Without even trying to dig into the math, that analysis is fundamentally flawed because you're not looking to exhaust the search space. There is not only one solution, nor do you have to find all the solutions. You only need one.
One out of X, where X is a ridiculously large search space. The computational complexity of the direct brute force approach is O(n!); that is precisely what my post demonstrated.
thejeff wrote:
Others have posted fairly simple and elegant approaches to finding those solutions. It's not really difficult, but will definitely slow down play.
Links? The only "solution" I've come across fairly consistently produces incorrect results.
EDIT: n/m, found a way to reduce the search space significantly.
You shouldn't underestimate the importance of carrying capacity. With a dexterity build, try pushing giant chunks of rock and stone from a collapsed ceiling, each weighing several tons, off of you and then digging the rest of your party out of the rubble after your GM has the bad guy (who can earthglide to safety) cast Earthquake while you're underground. If my character at the time hadn't been strength-based, it would've been a TPK.
According to the spell descriptors, charms and compulsions are neither ethically nor morally aligned. Based on how the PF alignment system is supposed to work, compulsions really ought to carry the lawful descriptor, because they override the subject's free will. Charms don't directly override the subject's free will, they skew its perceptions, which in turn skews how it exercises its free will. Good and evil, as defined by the PF alignment system, don't enter into it.
As for trying to apply real world moral or ethical judgments to the game world, it's generally a bad idea. In the game world, the "heroes" are constantly killing other intelligent beings, without any semblance of remorse, nor any hint of post traumatic stress afterwards; PCs are psychopaths.
Unless there are some clever optimizations possible to significantly reduce its complexity class (I suspect not), the calculations required to use this feat have a computational complexity of O(n!). That is, the size of the search space of the problem grows with the factorial of the number of dice you need to roll, which must equal your ranks in Knowledge (engineering).
Analysis:
The feat calls for n number of dice to be rolled (where n is your ranks in KE), the results then being combined in some way using only the four basic binary arithmetic operators, '+', '-', '*', and '/', and the result must match one of three prime numbers. Any such arithmetic expression can be modeled by a binary expression tree, which is a full binary tree wherein the internal nodes are the operators and the leaves are the operands (i.e. the numbers which came up on the dice). Unfortunately, I can't really illustrate it here because this messageboard apparently lacks anything like a 'pre' or 'code' bbcode tag, and it's probably confusing without illustration. Wikipedia's page on the subject fortunately includes diagrams which should make it clear: binary expression tree
To solve the problem presented by the feat we need to find one such binary expression tree, which evaluates to one of the three prime constants, out of all the possible structurally distinct binary expression trees which could be constructed using the four binary arithmetic operators available to us (for the internal nodes) and the numbers which came up on the dice (for the leaf nodes).
How many such trees are there?
Let SDT(n) be the number of structurally distinct full binary trees with n leaves; note this is independent of assigning any particular operators or dice results to the nodes of any tree (yet).
To find SDT(n), this page has a nice easy-to-use lookup table for, and a good explanation of how to find, the number of structurally distinct binary trees, however it includes all structurally distinct binary trees (not just the full ones) and works by a count of all the nodes (vertices), not just the number of leaves. Fortunately it's easy to convert to what we need here, since the internal nodes for all of our full trees are themselves binary trees with n-1 nodes. Therefore, SDT(n) can be found simply by finding the row in the table on that page for n-1 number of vertices, and then looking at its rightmost column. For example, if n=10, according to that table SDT(10)=98 structurally distinct trees. SDT(n) grows exponentially with n, so we're already at O(c^n)|c>1 (exponential complexity).
Let m be the number of operator nodes in any of our binary expression trees; this must always be n-1, so m=n-1.
The number of possible arrangements of operators in any given tree structure must be 4^m, since any of the 4 operators could be used at any particular operator node. This means we're up to SDT(n) * 4^m for our search space (still exponential complexity), but we haven't populated the leaves with dice results yet.
For each of those SDT(n) * 4^m trees, we have to try populating the leaves with every permutation of our n dice results, so that brings us up to SDT(n) * 4^m * n! trees in our search space.
This brings us up to O(n!), factorial complexity.
What does this mean in terms of the actual numbers:
With 5 ranks in Knowledge (engineering), SDT(5)=3, 4^4=256, 5!=120, so our search space is 3*256*120=92160 different configurations.
With 10 ranks in Knowledge (engineering), SDT(10)=98, 4^9=262144, 10!=3628800, so our search space is 98*262144*3628800, which is greater than 93 trillion different configurations.
With 15 ranks in Knowledge (engineering), SDT(15)=4850, 4^14=268435456, and 15! is over 1.3 trillion, so that works out to over 1.7 septillion (i.e. 1.7*10^24) different configurations in our search space.
The feat is fundamentally broken, but not because it's overpowered.
I'm currently the GM (we rotate) for my group, and right now we're going through Sins of the Saviors (AE).
The major changes thus far have been:
Ravenous Crypts of Gluttony:
The PCs were welcomed to Runeforge by an ancient advanced fiendish vampiric sinspawn of gluttony, on PCP. While joking about all the nasty things I was going to throw at them once they entered Runeforge, a few days prior to the session wherein they would first go there (they'd just opened the portal at the end of the previous session), one of my players suggested it, so I actually stat'd it up (including the PCP, which granted immunity to pain [such as from thirst for blood] for a couple of hours, but had some ugly side effects for those not immune to Fort saves) and had it waiting for them near the entrance. The party's got a full paladin as well as an oradin, so it was forced into gaseous form pretty quickly, which then led them into the gluttony wing. A couple of sessions (and trips to the material plane and back) later they made it to area F6, the alchemical research room, which (unlike the published version) was still very much in use, as an alchemy (primarily PCP) lab. I added a coffin in a little alcove in the back wall of F6, which turned out to be a decoy filled with a vampiric fleshdreg swarm, and it had a couple of small pipes leading into the wall (presumably so the vampire could use them to move in gaseous form to/from its real coffin, which I placed in the floor of F9, behind the secret door from F8). The vampiric sinspawn had a telepathic bond to Azaven just like Xyoddin and the mummies, so I had it try to provide back up for Xyoddin when the PCs found him in his lab in F7 (though naturally as a lone NPC, Xyoddin didn't last long enough for it to matter). The players seemed to enjoy the addition.
For the Xyoddin fight, aside from the vampiric sinspawn as backup, I added a free action on command supernatural darkness effect to his lab, because it was his only chance to even last more than one round (he still only lasted two, because paladins). It's also a no-brainer defensive measure I would expect a bunch of super-genius magic users with 10k+ years of prep-time to have managed to implement.
For the Azaven fight, I added a mechanical alarm trap to the secret door from F9 to F10 (which easily could've been put in place by Xyoddin for him, and is another no-brainer defensive measure). The party used hit-and-run tactics for the gluttony wing, and this fight was no different, so Azaven needed the alarm to avoid being caught completely by surprise; thanks to the warning, he had time to throw up a wall of force and buff up before facing the party. I also had the necromantic death trap fire off every round instead of every other round, and Azaven dimension doored himself to F11 and activated it as soon as he'd taken damage from the paladins, instead of waiting to get smited to death. The devourer was of course smited into oblivion in a single round, and the lone lich had no hope of winning against a pair of paladins, but he did manage to survive long enough to keep it interesting.
Vault of Greed:
In the greed wing, I added a 20-ft range mass flesh to stone trap on each of the doors from G4 to G6. This saved Ordikon's life, for a day, by forcing the party to retreat after nearly killing him and forcing him to dimension door out of one of the rear G4 rooms (where he and two of the stone golems faced off against the party). That was the session before last.
For this week's session, the party was going to kill or capture Ordikon, and they did manage to kill him in the end, but I didn't make it easy. The wizard used one of his new spells copied from Azaven's spellbooks which allowed him to turn incorporeal to scout the greed wing beyond G6. What he found were permanent walls of force sandwiched between the stone of the walls separating G6 from G7; the only part not covered by the walls of force was the actual door to G7. The player already expected it, since we'd discussed the spell he used and potential defenses against it a few weeks earlier, but it made sense for his character to try it anyway. Most of the other wings of my version of Runeforge have similar wall of force protections erected around key areas.
After moving through the door into G7, the party's wizard found a very different room from the published version. I expanded it to be 60-ft wide, 60-ft high, and 120-ft long, with the gilded "statues" placed in five 10-ft square alcoves on each side of the room starting from the entrance. The published description of the room was kept largely intact, but I added several streams of various molten metals flowing from holes in the ceiling into drains in the floor directly beneath them; essentially they were columns of molten metal (high value metals, like gold, platinum, and mithral; which will all of course revert to lead and other cheap metals if ever brought outside the greed wing). The alcoves with the gilded "statues" each had similar molten metal column ports above them, with drainage set up at the base, and in the first three alcoves on each side of the room the ports were closed, but on the last two on each side the statues had been moved outside their alcoves to the center of the room (positioned looking in the direction of the door from G6), and the column ports were open and molten gold flowed from them. The gilded "statues" also all had greed runes on their foreheads made of mithral for the claw/hand-like part of the rune and a precious gem for the gem part of the rune; aside from the added flavor, this also had a function, as the current leader of the wing (Ordikon at this point) could use the gems as additional eyes, essentially giving my BBEG a good view of the room, even while he hid in the ceiling in earth elemental form.
Instead of only one bound nalfeshnee eager to get chatty with the party and beg for their help, I placed two in G7, bound to defend G7 and G8 until the time of Karzoug's return. The path to G8, where Ordikon (who'd been scrying on the party the whole session) was busy buffing up while the party wizard was on his incorporeal scouting run, was moved to the back of G7. The nalfeshnees, with constant true seeing and +31 perception easily noticed the scouting wizard, so he retreated back to the rest of the party, and then they buffed up before trying to dimension door into G7. While the party was busy buffing, the nalfeshnees summoned two vrocks each, and Ordikon finished buffing and moved into the ceiling of G7 in earth elemental form. The nalfeshnees each moved into position, each flanked by their two vrocks, flying 50-ft up in front of the alcoves with molten gold columns, so the vrocks could each use telekinesis to throw 300lb blobs of that stuff at the party each round. I couldn't find any existing rules for flinging molten metal at someone with telekinesis, so for the effects I used wall of lava as a guide; each 300lb blob carried with it 12d6 fire damage, plus half damage on the following round, and hit on a ranged touch attack. The nalfeshnees of course were to just continuously cast feeblemind at the party wizard until it stuck, and use their free action unholy nimbus ability whenever party members came in range. The nalfeshnees also had golden greed runes embedded in their foreheads (similar to Teal'c from Stargate SG-1), which worked similarly to the greed runes on the gilded statues (i.e. Ordikon could see everything in their field of vision, as needed), but also allowed them to reliably summon the two vrocks each I needed, without the percentile roll or the d4 roll.
When the party tried to dimension door into G7, the party wizard found he had to make two DC30 will saves to get past the permanent persistent teleport trap that I had covering all of G7 and G8 (because super-geniuses with magic would've thought to do that). Fortunately for the party, he made both saves; had he not, they would've all been redirected to the exterior surface of runeforge, at which point a reverse gravity would've kicked in to nudge them away from runeforge and into the endless void. Towards the end of the encounter, the summoner in the party forgot about the teleport trap and tried to DD himself and his eidolon over to Ordikon (a move he's used before to great effect, and the very reason I'd added the teleport trap in the first place), and he failed to make the save, so he and his eidolon spent the next few rounds trying to get back to G7 and missed the tail end of the fight.
Ordikon wore a modified sihedron medallion which had a gold and mithral greed rune laid over the sihedron on its face, and marked him as the current leader of the greed wing (allowing him to use the greed runes on the nalfeshnees and "statues" as extra eyes, and to bypass the teleport trap). It also prevented his capture, causing his body to start to calcify once he hit negative HP (starting with the spot on his chest where the medallion hung and radiating outward, thereby irreparably damaging his heart almost immediately upon activation); the party removed it before it could calcify his head, so they can still use speak with dead on him, but he won't be very cooperative and their options are far more limited with that than if he were still alive. He also had a few quicken metamagic gems, because his action economy was terrible otherwise, and a rod would've been too valuable to permit it as loot from the encounter.
Ordikon's tactics were to stick to earth elemental form and to pop out from the ceiling or walls just enough to cast a greater dispel magic at any melee types trying to fly up to the nalfeshnees (which cost one of the two party paladins a potion of fly, and managed to deter any further attempts at flying up to melee them, though the gargantuan ape eidolon had reach), and to try to flesh to stone the wizard (if not already feebleminded), followed by the summoner, and then the paladins, once air dominance had been achieved.
By the end, the summoner and his eidolon were floating in space outside runeforge thanks to the teleport trap (though not before the eidolon had killed one nalfeshnee, and grappled the other to bring it in range of paladin smites, from which it also promptly died), one paladin was feebleminded (due to a paladin sacrifice for the wizard), the other paladin was turned to stone (also due to a paladin sacrifice for the wizard), and it was pretty much a wizard vs wizard fight, which ended with Ordikon being black tentacled to death. The entire session was full of suspense, with the party members all feeling like they were about to be TPK'd right up until they won, which is exactly what I wanted.
All the players seemed to really enjoy the session, and I had a lot of fun with it too.
The envy wing was played as written, and the party hasn't gotten to the sloth, lust, pride, or wrath wings yet, though I'm currently working on some changes for those. Hopefully they'll be as successful as what I did with greed.
Now the only way to heal your Eidolon is by casting a spell. So is there a way to regain those temp HP EVER?
Lose all your temporary HP. The eidolon will go back to whatever plane it's from, and then you can resummon it. It'll come back with half its maximum HP. You'll have to settle for half or less of your maximum temporary HP so long as you don't have arms in fused form and don't have some other way to have one of the Rejuvenate Eidolon spells cast on your fused form. If you really want to get back to your maximum temporary HP but don't want arms in fused form, and you don't have someone else in your party who can cast one of the Rejuvenate Eidolon spells on you, Eldritch Heritage with the Arcane bloodline could get you a familiar, and some familiars can use wands. Leadership with a high-UMD cohort is another option. Still Spell, in either rod or feat form, is yet another option for you.
I will have to disagree with the above poster on a couple things. Advanced Firearms use Metal Cartridges Only as ammunition. Metal Cartridges are alchemical cartridges. Therefore I strongly feel that this is already included in their action to load.
Otherwise an advanced firearm would ALWAYS be a free action and never a move action to load.
Of course this is up to your GM to interpret, but I feel confident this is how it is meant to be played.
Correct.
Ishyna wrote:
Secondly, by RAW, Rapid Reload has no affect on Advanced Firearms. It does not reduce the step to reload a firearm.
Just looking at the Rapid Reload feat alone, it would actually appear to increase the time it takes to reload two-handed advanced firearms from a move action to a standard action, which makes the feat's benefit text self-contradictory.
The Loading Firearms section of UC states that the feat reduces the time required to "load one-handed and two-handed firearms", while the feat's benefit text begins by stating that it reduces the time required to "reload your chosen type of weapon". In both cases, the reloading time is said to be reduced, not kept the same or increased.
The remainder of the feat's benefit text can be interpreted in one of two ways: either it contradicts both the "Loading Firearms" section and the beginning of its own benefit text, or the weapons listed in parentheses must be viewed only as examples of weapons which normally have greater loading times (i.e. one-handed early firearms are normally a standard action to load, hence the presence of "one-handed firearm" in parentheses next to "move action"). The former interpretation would mean that you are correct, while the latter would mean that Starbuck_II is correct; problem is, neither interpretation is technically RAW.
There actually is no way to determine the RAW in this case, because even if it were clear which text qualifies as the more "specific" text (and it most certainly is not clear), neither reading can just be taken at face value. No matter how the feat's benefit text or the "Loading Firearms" section of UC are read, it's obvious there's an error somewhere, because the very existence of advanced firearms renders the feat's benefit text nonsensical.
Unfortunately, despite the frequency with which this question comes up, and despite the logical incoherence and ambiguity being glaringly obvious, Paizo has yet to faq/errata this. Until they do, there is no RAW for how the feat works with advanced firearms, if it does at all, there's only guessing about RAI.
I've got a whole bunch of fixes (in .MOD form) for the v6.00.0 pathfinder data sets sitting in my homebrew directory, and maybe I'm just getting old/going blind, but I'm not seeing anything like a "Contribute" link on the pcgen homepage. What's the best way to submit something like that? Should I just send the .lst files (they're all organized as fixed__<source lst>.lst, e.g. fixed__pfcr_feats.lst) as attachments to one of the mailing lists (pcgen_experimental, perhaps?), or send them in a bug report through JIRA, or is there some other method I should use to submit them? Or will fixes only be accepted in patch form (which one? diff -u?), and if so, what's the usual turnaround time for upstream acceptance of data changes? (If I'm going to need to maintain a local fork for awhile, I'd like to know ahead of time so I can get git set up to do most of the branching/merging work for me.)
Also, what's the process for becoming a regular contributor? I've gotten pretty good at reading/modifying LST files in the few months since I started using PCGen, and I'm also a software developer.
