes a while just to work your way out of the top BAB tiers into your iterative attacks.

It probably would have been helpful if Lightning Maces indicated that critical threats generated from extra attacks provided through the feat would not generate extra attacks.

AtomicKitKat
2010-12-23, 02:11 PM
It probably would have been helpful if Lightning Maces indicated that critical threats generated from extra attacks provided through the feat would not generate extra attacks.

Except that they probably intended it to keep generating, because who would expect the lowly Light Mace to get that 1/400 chance to generate more than 1 bonus attack in a round? Or 1/361, if you had Improved Critical/Bludgeoning equivalent of Keen. Incidentally, I proposed(or maybe I never posted it, but did write a draft for) the "Octopus Drummer" several years back. Character abusing the hell out of the Tentacle Arm(might have been a different arm) Fiendish Grafts(RAW, Grafts only cost you the base body part if they say they do), in combination with Lightning Mace. Essentially, a drum soloist.:smallcool:

Forged Fury
2010-12-23, 02:13 PM
Except that they probably intended it to keep generating, because who would expect the lowly Light Mace to get that 1/400 chance to generate more than 1 bonus attack in a round? Or 1/361, if you had Improved Critical/Bludgeoning equivalent of Keen.
LOL! True, I guess my mind has been warped so much reading through some of these manipulations of the feat that I forgot that you generally only have a 10% chance of rolling a threat, assuming improved crit or impact.

Mongoose87
2010-12-23, 02:21 PM
Except that they probably intended it to keep generating, because who would expect the lowly Light Mace to get that 1/400 chance to generate more than 1 bonus attack in a round? Or 1/361, if you had Improved Critical/Bludgeoning equivalent of Keen. Incidentally, I proposed(or maybe I never posted it, but did write a draft for) the "Octopus Drummer" several years back. Character abusing the hell out of the Tentacle Arm(might have been a different arm) Fiendish Grafts(RAW, Grafts only cost you the base body part if they say they do), in combination with Lightning Mace. Essentially, a drum soloist.:smallcool:

For extra profit, start with a Thri-Kreen.

Thespianus
2010-12-23, 02:25 PM
For all practical purposes, it's as good as infinite.
For mathematical purposes, it's about as infinite as the number 1. ;)

AyeGill
2010-12-23, 02:30 PM
It's nigh infinite. Since the probability of it ending is fairly small to begin with, and keeps approaching zero as the attack sequence progresses, mathematically, there's no amount of hp that will statistically be able to withstand it.

While it will always end at some point, thus preventing true infinity, it will make an arbitrarily high amount of attacks before finishing.

Thespianus
2010-12-23, 02:35 PM
While it will always end at some point, thus preventing true infinity, it will make an arbitrarily high amount of attacks before finishing.
There's a very high probability that the chain reaction can inflict an arbitrarily high number of attacks. ;)

Ok, I'm done with the nit picking.

EDIT: How does DR work, by the way? If I pull this trick, and attack a monster with 400 billion attacks during one round, and I never score above his DR value, will he take any damage? That is: Is DR counted "per attack" or "per round"?

I should know this.

Kansaschaser
2010-12-23, 02:37 PM
I'd like to know what (ability/feat/spell/magical property) allows you to get an extra attack after you land a critical. I can't seem to find anything in 3.5. Is this a Pathfinder or D&D 3.0 ability/feat/spell/magical property?

Thespianus
2010-12-23, 02:39 PM
I'd like to know what (ability/feat/spell/magical property) allows you to get an extra attack after you land a critical. I can't seem to find anything in 3.5. Is this a Pathfinder or D&D 3.0 ability/feat/spell/magical property?

The Feat "Roundabout Kick", Complete Warrior.

With Aptitude Weapons you can gain the benefit of Roundabout Kick with a weapon with a higher crit range.

AyeGill
2010-12-23, 02:40 PM
There's a very high probability that the chain reaction can inflict an arbitrarily high number of attacks. ;)

Ok, I'm done with the nit picking.

no, its a fair point. Although 'arbitrarily high' doesn't really mean anything except from 'enough to kill everything within range, and cut the corpses into a paste for good measure' in this context. But your point stands.

Also, DR is per attack.

ShneekeyTheLost
2010-12-23, 02:42 PM
There's a very high probability that the chain reaction can inflict an arbitrarily high number of attacks. ;)

Ok, I'm done with the nit picking.

EDIT: How does DR work, by the way? If I pull this trick, and attack a monster with 400 billion attacks during one round, and I never score above his DR value, will he take any damage? That is: Is DR counted "per attack" or "per round"?

I should know this.

You need to penetrate DR, otherwise you are dealing no damage, and damage is per attack.

However, this is what Shards of Granite was designed to negate.

Thespianus
2010-12-23, 02:46 PM
You need to penetrate DR, otherwise you are dealing no damage, and damage is per attack.

However, this is what Shards of Granite was designed to negate.

I hadn't studied the build all the way through. Using Shards of Granite is indeed clever. :)

ShneekeyTheLost
2010-12-23, 02:49 PM
I hadn't studied the build all the way through. Using Shards of Granite is indeed clever. :)

Also keep in mind you are racking up damage bonuses from Blood In The Water as well. Eventually, you'll bust through DR. And when you do...

true_shinken
2010-12-23, 03:00 PM
How does DR work, by the way? If I pull this trick, and attack a monster with 400 billion attacks during one round, and I never score above his DR value, will he take any damage?
Imagine this. You go there, all catsuit-y (loved the term, AtomicKitKat), hit him 30 times or something... for no damage.
No effect?! I need a bigger sword...

Thespianus
2010-12-23, 03:01 PM
Also keep in mind you are racking up damage bonuses from Blood In The Water as well. Eventually, you'll bust through DR. And when you do...
I just noticed that the Blood in the Water gives you bonuses for any attacks, not just on the same opponent.

Is there a good way to move from one opponent to another with the build? Cleave, Great Cleave and Supreme Cleave would be one way, but I figure the build is feat starved as it is?

I guess I should just google up the build myself.

Thespianus
2010-12-23, 03:02 PM
Imagine this. You go there, all catsuit-y (loved the term, AtomicKitKat), hit him 30 times or something... for no damage.
No effect?! I need a bigger sword...

Once you throw away your Aptitude Kukri for a bigger sword, you're just a sweaty harmless guy panting for breath from waving your arms around at supersonic speeds. ;)

Coidzor
2010-12-23, 03:09 PM
And DR is why you have a dragonfire inspiration bard friend, I imagine.

I mean, you're breaking the sound barrier anyway, might as well add some flames to the mix.

Also, wow. And it doesn't freeze time like the 1d2 crusader, either.

tyckspoon
2010-12-23, 03:32 PM
And DR is why you have a dragonfire inspiration bard friend, I imagine.

I mean, you're breaking the sound barrier anyway, might as well add some flames to the mix.

Also, wow. And it doesn't freeze time like the 1d2 crusader, either.

That or as mentioned you can just wait until the Blood In The Water bonuses outweigh the DR. And you're critting a lot anyway; doubling your Strength and enhancement bonuses should be enough to get you chip damage even before your bonuses start stacking.

JaronK
2010-12-23, 03:49 PM
I just noticed that the Blood in the Water gives you bonuses for any attacks, not just on the same opponent.

Is there a good way to move from one opponent to another with the build? Cleave, Great Cleave and Supreme Cleave would be one way, but I figure the build is feat starved as it is?

Lion Totem Barbarian means you kill one opponent, then charge and destroy the next. Getting Large gives you more reach to work with.

But yeah, with Blood in the Water eventually you'll crack any DR. I also liked making the guy Necropolitan with a pair of Enfeebling Aptitude Lifedrinker Kukris, so your hits cause negative levels in addition to ever increasing damage.

JaronK

Thespianus
2010-12-23, 03:56 PM
Lion Totem Barbarian means you kill one opponent, then charge and destroy the next. Getting Large gives you more reach to work with.
But you still can only make one charge per round, right?

Going Large means a bigger reach, but with the extra attacks you get from Roundabout Kick, you can only attack the same opponent.

However, yes I see it now: the Lightning Mace feat gives you an extra attack period. Doesn't mention "the same opponent" as the Roundabout Kick-feat does.

My thought was to be able to go between multiple enemies in the same round, and I thought the only way to do that was to use Cleave-like feats. But Lightning Mace gives you more attacks that you can target a different opponent with.

Claudius Maximus
2010-12-23, 04:03 PM
You could 5 (or 10) step, or use something like Travel Devotion between attacks in the full attack. You can take out more than one person this way.

Forged Fury
2010-12-23, 04:28 PM
Could you dervish dance this?

Thespianus
2010-12-23, 04:37 PM
You could 5 (or 10) step, or use something like Travel Devotion between attacks in the full attack. You can take out more than one person this way.

Yeah, I wanted something that could make the Lightning Mace:er take out a whole band of enemies, using his insane amount of attacks against alot of enemies, but I can't make it happen. 5/10 ft step, Abrupt Jaunt and Cleave-variants are the few exceptions. Travel Devotion might work too, but then you lose your 5 ft step.

true_shinken
2010-12-23, 04:46 PM
Yeah, I wanted something that could make the Lightning Mace:er take out a whole band of enemies, using his insane amount of attacks against alot of enemies, but I can't make it happen. 5/10 ft step, Abrupt Jaunt and Cleave-variants are the few exceptions. Travel Devotion might work too, but then you lose your 5 ft step.

Shadow Jaunt. Or Sun School, since you're in a Catsuit anyway.

AyeGill
2010-12-23, 04:51 PM
Yeah, I wanted something that could make the Lightning Mace:er take out a whole band of enemies, using his insane amount of attacks against alot of enemies, but I can't make it happen. 5/10 ft step, Abrupt Jaunt and Cleave-variants are the few exceptions. Travel Devotion might work too, but then you lose your 5 ft step.

extending aptitude kukri + large size?

FMArthur
2010-12-23, 04:58 PM
Kukris have a nice threat range, but you're just better off getting Aptitude Splitting Hand Crossbows instead to generate two attack rolls from every attack. It turns from a possibly-endless loop attack into a nigh-infinite fractal attack.

Kansaschaser
2010-12-23, 05:01 PM
If there is someway of taking a few levels of Bloodstorm Blade, then you can make melee attacks as ranged attacks. I think 4 levels should do it. But again, that takes some more feats. :smallfrown:

Claudius Maximus
2010-12-23, 05:42 PM
Kukris have a nice threat range, but you're just better off getting Aptitude Splitting Hand Crossbows instead to generate two attack rolls from every attack. It turns from a possibly-endless loop attack into a nigh-infinite fractal attack.

Though sometimes you don't want it to. (http://www.giantitp.com/forums/showthread.php?p=6968457)

Thespianus
2010-12-23, 05:43 PM
Shadow Jaunt. Or Sun School, since you're in a Catsuit anyway.
Shadow Jaunt is a Standard Action, right?

FMArthur: You need a way to reload the hand crossbows, real fast, and I'm not sure you get that as a free action? At least not a free action that doesn't violate the gist of Free Actions in the PHB. ;)

Greenish
2010-12-23, 05:52 PM
Shadow Jaunt is a Standard Action, right?Maybe he meant stride or blink.

Claudius Maximus
2010-12-23, 05:56 PM
I think he means Shadow Pounce, an ability that lets you full attack every time you teleport. It's like Sun School but better, so it fits in his post that mentions Sun School.

true_shinken
2010-12-23, 09:34 PM
I think he means Shadow Pounce, an ability that lets you full attack every time you teleport. It's like Sun School but better, so it fits in his post that mentions Sun School.

Yeah, that's what I meant. Sorry for confusing everyone.

FMArthur
2010-12-23, 11:21 PM
Shadow Jaunt is a Standard Action, right?

FMArthur: You need a way to reload the hand crossbows, real fast, and I'm not sure you get that as a free action? At least not a free action that doesn't violate the gist of Free Actions in the PHB. ;)

You can do it easily with Rapid Reload (by which I mean Hand Crossbow Focus) and a Spare Hand, or Gloves of Storing, or a prehensile tail, or extra limbs, or a creature of yours who acts on your turn, or use that one spell whose sole purpose is reloading crossbows, etc etc.

You could probably just do it with one crossbow instead of TWFing. The real wierd thing you'd want to do is get one made primarily out of iron to take advantage of Disciple of Dispater.

Thespianus
2010-12-24, 05:46 AM
Yeah, that's what I meant. Sorry for confusing everyone.
I'm easily confused. ;) I still don't have ToB internalized in my 3.5 brain.

JaronK
2010-12-24, 05:52 AM
Large size means reach, which may let you kill multiple opponents. Also, Sudden Leap is a first level maneuver that will let you leap part way through an attack to engage a new target.

JaronK

2xMachina
2010-12-24, 08:47 AM
If there is someway of taking a few levels of Bloodstorm Blade, then you can make melee attacks as ranged attacks. I think 4 levels should do it. But again, that takes some more feats. :smallfrown:

For feats, just need Point Blank Shot to enter that PrC.

term1nally s1ck
2010-12-24, 09:16 AM
Re: Infinite discussions, the loop in the better builds does indeed tend to infinity. (assuming roundabout kick.)

Basically, it is trivial to prove (Sum-to-infinity of a simple formula) that there is a finite chance of the loop ending in finite iterations, depending on the exact numbers. IIRC, the 8-20 has about a 1/3 chance of ending at a finite number of attacks, which means that in 2/3 of the times you use it you will make a number of attacks that is not finite.

EDIT: I'm not sure if the standard builds for this stuff have ANY feats spare. the crit-boosting PrCs are VERY feat-intensive.

AtomicKitKat
2010-12-24, 11:41 AM
So now we have something that looks like Quicksilver(catsuit) wielding Wolverine Claws(of Aptitude), slashing discorporating his opponents in a blur of flashing white attacks, while bouncing about between groups like Nightcrawler(Shadow Pounce)? I like it!:smallcool:

Thespianus
2010-12-24, 04:46 PM
Basically, it is trivial to prove (Sum-to-infinity of a simple formula) that there is a finite chance of the loop ending in finite iterations, depending on the exact numbers. IIRC, the 8-20 has about a 1/3 chance of ending at a finite number of attacks, which means that in 2/3 of the times you use it you will make a number of attacks that is not finite.

No. Infinity is tricky. If there's a possibility > 0 that the series will end in a finite number of steps, the series *will* end in a finite number of steps.

Look at it this way: The possibility for the series to end in a finite number of steps is only dependent on the random event of rolling a sufficient number of 1s. That will eventually happen, no matter how many 1s you have to roll, and the series will therefore terminate in a finite number of steps.

The number of steps will be insanely high, yes, but it is a finite number for each run of the series.

EDIT: Does anyone have a link to "the" build? I can't seem to find it... :(

term1nally s1ck
2010-12-24, 06:14 PM
No, it's not. I can PROVE that the total probability of the sequence terminating over all of the finite numbers is bounded above by a number that is smaller than 1. That means that there is >0 probability of the number of attacks you make being completely unbounded. That means that the cumulative number of attacks you make has NO least upper bound with positive probability, and hence it tends to infinity.

Infinity is tricky yes, but when dealing with sequences and series, the rules are set out VERY rigorously, and this particular sequence tends to infinity with positive probability provided your crit range is at least 10-20. (Or for smaller ranges if you have another way to gain attacks with each crit, like Splitting.)

jseah
2010-12-25, 01:16 AM
RE: infinity.

I've asked a 3rd year math student at my university here.
You have one coin. You flip it. If it lands heads, you flip twice more, if it lands tails, you don't. Additional flips generate more flips.
Do you get an infinite number of flips?

He said this: (paraphrased heavily, it's been some time)
"Yes. On average you get an infinite number of flips.
Average number of coin flips generated per coin flip = 1. Hence, on average, you get an infinite number.
More precisely, the probability you get at least X number of flips in total drops towards zero slower than the number X increases towards infinity.

However, at the same time, the flips WILL stop. Assuming you do the flips one at a time, the number of flips you have left to do can be plotted over time.
Imagine, Y is the number of flips you have. Every time you go 1 along the X axis, you take one off the Y axis and 1/2 the time add 2.
You start at 1 and end if you ever hit zero.

It will stop, and it will stop at 0. Simply because that line travels up and down and only ever stops at 0. Since the algorithm can make it go to any positive integer (and 0) from any other positive integer, eventually, it will hit zero.
After an infinite number of tosses, all trials will be at zero. "

**************

That was a bit confusing and I can't remember the names of some of the terms he used. But basically, just because the average number of attacks is infinity, doesn't mean that the attacks cannot stop. The average can be infinity and the attacks can stop.

It just means that you can have any finite amount of hp, no matter how high, and this has a good chance of destroying it.

term1nally s1ck
2010-12-25, 01:32 AM
That is known as a Random Walk and is known to have probability 1 of ending.

However, that isn't true when the probabilities are not 50-50. The Expected value at every step in the 50-50 version is 1. If there's a bias towards increase, then the sequence of 'attacks remaining to be made' tends to infinity, as does the cumulative sequence, which is the number of attacks you make.

Put it this way, by proving that there's an upper bound of 1/2 (which is trivial to do for the 8-20 case, iirc) on the sum of probabilities, we can set up an infinite number of players all starting an attack routine, and over the entire set of integer rounds, at least 1/2 of the players will continue to roll dice without end.

The number of attacks made is not an infinite number, because you can only have an infinite number when there's a clearly defined set for you to compare to known infinities. However, the number of attacks made tends to infinity, and is not certain to finish in finite time.

Thespianus
2010-12-25, 10:35 AM
That is known as a Random Walk and is known to have probability 1 of ending.
I believe this is what I argued for above, and it's possible I didn't do my math right when I thought the non-"50%-50%" problem through in my head.

When I get time, I'll see if I can perform some deeper necromancy on my math skills. Until then, I concede defeat. ;)

Thespianus
2010-12-26, 02:45 PM
When I get time, I'll see if I can perform some deeper necromancy on my math skills. Until then, I concede defeat. ;)
It seems that I'm wrong and I present to you all Doob's Optional Stopping Theorem, and some relevant math:

http://en.wikipedia.org/wiki/Optional_stopping_theorem

http://en.wikipedia.org/wiki/Martingale_(probability_theory)

And in the Martingale article, I'd like to refer to the Amoeba example, which correlates nicely to the Lightning Mace abuse with Roundabout Kick:


Suppose each amoeba either splits into two amoebas, with probability p, or eventually dies, with probability 1 − p. Let Xn be the number of amoebas surviving in the nth generation (in particular Xn = 0 if the population has become extinct by that time). Let r be the probability of eventual extinction. (Finding r as function of p is an instructive exercise. Hint: The probability that the descendants of an amoeba eventually die out is equal to the probability that either of its immediate offspring dies out, given that the original amoeba has split.) Then

is a martingale with respect to { Xn: n = 1, 2, 3, ... }.

Substitute with p = 0.95 in this case.

AtomicKitKat
2011-01-02, 10:58 AM
So, more craziness. Assuming Aptitude works, Chakram Ricochet(Champions of Ruin), Boomerang Ricochet(Eberron CS), Roundabout Kick, should generate additional attacks, not just on the main target(360Kick), but also on secondary targets(the 2 Ricochets). Any others I'm missing?:smallcool:

Keld Denar
2011-01-02, 11:28 AM
Dunno...

I did find the code though! This is according to JaronK's build which just has LM and Roundabout Kick.



[totdmg,totswing]=function(numatt,avgdmg)
p=0;
while numatt > 0;
numatt = numatt - 1; % reduces # of attacks each iteration
p = p + 1;
roll = 1+floor(rand*20);
if roll > 8;
roll = 1+floor(rand*20);
if roll > 1;
swing(p) = 2; % Change 2 to 3 or 4 if using higher crit multipliers
numatt = numatt + 2; % Generate 2 attacks per crit
else
swing(p) = 1; % Noncrit
end
else
roll = 1+floor(rand*20);
if roll > 1;
swing(p) = 1;
else
swing(p) = 0;
end
end

% This creates an array of 0s, 1s, and 2s, which corresponds to misses, hits, and crits respectively.

totswing = length(swing);
% The length of the variable swing gives the total number of attack rolls made
% as the counter p increases at every iteration.

totdmg = sum(swing) * avgdmg;
% Summing swing will give us the total number of effective hits
% Hits count a 1 effective hit, crits count as 2 effective hits
% Multiplying avgdmg by the sum of hits will give total damage for the run


Anyone have MatLab and a decent computer that can run this script a few times?

drakir_nosslin
2011-01-02, 11:39 AM
Dunno...

I did find the code though! This is according to JaronK's build which just has LM and Roundabout Kick.



[totdmg,totswing]=function(numatt,avgdmg)
p=0;
while numatt > 0;
numatt = numatt - 1; % reduces # of attacks each iteration
p = p + 1;
roll = 1+floor(rand*20);
if roll > 8;
roll = 1+floor(rand*20);
if roll > 1;
swing(p) = 2; % Change 2 to 3 or 4 if using higher crit multipliers
numatt = numatt + 2; % Generate 2 attacks per crit
else
swing(p) = 1; % Noncrit
end
else
roll = 1+floor(rand*20);
if roll > 1;
swing(p) = 1;
else
swing(p) = 0;
end
end

% This creates an array of 0s, 1s, and 2s, which corresponds to misses, hits, and crits respectively.

totswing = length(swing);
% The length of the variable swing gives the total number of attack rolls made
% as the counter p increases at every iteration.

totdmg = sum(swing) * avgdmg;
% Summing swing will give us the total number of effective hits
% Hits count a 1 effective hit, crits count as 2 effective hits
% Multiplying avgdmg by the sum of hits will give total damage for the run


Anyone have MatLab and a decent computer that can run this script a few times?

I can do it once I get back home, but it'll be a few more days. If my computer can't handle it, I can always borrow a machine at the uni.

Keld Denar
2011-01-02, 11:54 AM
I haven't run it, so I haven't been able to debug it, but I did have one of my college proffessors help me with the syntax of a few of the commands I didn't know already (like randint). It SHOULD work. If it doesn't, run it in debug and lemme know what lines are broken.

term1nally s1ck
2011-01-02, 03:05 PM
Atomik: The build needs 2 specific classes to boost the crit range so much. You need a LOT of feats to qualify for both and still to take lightning maces and roundabout kick. I think I could only just do it at 20 with 1 feat spare.

Ernir
2011-01-02, 03:17 PM
I found it too. (http://www.giantitp.com/forums/showpost.php?p=10024417&postcount=17) :smallbiggrin:

You don't need Matlab (GNU octave does the job just fine in this case), or a decent computer (I'm running this on a 'puter that threatens me with death and destruction when opening too many browser tabs). What you do need is to change the function header to something like

function [totdmg,totswing]= KeldFunction(numatt,avgdmg)
and add an "end" to the third "if".

I'd also add a cap to the number of attacks you can actually simulate. Infinity is cool when you're a mathematician, less so when you have to wait for it to finish. I went for 100000.
This made the final code I ran...


%[totdmg,totswing]=function(numatt,avgdmg)
function [totdmg,totswing]= KeldFunction(numatt,avgdmg)
p=0;
while numatt > 0 && numatt < 100000;
numatt = numatt - 1; % reduces # of attacks each iteration
p = p + 1;
roll = 1+floor(rand*20);
if roll > 8;
roll = 1+floor(rand*20);
if roll > 1;
swing(p) = 2; % Change 2 to 3 or 4 if using higher crit multipliers
numatt = numatt + 2; % Generate 2 attacks per crit
else
swing(p) = 1; % Noncrit
end
else
roll = 1+floor(rand*20);
if roll > 1;
swing(p) = 1;
else
swing(p) = 0;
end
end
end

% This creates an array of 0s, 1s, and 2s, which corresponds to misses, hits, and crits respectively.

totswing = length(swing);
% The length of the variable swing gives the total number of attack rolls made
% as the counter p increases at every iteration.

totdmg = sum(swing) * avgdmg;
% Summing swing will give us the total number of effective hits
% Hits count a 1 effective hit, crits count as 2 effective hits
% Multiplying avgdmg by the sum of hits will give total damage for the run


And some sample runs...

octave:4> [damage, attacks] = KeldFunction(7,10)
damage = 11222960
attacks = 724645
octave:5> [damage, attacks] = KeldFunction(3,10)
damage = 320
attacks = 23
octave:6> [damage, attacks] = KeldFunction(5,10)
damage = 10899570
attacks = 702557
octave:7> [damage, attacks] = KeldFunction(2,10)
damage = 50
attacks = 4
octave:8> [damage, attacks] = KeldFunction(2,10)
damage = 200
attacks = 14
octave:9> [damage, attacks] = KeldFunction(2,10)
damage = 530
attacks = 36
octave:10> [damage, attacks] = KeldFunction(2,10)
damage = 10994720
attacks = 709056


What were we trying to show, again?

Thespianus
2011-01-02, 03:26 PM
What were we trying to show, again?
That the number of attacks heads off towards inifinity (ie , they reach your "max number of attack"-limit) for almost all of the runs.

From what I can see from your output, it doesn't seem like they do.

EDIT: The code above does not handle the Blood in the Water-effect: The code compares the roll with "8" constantly. That value should actually go down with eac accumulated critical hit, if I'm not mistaken

JaronK
2011-01-02, 03:46 PM
I remember when I first put together this build and we were trying to calculate if it went to infinity (it was for a TO attempt to get the most number of attacks in a round without having an infinite loop). Standard statistical models don't work very well here, so we tried to run a simulation... but back then, nobody actually did it.

In the end, because of blood in the water you'll be confirming on a 2 after the first 20+ attacks. The code has to have that effect built in or it won't be accurate. The real question is exactly what crit range you need for it to never go off to infinity, and yet still generate the maximum number of attacks.

JaronK

Ernir
2011-01-02, 04:09 PM
EDIT: The code above does not handle the Blood in the Water-effect: The code compares the roll with "8" constantly. That value should actually go down with eac accumulated critical hit, if I'm not mistaken

Eh, looks to me like it's assuming auto-confirmations (excepting nat 1's) on crit threats anyway. That's what I am reading out of it, but I could be picking Keld's brain wrong.

What is it that you would like to see? Adjusting the code to include ACs, to-hit rolls, confirmation rolls, and whatever else you can think of should be rather easy.

eggynack
2011-01-02, 04:28 PM
octave:4> [damage, attacks] = KeldFunction(7,10)
damage = 11222960
attacks = 724645
octave:5> [damage, attacks] = KeldFunction(3,10)
damage = 320
attacks = 23
octave:6> [damage, attacks] = KeldFunction(5,10)
damage = 10899570
attacks = 702557
octave:7> [damage, attacks] = KeldFunction(2,10)
damage = 50
attacks = 4
octave:8> [damage, attacks] = KeldFunction(2,10)
damage = 200
attacks = 14
octave:9> [damage, attacks] = KeldFunction(2,10)
damage = 530
attacks = 36
octave:10> [damage, attacks] = KeldFunction(2,10)
damage = 10994720
attacks = 709056
[/code]


That makes some sense, because the series would have to end early if it was going to end at all. My only question is why it always ends right at the 700000 range every time.

term1nally s1ck
2011-01-02, 04:31 PM
The constant ending at the 700000 range is suspicious...the probability of it ending after that many attacks is minimal....possibly a flaw in the random number generator.

Thespianus
2011-01-02, 04:33 PM
Since I couldn't spare the 5 minutes needed to learn Octave, I used C# to recreate the algorithm. I hope nobody minds:


using System;
using System.Text;

namespace ConsoleApplication1
{
class Program
{
static Random random = new Random(DateTime.UtcNow.Millisecond);
static int D20() { return 1 + random.Next(20); }

static void Main(string[] args)
{
int numRounds = 1000;
if (args.Length > 0)
numRounds = Int32.Parse(args[0]);
int infiniteCount = 0;
int finiteCount = 0;
for (int i = 0; i < numRounds; i++)
{
if (PerformFullAttack(1))
infiniteCount++;
else
finiteCount++;
Console.WriteLine("Infinite: " + infiniteCount + " Finite: " + finiteCount);
}
Console.WriteLine("Runs: " + numRounds);
Console.WriteLine("Infinite: " + infiniteCount + " Finite: " + finiteCount);
Console.ReadLine();
}

static bool PerformFullAttack(int startingAttacks)
{
int numAttacks = startingAttacks;
int maxNumAttacks = 100000;
int rollNeededToHit = 8;
int rollNeededToCrit = 8;
int performedAttacks = 0;

while (numAttacks > 0 && performedAttacks < maxNumAttacks)
{
numAttacks--;
performedAttacks++;
int roll = D20();
if (roll >= rollNeededToCrit)
{ // we threaten a critical. Gain one attack from Lightning Maces
numAttacks++;
if (D20() >= rollNeededToHit)
{ // we have scored a critical hit. Gain one attack more from Roundabout Kick
rollNeededToHit--; // Reduce the roll needed to hit due to Blood in the Water
if (rollNeededToHit < 0)
rollNeededToHit = 0;
numAttacks++;
}
}
}
return performedAttacks == maxNumAttacks;
}
}
}



EDIT: Code has been updated to correct for my lack of understanding of Lightning Maces awarding an attack on a roll for a critical threat.

term1nally s1ck
2011-01-02, 04:43 PM
You should still get some. For 8-20, I think it's about a 13/40 chance.

Thespianus
2011-01-02, 04:54 PM
You should still get some. For 8-20, I think it's about a 13/40 chance.

Are you sure? Without the blood in the water-effect, you need to roll 8 or better twice (first to get a critical hit and secondly to confirm the hit) to get two extra attacks.

That's a 12/20 * 12/20 chance, or a measly 0.36 chance to get 2 new attacks. Maybe I'm missing something, but the only way to gain new attacks is to perfrom a successful critical hit, right? Or am I missing something?

You need to go above a 0.50 chance to gain 2 new attacks, otherwise it will stop. Blood In The Water gets you above 0.50 after a few sucessful attacks.

EDIT: My bad. I misunderstood Lightning Maces

Keld Denar
2011-01-02, 04:55 PM
I'm not smart enough to code in blood in the water. I just assumed anything 2+ would confirm. The damage part was added in after, the main goal of the function was to sum number of swings proced. As someone mentioned, computer RNGs tend to even out more than a real unbiased simulation, so you are less likely to get strings of ones. Otherwise, I'm relatively happy that the code works decently...I'm a mechanical engineer dabbling in a programmers world...

Greenish
2011-01-02, 04:55 PM
Are you sure? Without the blood in the water-effect, you need to roll 8 or better twice (first to get a critical hit and secondly to confirm the hit) to get two extra attacks.

That's a 12/20 * 12/20 chance, or a measly 0.36 chance to get 2 new attacks. Maybe I'm missing something, but the only way to gain new attacks is to perfrom a successful critical hit, right? Or am I missing something?Lightning Maces only requires a critical threat to trigger, not a confirmed critical.

term1nally s1ck
2011-01-02, 05:17 PM
Are you sure? Without the blood in the water-effect, you need to roll 8 or better twice (first to get a critical hit and secondly to confirm the hit) to get two extra attacks.

That's a 12/20 * 12/20 chance, or a measly 0.36 chance to get 2 new attacks. Maybe I'm missing something, but the only way to gain new attacks is to perfrom a successful critical hit, right? Or am I missing something?

You need to go above a 0.50 chance to gain 2 new attacks, otherwise it will stop. Blood In The Water gets you above 0.50 after a few sucessful attacks.

it's a 13/20, actually.

And that's a 140/400 chance of going -1, a 169/400 chance of going +1, and the other 91/400 is unconfirmed crits, which generate 1 attack and break even.

Let the probability of an attack string ending, starting with one attack, be equal to P.

P = 140/400 + 169/400 * p * p + 91/400 * p

Using the quadratic formula, I get a 140/169 probability, which means that there's a 29/169 chance of it going to infinity.

(If you actually mean >8, so a 9-20, 12/20 is not enough a chance, barely.)

Thespianus
2011-01-02, 05:18 PM
Lightning Maces only requires a critical threat to trigger, not a confirmed critical.
That would explain it. My bad.

Yeah, with this modification, the numbers look quite different. ;)

Edit: Ok, this is turning into a nerd-fest, but.. Just to be sure I'm thinking correctly now:

A critical threat results in 1 extra attack from Lightning Maces
A confirmed critical results in 1 extra attack from Roundabout Kick?

I'll update the code... ;)

cd4
2011-01-02, 05:24 PM
Roundabout Kick which is also used does require a critical confirmed to give you an extra attack. So you have the situation:

Attack Roll:

1-7 miss and stop (might hit with Blood in the Water but will still stop the progression)
8-20 - hit and threaten

Confirm Roll
1-7 not critical and only have one attack (Reduced by Blood in the Water so it is almost certain of getting 2 attacks)
8-20 critical and gain 2 attacks.

Oh and just for the two main parts of this build the feats required are:

Improved Unarmed Strike
Power Attack
Roundabout Kick
Combat Reflexes
Two Weapon Fighting
Weapon Foucs (Light Mace)
Lightning Mace

term1nally s1ck
2011-01-02, 05:32 PM
Disciple of Dispater has another 2-3 feat pre-reqs, as does the exotic weapon master variant that does work.

Thespianus
2011-01-02, 05:53 PM
I've updated the code, if someone is nerdy enough to go back and check it out.

The relevant algorithm in the code is the PerformFullAttack - method, that takes one argument: the number of initial attacks ( 1 for a standard attack, more for Full attacks)

With a standard attack (1 initial attack ) I get infinite results in about 1/3rd of the simulated attacks. This is with a 8-20 critical range and an initial requirement of needing to roll 8 or better to actually hit. Without Blood in the Water, I get about 1/6 infinites.

With more initial attacks (simulating a Full Attack round), the numbers go way up. With 4 attacks, I get over 4/5 infinites, and with 8 attacks (Is TWF possible?) 97% of all attacks go to infinity.

AtomicKitKat
2011-01-03, 12:47 PM
I remember when I first put together this build and we were trying to calculate if it went to infinity (it was for a TO attempt to get the most number of attacks in a round without having an infinite loop). Standard statistical models don't work very well here, so we tried to run a simulation... but back then, nobody actually did it.

In the end, because of blood in the water you'll be confirming on a 2 after the first 20+ attacks. The code has to have that effect built in or it won't be accurate. The real question is exactly what crit range you need for it to never go off to infinity, and yet still generate the maximum number of attacks.

JaronK

For the purposes of the TO ruling, it should be considered "non-infinite", if only because it can end. Just that the odds are "infinitesimally" small, and shrinking the further along you go.:smallwink:

JaronK
2011-01-03, 03:34 PM
For the purposes of the TO ruling, it should be considered "non-infinite", if only because it can end. Just that the odds are "infinitesimally" small, and shrinking the further along you go.:smallwink:

I think I'd go for "median behavior terminates" as a guideline.

JaronK

AtomicKitKat
2011-01-03, 09:57 PM
I think I'd go for "median behavior terminates" as a guideline.

JaronK

No idea what that means. :/

Greenish
2011-01-03, 10:30 PM
No idea what that means. :/If you order all the results (ie. numbers of attacks) in a line by size, and then pick the one that's in the middle, you've got the median.

He's saying that for his purposes, this number being less than infinite should work.

term1nally s1ck
2011-01-03, 11:38 PM
Probability of reaching infinity <1/2.

However, that does mean that the mean value would actually 'be' infinity, and taking the median value would result in quite a small number as the actual submission.

Gralamin
2011-01-03, 11:55 PM
The real question is exactly what crit range you need for it to never go off to infinity, and yet still generate the maximum number of attacks.

To never go off to infinity, you will need to be unable to crit. I haven't done the math (Self-referral Random Walks, which Blood in the Water would cause, are a pain. Specifically it would be a one-way Reinforced Random Walk with a maximum weight Edit: Whoops, that would require increasing your chance to crit as you go. Ignore that.), but to get a statistically small chance with Blood in the Water, I imagine you would need a Critical Range of 20 (and even then, it might still be a very large chance). Drop Blood in the Water and you can probably get away with a bit more - I'd have to remember the equations for Random Walks to say for sure.

term1nally s1ck
2011-01-04, 12:04 AM
Crit range of 9-20 and hitting on 9s would never hit infinity. Probability 1 of ending in finite time.

However, since it's got no cap, it has a positive probability of reaching any finite number, so it can beat any other record out there....once in a blue moon.

As for hitting on 2s, you need a crit range of 11-20 or smaller to stay finite then.

Gralamin
2011-01-04, 12:12 AM
Crit range of 9-20 and hitting on 9s would never hit infinity. Probability 1 of ending in finite time.

However, since it's got no cap, it has a positive probability of reaching any finite number, so it can beat any other record out there....once in a blue moon.

As for hitting on 2s, you need a crit range of 11-20 or smaller to stay finite then.

Mind PMing or posting the math for that? I must be misremembering how Random walks work.

Tvtyrant
2011-01-04, 12:24 AM
...This thread wins so hard.

term1nally s1ck
2011-01-04, 02:26 AM
Mind PMing or posting the math for that? I must be misremembering how Random walks work.

Check the expected number of attacks after making the first one. With a 12/20 to crit and 12/20 to confirm (without blood in the water), you average less attacks than you had to start with. The random walk is biased towards 0 for that probability. (10/25 to go -1, 9/25 to go +1. Expected value is 24/25 attacks.)

If the probability of the walk increasing is equal or less than the probability of it decreasing, it hits 0 with probability 1.

EDIT: TV, this happens in every single lightning maces thread. Every. Single. One.

Gralamin
2011-01-04, 02:47 AM
Check the expected number of attacks after making the first one. With a 12/20 to crit and 12/20 to confirm (without blood in the water), you average less attacks than you had to start with. The random walk is biased towards 0 for that probability. (10/25 to go -1, 9/25 to go +1. Expected value is 24/25 attacks.)

If the probability of the walk increasing is equal or less than the probability of it decreasing, it hits 0 with probability 1..

Ah right, that makes sense. I really should review probability...

Wait a second, wheres the other 6/25? Is that just the chance of not changing? (IE: critting but not confirming, giving you an additional attack).

Let's see:
8/20 chance to not threat: -1.
12/20 chance to threat, then a 12/20 chance to confirm for +1, or a 8/20 chance to not confirm for only a single attack.

P(Threat, Confirm) = P(Threat) * P(Confirm|Threat) = (12/20)^2 = 9/25. 9/25 for +1

8/20 = X/25 -> X = 10. 10/25 for -1.

P(Threat, NotConfirm) = P(Threat) * P(NotConfirm|Threat) = 12/20 * 8/20 = 6/25

So yeah, those numbers work out... But this isn't a true random walk is it? Generally you cannot "stay" the same value in a random walk.

In fact, wouldn't splitting crossbows make it a true random walk, since then we either get +1 or -1? Then the optimal amount of attacks would be at the 50% mark, or an 11-20 crit range.

term1nally s1ck
2011-01-04, 02:57 AM
It can be treated as one when you're checking to see if it 'goes to infinity' or not, but it doesn't quite follow the rules, no.

Of course, that just changes the total attacks made slightly.

Gralamin
2011-01-04, 03:03 AM
It can be treated as one when you're checking to see if it 'goes to infinity' or not, but it doesn't quite follow the rules, no.

Of course, that just changes the total attacks made slightly.

Yeah, I expected it wouldn't change the infinite case. It's good to realize though.