Talk:Experience

The erratic formula

Erratic (600000) E = -1/50*l^4 + 2*l^3 for level<=50

http://www.math.miami.edu/~jam/azure/forum/tuff/ultimatebb.php?ubb=get_topic;f=1;t=001260

It's mentioned in the thread that level 45 does not work for that. Level 45 is 100273, not 100237. Sheep 16:42, 21 Feb 2005 (UTC)

Nevermind the previous message, I figured it out myself. Sheep 18:09, 21 Feb 2005 (UTC)

Simplify the formulas

Was that copied from matlab or something? It can be simplified...

((0.814 - 0.02*(((n - 69) / 3) - (((n - 69) / 3) modulo 1)) - ep(((n - 69) modulo 3))(n^3)

((0.814 - 0.02*(((n - 69) / 3) - float_part((n - 69) / 3) ) - ep(((n - 69) modulo 3))(n^3)
((0.814 - 0.02*( int((n - 69) / 3) ) - ep(((n - 69) modulo 3))(n^3)
((0.814 - 0.02*( int(n/3 - 23) ) - ep(((n - 69) modulo 3))(n^3)
((0.814 + 0.46 - 0.02*( int(n/3) ) - ep(((n - 69) modulo 3))(n^3)

int(  (  1.274 - 0.02*(int(n/3)) - ep(n%3)  ) * (n^3)  )
That would be easier to read.

2*(1 - 0.01n)*(n3)
= int( 2n3 - 0.02n4 )

(  1 - ((n - 50)*0.01)  ) * (n^3)
= int( 1.5n3 - 0.01n4 )

(1.6 - 0.01n) * (n3)
= int( 1.6n3 - 0.01n4 )

Qgpr03:05, 01 Mar 2005

I really want to avoid using expressions that are not standard math notation. I don't even know if there's a way to represent some integer rounding using standard math notation, just some pseudo-code, that's why I stuck mod 1 subtraction in there. My plan was to simplify/clean it up once TeX was implemented on bulbapedia.

In the meanwhile you can simplify them. Just keep "mod" or "modulo" instead of "%" and avoid "int()." Sheep 20:20, 1 Mar 2005 (UTC)

The big problem is that when I first read that I got confused. No person that would bother to understand that formula wouldn't know how to program anyways, to represent that rounding you use ||x|| I think, maximum integer, unless I got the symbol wrong, long time I don't touch math. I'll leave that formula but also add the simplification for anyone that wants someting simplier to read. I will also simplify your standard math formulas, but won't add the ||x|| because I am not sure if its the right symbol, those - 69 can be taken out of modulo since they will be equal to zero. Qgpr 20:03, 5 Mar 2005 (GMT-5)

|x| is modulus - gives the absolute value of function, i.e. |-0.5| = 0.5. Modulo is simply represented as mod in mathematics. Anyway, while we do not have TeX at the moment, you can try this instead - leave the original math formula as a <!-- COMMENT -->, and copy the image from Wikipedia (it will generate the image on preview, so you can download that and reupload it here. Don't overdo it of course.)

||x|| ( double | ) is what I was taught here at college, but I guess is not an standard, however I found what it seems is the standard, and used the UTF code to show it. However <*pre> is not happy with <*sub>, you should find a way around because mine doesn't look that good. By the way it was "greatest integer" not maximum, problems for using a direct translation. Qgpr 22:16, 5 Mar 2005 (GMT-5)

First shot at a TeX markup image. Using

\begin{math}
b \star x \rightarrow \frac{b}{x} - (\frac{b}{x}\;mod\;1)
\end{math}

\addvspace{.2cm}

\begin{math}
e(n) = \left\{
\begin{array}{clrr}
n^3(\frac{100 - n}{50}); & 0 < n \leq 50 \\*
n^3(\frac{150 - n}{100}); & 51 \leq n \leq 68 \\*
n^3(1.274 - \frac{1}{50}(n\star3) - p(n\;mod\;3)); & 69 \leq n \leq 98 \\*
n^3(\frac{160 - n}{100}); & 99 \leq n \leq 100 \\*
\end{array}
\right\}
\end{math}

\addvspace{.2cm}

\begin{math}
p(x) = \left\{
\begin{array}{clrr}
0.000; & x = 0 \\*
0.008; & x = 1 \\*
0.014; & x = 2 \\*
\end{array}
\right\}
\end{math}

I came up with

Kind of cramped. Does it meet the approval of you two? Looking for revisions before I even post it. Especially since I forgot how to properly represent a custom operator (\star). Is that how? Sheep 03:32, 6 Mar 2005 (UTC)

This seems better:

\begin{equation}
E(n) = \left\{ \begin{array}{ll}
\frac{n^{3} \left(100 - n\right)}{50}, & \textrm{if $0 < n \leq 50$} \\
\frac{n^{3} \left(150 - n\right)}{100}, & \textrm{if $51 < n \leq 68$} \\
n^{3} \left(1.274 - \frac{1}{50}\left\lfloor\frac{n}{3}\right\rfloor -p\left(n \bmod 3\right)\right), & \textrm{if $69 < n \leq 98$} \\ 
\frac{n^{3} \left(160 - n\right)}{100}, & \textrm{if $99 < n \leq 100$}
\end{array}\right.
\end{equation}

\begin{equation}
p(m) = \left\{ \begin{array}{ll}
0.000, & \textrm{if $m = 0$} \\
0.008, & \textrm{if $m = 1$} \\
0.014, & \textrm{if $m = 2$}
\end{array}\right.
\end{equation}

File:Erratic experience levels equation 1.png
File:Erratic experience levels equation 2.png But when I tested it on Wikipedia, there were a few problems, so if we do get around to installing Texvc, we'll have to go hammer those problems out. Also, I'd appreciate it if we could find the exact fractions for those decimalised numbers - after all, the computer works in binary, not decimal - hence we won't have 0.008, we might instead have 523/65536 (or maybe 8/1000, but the result would still be stored as a binary fraction) - 刘 (劉) 振霖 07:22, 6 Mar 2005 (UTC)

I updated texerratic.png up there with something that looks more like yours. I didn't use Wikipedia for mine, I was using TeXnicCenter. I think the decimals are all right how they are, because they actually are more correct. I don't know exactly how the games handle decimal arithmetic, but it appears that the formulas use values accurate to 3 decimal places. (525/65536) or (523/65536) are not exactly 0.008, and if you put in those values, you'll get numbers that are off (I got about 6 points off for level 70). Sheep 13:47, 6 Mar 2005 (UTC)

Mmm. Since the experience points have to be accurate to about 7sf - I think the fractions should therefore be accurate to 7sf as well. But it might well be that they did use integer multiplication + division (× 8 ÷ 1000) rather than a simpler constant floating point multiplication (× (1 + 402653/2²⁴) × 2^-7) - which is approximately how accurate a single-precision IEEE 754 binary fraction is. Funnily enough, IEEE 754 single-precision fractions are accurate to approx 7sf (without exponentiation) by defintion (this one is 0.008 correct to 11sf)). - 刘 (劉) 振霖 14:40, 6 Mar 2005 (UTC)

Hmm. I just learned from Meowth that the game simply stores the values for these as constants in the game, so the fractions are just best-fit and the formula is not used in-game - 刘 (劉) 振霖 14:44, 6 Mar 2005 (UTC)

So we need to reach a conclusion. There will be a note added that says these values are stored in the game as constants, not calculated. The formulas, I guess, do not have to be specific to any system of number storage now, so are we saying that the decimal constants are acceptable? Sheep 15:04, 6 Mar 2005 (UTC)

quick question

Why does Level redirect here? An even better question is, why does this page link to Level, thus linking back to itself? --greengiant

Sorry my reply is formatted wrong, I don't know how to do it normally yet. Anyway, the last time I checked, Level is its own, seperate article. Maybe they fixed the problem you saw before. {{SUBST:Superbreeder]] What's up? 23:30, 16 October 2008 (UTC)

Modify this!

Can someone modify the erratic and the fluctuating part of experience? I want to understand it like all the other ones!

I think those formulas should be explained. It would help quite a lot of people who want to understand those formulas but haven't seen them before.Dullstar 02:16, 26 June 2010 (UTC)

Color-Coded

I color-coded the exp type descriptions to more-or-less match the colors used on the graph. I did this to make it easier to distinguish them from eachother. I wonder where I can find the look-up table in the game, I bet I can simplify those two honking large piecewise formulas. Twigpi 15:51, 20 November 2007 (UTC)

See http://www.upokecenter.com/games/rs/guides/exptable.html. The "1,050,000" in one of the top columns is a typo (should be "1,059,860"). At Level 1, the Experience is always "0" (here, they have it at "1"). Ultraflame 22:53, 10 December 2007 (UTC)

Organize Pokémon

I really think there should be either a list of Pokémon by experience types or a category for each type. You can find out on each particular Pokémon's page, but there is no way to find Pokémon based on their experience type. Cheesus Is Lord 13:23 23 January 2008

Here is the whole list. Someone could modify it and put it into the main article.

600000
Nincada/Ninjask/Shedinja
Volbeat
Swablu/Altaria
Zangoose
Lileep/Cradily
Anorith/Armaldo
Feebas/Milotic
Clamperl/Huntail/Gorebyss
Cranidos/Rampardos
Shieldon/Bastiodon
Finneon/Lumineon

800000
Cleffa/Clefairy/Clefable
Igglybuff/Jigglypuff/Wigglytuff
Happiny/Chansey/Blissey
Ledyba/Ledian
Spinarak/Ariados
Togepi/Togetic/Togekiss
Marill/Azumarill
Aipom/Ambipom
Misdreavus/Mismagius
Snubbull/Granbull
Corsola
Delibird
Smeargle
Skitty/Delcatty
Mawile
Spoink/Grumpig
Spinda
Lunatone
Solrock
Shuppet/Banette
Duskull/Dusclops/Dusknoir
Chingling/Chimecho
Luvdisc
Glameow/Purugly

1000000
Caterpie/Metapod/Butterfree
Weedle/Kakuna/Beedrill
Rattata/Raticate
Spearow/Fearow
Ekans/Arbok
Pichu/Pikachu/Raichu
Sandshrew/Sandslash
Vulpix/Ninetales
Zubat/Golbat/Crobat
Paras/Parasect
Venonat/Venomoth
Diglett/Dugtrio
Meowth/Persian
Psyduck/Golduck
Mankey/Primeape
Ponyta/Rapidash
Slowpoke/Slowbro/Slowking
Magnemite/Magneton/Magnezone
Farfetch'd
Doduo/Dodrio
Seel/Dewgong
Grimer/Muk
Onix/Steelix
Drowzee/Hypno
Krabby/Kingler
Voltorb/Electrode
Cubone/Marowak
Tyrogue/Hitmonlee/Hitmonchan/Hitmontop
Lickitung/Lickilicky
Koffing/Weezing
Tangela/Tangrowth
Kangaskhan
Horsea/Seadra/Kingdra
Goldeen/Seaking
Mime Jr./Mr. Mime
Scyther/Scizor
Smoochum/Jynx
Elekid/Electabuzz/Electivire
Magby/Magmar/Magmortar
Ditto
Eevee/Vaporeon/Jolteon/Flareon/Espeon/Umbreon/Leafeon/Glaceon
Porygon/Porygon2/Porygon-Z
Omanyte/Omastar
Kabuto/Kabutops
Sentret/Furret
Hoothoot/Noctowl
Natu/Xatu
Bonsly/Sudowoodo
Yanma/Yanmega
Wooper/Quagsire
Unown
Wobbuffet
Girafarig
Dunsparce
Qwilfish
Teddiursa/Ursaring
Slugma/Magcargo
Remoraid/Octillery
Phanpy/Donphan
Poochyena/Mightyena
Zigzagoon/Linoone
Wurmple/Silcoon/Beautifly/Cascoon/Dustox
Wingull/Pelipper
Surskit/Masquerain
Nosepass/Probopass
Meditite/Medicham
Plusle
Minun
Numel/Camerupt
Torkoal
Barboach/Whiscash
Baltoy/Claydol
Castform
Snorunt/Glalie/Froslass
Bidoof/Bibarel
Burmy/Wormadam/Mothim
Pachirisu
Buizel/Floatzel
Cherubi/Cherrim
Shellos/Gastrodon
Buneary/Lopunny
Stunky/Skuntank
Bronzor/Bronzong
Spiritomb
Croagunk/Toxicroak
Rotom

1059860
Bulbasaur/Ivysaur/Venusaur
Charmander/Charmeleon/Charizard
Squirtle/Wartortle/Blastoise
Pidgey/Pidgeotto/Pidgeot
Nidoran-F/Nidorina/Nidoqueen
Nidoran-M/Nidorino/Nidoking
Oddish/Gloom/Vileplume/Bellossom
Poliwag/Poliwhirl/Poliwrath/Politoed
Abra/Kadabra/Alakazam
Machop/Machoke/Machamp
Bellsprout/Weepinbell/Victreebel
Geodude/Graveler/Golem
Gastly/Haunter/Gengar
Mew
Chikorita/Bayleef/Meganium
Cyndaquil/Quilava/Typhlosion
Totodile/Croconaw/Feraligatr
Mareep/Flaaffy/Ampharos
Hoppip/Skiploom/Jumpluff
Sunkern/Sunflora
Murkrow/Honchkrow
Gligar/Gliscor
Shuckle
Sneasel/Weavile
Celebi
Treecko/Grovyle/Sceptile
Torchic/Combusken/Blaziken
Mudkip/Marshtomp/Swampert
Lotad/Lombre/Ludicolo
Seedot/Nuzleaf/Shiftry
Taillow/Swellow
Whismur/Loudred/Exploud
Sableye
Budew/Roselia/Roserade
Trapinch/Vibrava/Flygon
Cacnea/Cacturne
Kecleon
Absol
Spheal/Sealeo/Walrein
Turtwig/Grotle/Torterra
Chimchar/Monferno/Infernape
Piplup/Prinplup/Empoleon
Starly/Staravia/Staraptor
Kricketot/Kricketune
Shinx/Luxio/Luxray
Combee/Vespiquen
Chatot
Riolu/Lucario
Shaymin

1250000
Growlithe/Arcanine
Tentacool/Tentacruel
Shellder/Cloyster
Exeggcute/Exeggutor
Rhyhorn/Rhydon/Rhyperior
Staryu/Starmie
Pinsir
Tauros
Magikarp/Gyarados
Lapras
Aerodactyl
Munchlax/Snorlax
Articuno
Zapdos
Moltres
Dratini/Dragonair/Dragonite
Mewtwo
Chinchou/Lanturn
Heracross
Swinub/Piloswine/Mamoswine
Mantyke/Mantine
Skarmory
Houndour/Houndoom
Stantler
Miltank
Raikou
Entei
Suicune
Larvitar/Pupitar/Tyranitar
Lugia
Ho-Oh
Ralts/Kirlia/Gardevoir/Gallade
Slakoth/Vigoroth/Slaking
Aron/Lairon/Aggron
Electrike/Manetric
Carvahna/Sharpedo
Tropius
Relicanth
Bagon/Shelgon/Salamence
Beldum/Metang/Metagross
Regirock
Regice
Registeel
Latias
Latios
Kyogre
Groudon
Rayquaza
Jirachi
Deoxys
Gible/Gabite/Garchomp
Hippopotas/Hippowdon
Skorupi/Drapion
Carnivine
Snover/Abomasnow
Uxie
Mesprit
Azelf
Dialga
Palkia
Heatran
Regigigas
Giratina
Cresselia
Phione
Manaphy
Darkrai
Arceus

1640000
Shroomish/Breloom
Makuhita/Hariyama
Illumise
Gulpin/Swalot
Wailmer/Wailord
Seviper
Corphish/Crawdaunt
Drifloon/Drifblim

Ultraflame 20:45, 23 January 2008 (UTC)

I like this idea and think it would be useful. Anyone else agree? Eric the espeon 19:43, 25 November 2008 (UTC)

Yes, but I don't know if it belongs on this page. We could create categories, but I think I like the idea of another page better, like List of Pokémon by experience requirement, or something. — Laoris _(Blah) 19:52, 25 November 2008 (UTC)

Makes sense, and link to that list from here? Eric the espeon 18:19, 4 December 2008 (UTC)

???

If the formula returns a decimal, is the result rounded up or down? --S^hi_ny ^No_ct^ow_l 15:39, 26 May 2008 (UTC)

Always rounded down, I believe. Ultraflame 22:59, 26 May 2008 (UTC)

Hmm...

Does anyone else think that a chart that lists the experience needed to reach each level in an experience group would be a good addition to this page? TTEchidna 05:20, 1 July 2008 (UTC)

yes MathijsP 07:22, 1 July 2008 (UTC)

I added the charts, but, as I had a computer program generate them, I haven't been able to check the piecewise ones yet. It would be good if someone else could check the piecewise functions ("erratic" and "fluctuating") to make sure they're correct. --S^hi_ny ^No_ct^ow_l 00:15, 29 September 2008 (UTC)

the "Erratic" one is messed up. lvl 98 is 1.2 million, and lvl 99 is under 600,000. also, thats a lot of text, so, i added the show/hide ability. -- MAGNEDETH 00:24, 29 September 2008 (UTC)

It's still messed up. Can you fix it please? --S^hi_ny ^No_ct^ow_l 00:27, 29 September 2008 (UTC)

i dont know how to fix it, i just added the show/hide things. im just noting its not right. you said you had a computer do it. try again? -- MAGNEDETH 00:29, 29 September 2008 (UTC)

The formula for Level 98 comes out to 583539, which is correct. Ultraflame 23:05, 29 September 2008 (UTC)

well thats fine, but currently it says lvl 98 is 1185901, which it wrong. -- MAGNEDETH 23:21, 29 September 2008 (UTC)

Pokémon Gold/Silver Version ROM - Hack-O-Matic - "Secret" Experience functions

I'm sure someone else has noticed that if you use the ROM hack tool Hack-O-Matic to open a Pokémon Gold/Silver ROM and edit Pokémon, there will be eight "experience gradient" choices for each Pokémon as opposed to the four that were actually used in Generation II. I've figured out three of the four "secret" functions, and due to a glitch in Hack-O-Matic I can't examine the fourth one.
Hack-O-Matic displays the eight experience functions as eight "types" (the numbers go from Level 2 to 100):

1 Formula is 0.25n^3 + 15n^2 + 205n – 107

365 649 969 1324 1717 2148 2621 3135 3693 4295 4945 5642 6389 7186 8037 8941 9901 10917 11993 13128 14325 15584 16909 18299 19757 21283 22881 24550 26293 28110 30005 31977 34029 36161 38377 40676 43061 45532 48093 50743 53485 56319 59249 62274 65397 68618 71941 75365 78893 82525 86265 90112 94069 98136 102317 106611 111021 115547 120193 124958 129845 134854 139989 145249 150637 156153 161801 167580 173493 179540 185725 192047 198509 205111 211857 218746 225781 232962 240293 247773 255405 263189 271129 279224 287477 295888 304461 313195 322093 331155 340385 349782 359349 369086 378997 389081 399341 409777 420393

2

Same as Fifth Type

3 0.75n^3 + 10n^2 – 30

16 80 178 313 492 717 994 1326 1720 2178 2706 3307 3988 4751 5602 6544 7584 8724 9970 11325 12796 14385 16098 17938 19912 22022 24274 26671 29220 31923 34786 37812 41008 44376 47922 51649 55564 59669 63970 68470 73176 78090 83218 88563 94132 99927 105954 112216 118720 125468 132466 139717 147228 155001 163042 171354 179944 188814 197970 207415 217156 227195 237538 248188 259152 270432 282034 293961 306220 318813 331746 345022 358648 372626 386962 401659 416724 432159 447970 464160 480736 497700 515058 532813 550972 569537 588514 607906 627720 647958 668626 689727 711268 733251 755682 778564 801904 825704 849970

4 0.75n^3 + 20n^2 – 70

16 130 298 523 812 1167 1594 2096 2680 3348 4106 4957 5908 6961 8122 9394 10784 12294 13930 15695 17596 19635 21818 24148 26632 29272 32074 35041 38180 41493 44986 48662 52528 56586 60842 65299 69964 74839 79930 85240 90776 96540 102538 108773 115252 121977 128954 136186 143680 151438 159466 167767 176348 185211 194362 203804 213544 223584 233930 244585 255556 266845 278458 290398 302672 315282 328234 341531 355180 369183 383546 398272 413368 428836 444682 460909 477524 494529 511930 529730 547936 566550 585578 605023 624892 645187 665914 687076 708680 730728 753226 776177 799588 823461 847802 872614 897904 923674 949930

5

First Type

6

Fourth Type

7

Sixth Type

8

Unknown due to glitch in Hack-O-Matic. Hack-O-Matic provides graphs for each experience function, but they may not be accurate. Do you want me to upload that graph's picture anyway?
Are these "hidden" experience functions notable enough to be added to the article somewhere? Thanks. Ultraflame 20:44, 12 October 2008 (UTC)

I dont understand any of this? Could you make it clearer for the average user to be able to read, if an average person sees that they wont be able to understand it, only smart people will --Guardian of Earth |SGMS 2010

When a Pokémon gains experience it levels up. Different Pokémon level up at different rates, and take different amounts of experience to gain each level. Some Pokémon level up faster than others. The amount of experience a Pokémon needs to gain a level is determined by one of eight possible formulas. That's really the premise of the whole article. — Laoris _(Blah) 18:27, 4 December 2008 (UTC)

oh ok thank you, it was that last line "The amount of experience a Pokémon needs to gain a level is determined by one of eight possible formulas." I didnt really get till you explained it. I just thought they made the pokemon that way, i wasnt aware there was a formula, if this is in the article, can we have the easier to understand version, your version, underneath the complicated bit. --Guardian of Earth |SGMS 2010

I'll add some clarification to the article later. — Laoris _(Blah) 18:36, 4 December 2008 (UTC)

List of base experence yeald

Can someone add this in as a hide able table?- unsigned comment from Eric the espeon (talk • contribs) ; massive amount of data hidden by trom.

Already in List of Pokémon by effort value yield. Ultraflame 00:42, 21 November 2009 (UTC)

No, that is EV yield.. The list I provided was base EXP yield. Quite a major difference there. Eric the espeon 17:57, 21 November 2009 (UTC)

If you take a look at the article, there's a column marked "Exp.", which contains exactly the same data that you have provided. Confusing, I know. Ultraflame 19:06, 21 November 2009 (UTC)

Ok, fair enough. This article should probably link to that list then, no? Eric the espeon 21:59, 21 November 2009 (UTC)

Negative EXP

According to the "medium slow" formula, the EXP for level 1 should be -54. This is exploited in the Pokémon Blue TAS. However there's no sign of this in the article. Was it fixed in later games? - unsigned comment from Gyorokpeter (talk • contribs)

It was fixed as of Generation III at the latest, I believe. According to the article, "Due most likely to the issue of speed when using these formulas, the GBA games will simply use a lookup table for each value of any type instead of computing them. Because of this, these formulas are not actually part of the game mechanics." Ultraflame 05:41, 22 December 2009 (UTC)

Experience gain in battle

The section on the experience gained in battle only mentions the total experience gained, and not how experience is spread throughout multiple participants in the battle. That information seems like a useful mechanic and should be appended by someone who can add it. --Naokohiro 19:21, 9 January 2010 (UTC)

It is divided evenly among participants, unless some participants have EXP share or whatever. In this case, they receive their share plus the bonus from the held item. —darklordtrom 03:56, 10 January 2010 (UTC)

How is it divided among arbitrary amounts of Pokémon involved in the battle, and Pokémon with Exp. Share, including cases where Pokémon have both an Exp. Share and participated in the battle? There are many different cases. For example, when some Pokémon have Exp. Share and some don't, but some of the Pokémon with Exp. Share participated and some didn't, but also some Pokémon without Exp. Share that did participate, etc. --Naokohiro 04:41, 14 January 2010 (UTC)

I don't know exactly how experience is divided among multiple Exp. Share holders (does each Pokémon with Exp. Share receive (1/(number of Exp. Shares + 1))*(total experience), or does each receive (1/2)*(total experience)*(1/number of Exp. Shares), or something else?), but the experience that is not allocated solely as a result of Exp. Share being held is always divided evenly among the participants of the battle, regardless of whether or not those participants hold Exp. Share. Ultraflame 06:22, 14 January 2010 (UTC)

Switching???

I'll give an example.

Say I have a Charizard out, and the opponent has Venusaur. The opponent switches to Blastoise, and later I switch to Raichu. If the opponent switches back to Venusaur, and I KO it, will my Charizard still gain exp? - ^Axxonnthe_{Awesometrainer} 20:59, 4 May 2010 (UTC)

No, at least not in my experience. The Pokémon must face the Pokémon since it has last been sent out from its Poké Ball. Werdnae (talk) 21:40, 4 May 2010 (UTC)

Nope. Tried it before. ^ht14 21:59, 4 May 2010 (UTC)

Errors with formulas?

I was just double-checking the erratic output based on the formula given, when I realized that the given formula is impossible, in programming terms. Specifically, the game would never output two values for levels 50, 68, and 98, as the current formula shows it does. So I checked to see what the correct distribution of the formulas was, and came up with the following: level 50 could go either way, as both formulas it is attributed to result in the same output, level 68 is calculated based on the second formula given, and level 98 is calculated on the final formula. In other words, the actual division of the formulas should be something along the lines of the first applying to levels below or equal to 50, the second applying to levels greater than 50 but less than or equal to 68, the third applying to levels greater than 68 but less than to 98 (or, to keep similar formatting, apply to levels greater than 68 but less than or equal to 97), and the final one applying to levels greater than or equal to 98 (or, to keep formatting, apply to levels greater than 97). Glitchfinder 03:31, 6 May 2010 (UTC)

It would also appear that quite a bit has been left out of the fluctuating piecewise function. Specifically, it would appear that the formula used for levels one through 14 will change based on the level modulo 3, resulting in three different yet very similar formulas for these levels. The middle levels, from 15 to 34, remain with the same formula that is given. However, it would appear that there are two separate formulas used for levels 35 to 100, determined by the level modulo 2. Here is the code I was using to test the function, which is written in Ruby:

test.fluctuating[1] = 0
for i in 2...101
  if i >= 1 && i < 15 && i % 3 == 0
    val = ((i.to_f ** 3.0) * ((24.0 + ((i.to_f) / 3.0)) / 50.0))
  elsif i >= 1 && i < 15 && i % 3 == 1
    val = ((i.to_f ** 3.0) * ((24.0 + ((i.to_f - 1.0) / 3.0)) / 50.0))
  elsif i >= 1 && i < 15 && i % 3 == 2
    val = ((i.to_f ** 3.0) * ((24.0 + ((i.to_f + 1.0) / 3.0)) / 50.0))
  elsif i >= 15 && i < 35
    val = ((i.to_f ** 3.0) * ((14.0 + i.to_f) / 50.0))
  elsif i >= 35 && i <= 100 && i % 2 == 0
    val = ((i.to_f ** 3.0) * ((32.0 + (i.to_f / 2.0)) / 50.0))
  elsif i >= 35 && i <= 100 && i % 2 == 1
    val = ((i.to_f ** 3.0) * ((32.0 + ((i.to_f - 1.0) / 2.0)) / 50.0))
  end
  test.fluctuating[i] = val.truncate
end

Please note that the code is given as a demonstration of how the formulas are chosen, and does not include the necessary setup for the test variable. I don't know the Wikipedia math functions well enough to replace the image, so it would be a good idea for someone else to do it instead. Glitchfinder 05:17, 6 May 2010 (UTC)

I believe that the modulo operator business is already covered in the formulas, using the floor function symbol (looks a little like square brackets). But I do agree that the erratic functions' domains should be clarified as you said. Ultraflame 05:25, 6 May 2010 (UTC)

That would certainly explain the problem. There should probably be a note stating that those values are actually encased in a floor function, and not simply placed within brackets. (I had to look quite closely to see that they were in fact the floor function symbols, and not simply brackets, like I had assumed they were) Glitchfinder 05:47, 6 May 2010 (UTC)

Understanding the formulas

Could anyone tell me what is ment with "N" in the formulas? I can't find it anywhere on the page. - unsigned comment from Ragnearoa (talk • contribs)

"n" generally means "number" when used in equations as a variable, in this case... I believe the "n" stands for the base experience given out by a Pokemon. ▫▫ティナ^♫★ 15:21, 20 May 2010 (UTC)

"n" actually represents the level of the Pokémon. "e(n)" is therefore the number of Exp. Points a Pokémon needs to get from level 1 to level "n". Ultraflame 01:21, 21 May 2010 (UTC)

Erratic formula

The erratic formula's piecewise functions are defined inclusively in all inequalities, and thus overlap.

File:Math-formula5.png

File:Expcalc erratic.png

The left formula is how I believe it should be, the right is the current one. -- Pokey 07:38, 27 May 2010 (UTC)

At the overlaps, the values are equal. Ztobor 22:34, 15 July 2010 (UTC)

Does Not Compute

The Trivia sections on the species pages for Arceus, Blissey and Chansey all state that they give 13,933 experience at level 100 when owned by a Trainer. But based on the way that the experience given in battle formula is written right now (with a being 1 plus an additional possible 1.2 depending on Trainer status, trade, and foreign language), it only equals 8,014, not 13,933.

a=2.2 b=255 L=100 a+b+L=56,100 56,100/7=8,014

Add a Lucky Egg (multiply by 1.5) onto that and you only get 12,021. The only way to get the 13,933 is to multiply 1.5 for the trade and 1.7 for the foreign language after the initial Trainer experience (with a being 1.5) has been calculated, not to add .5 and .2 to a. --PhantomJunkie 17:17, 17 July 2010 (UTC)

We'd better look into that. Perhaps you don't add 0.5 to a, but multiply a by 1.5 instead. I'll see about it. Ztobor 02:50, 22 July 2010 (UTC)

Hold on, you mean: if you calculate 1.5x for the Lucky Egg, 1.7x for the foreign language trade, and 1.5x for the Trainer battle as well.

If we're going by that logic, it might actually be that you multiply everything, instead of adding to the multiplier. In which case it would actually be 14,753 experience points. I'll still have to look into it. Ztobor 02:58, 22 July 2010 (UTC)

Okay, I found out. I tried battling against a trainer's Level 28 Mime Jr. with a German-traded Gabite holding a Lucky Egg. The base Exp. is exactly 312, and it gained 1193, which is 1.5 x 1.7 x 1.5. So yes, you actually do need to multiply 1.5 twice, and the article as it stands now is wrong. Ztobor 03:18, 22 July 2010 (UTC)

Fixed. Ztobor 03:57, 22 July 2010 (UTC)

Minor wording issue

I didn't want to just go in and edit this without asking. The section on the Experience Underflow glitch contains the following sentence: It is due to this bug that no level 1 Pokémon can be found in the wild without glitching or hacking the game, and why, even though level 2-4 Pokémon can be found wild, Pokémon hatched at level 5 in the first two generations. However, since breeding and, by extension, hatching eggs wasn't added until Generation II, shouldn't the sentence be changed to state that "Pokémon hatched at level 5 in the second generation"? The following paragraph refers to level 5 hatching in Generation III, so I don't feel it's necessary to add it here. --PhantomJunkie 08:08, 19 July 2010 (UTC)

Changed to "Pokémon hatched at level 5 when eggs were introduced in Generation II." Ultraflame 01:30, 20 July 2010 (UTC)

Use of asymptotic notation

It's not really appropriate to use asymptotic notation in the manner this article does currently: the notion is not really useful unless the functions tend to 0 or infinity, as we can write many statements that say equally true and unhelpful things. For example, all six functions in the article are O(1) at n=100, or O(n), or O(n^10) (using the lim sup definition in the Wikipedia article, for example), none of which show the detail we want.

In other words, the notation is effectively meaningless the way it is being used. Perhaps considering f(x)-f(100) would be better.

Small extra note: to be rather fussy, similar remarks could be made about the way the article used "continuous" before I removed it: these functions are defined on the integers, and since there is no way of talking about limits on a finite set of integers, you can't talk about continuity either. (The integers are called discrete or totally disconnected because of the way the topologies on them must be defined, if anyone wants more detail.) Of course, this also means you can't use asymptotic notation in this case at all... Chappers 00:32, 24 July 2010 (UTC)

Alright, I'll remove that part. A friend of mine also said it was inappropriate too.

Also, about the "continuous" function part, I can see why it's notation abuse now. I just meant to say that it wasn't piecewise, and could be represented as one function. I guess "polynomial" is a good compromise if we can't really find a suitable word for it. Ztobor 19:42, 24 July 2010 (UTC)

Where to put this?

A graph showing the number of experience points required to go from one level to the next.

I created a table with the number of experience points required to go from one level to the next level, not just from level 1 to a level, but the formula won't fit anywhere.

Image is to the right. Ztobor 20:26, 24 July 2010 (UTC)

Misleading names

The Medium Slow (a.k.a. Parabolic) function is not actually strictly slower than the Medium Fast (a.k.a Cubic) function.

The only functions that are actually strictly ordered in terms of speed of levelling up are the Slow, Medium Fast, and Fast functions. The other three are all quirky.

What I'd suggest is renaming the Medium Fast function to simply Medium, and rename the Medium Slow function to... something. I still need to come up with a name, hence why I haven't actually edited the page yet. Ztobor 00:44, 1 August 2010 (UTC)

You could just change it to "Parabolic", but do leave a reminder that there exist different names for the experience functions. In my opinion "Erratic" and "Fluctuating" are okay names for their respective functions. Ultraflame 04:51, 1 August 2010 (UTC)

Or, to go with the quirky nature of the function compared to the others, rename it to "Quirky". Because "parabolic" is a huge misnomer. What do you think? (I didn't say anything about Erratic or Fluctiating... o_O) Ztobor 17:25, 1 August 2010 (UTC)

Hard to say, since you're the first person I've ever seen to refer to that experience function as "Quirky". I just mentioned Erratic and Fluctuating because you didn't, and I wasn't sure what your own opinion was on those names. Ultraflame 22:07, 1 August 2010 (UTC)

Well, I didn't mention those, but I think they're perfectly fine. It was just the four that were "ranked" by speed that I had a problem with - and even then it's only the Medium Slow one simply because it doesn't even fit into that rank nicely. Ztobor 13:34, 2 August 2010 (UTC)

Actually, even those three names should probably be changed to "Cubic", "Fast Cubic", and "Slow Cubic". It just doesn't make sense to rank them all by speed if they vary so much. Ztobor 13:36, 2 August 2010 (UTC)

You could try something like "Fast (Cubic)", "Medium (Cubic)", and "Slow (Cubic)". As for 1059860, we might be able to get away with calling it "Parabolic" and adding a note that the name only refers to the existence of a "not-purely-cubic" part in the function. Ultraflame 01:06, 4 August 2010 (UTC)

Not to mention all the article-renaming we'd have to do. >_< Ztobor 14:48, 4 August 2010 (UTC)

Experience table for PMD

It seems random, at least from what I've played so far. I have a character named Vino the Torchic, and here's her experience table for levels 7 through 15:

Level	Exp.
7	410
8	700
9	1250
10	2250
11	4150
12	7290
13	10430
14	15430
15	(etc.)

I don't see a formula here at all. Does anybody have the full table? Ztobor 23:43, 4 August 2010 (UTC)

Okay, no, I did find a table, from UPokeCenter. But man, those numbers are arbitrary. From levels 16 through 24, the EXP. required to get to the next level is always 6000. o_O Ztobor 23:47, 4 August 2010 (UTC)

New Exp. gain formula

I'm doing research on it right now, by watching playthroughs. Apparently, there's actually a +1 in the formula >_> Ztobor 02:47, 25 October 2010 (UTC)

Okay, here it is:

The only differences in the formula is that the 7 has now been changed to a 5, and the d_L represents the level relationship factor, which I'm still trying to figure out. Also, a constant of 1 is now added to all experience point yields, probably to prevent defeated monsters from yielding 0 experience points due to a low d_L modifier. Ztobor 03:00, 25 October 2010 (UTC)

Do you know if any rounding down is involved in the calculation? Hexagon Theory 03:50, 7 November 2010 (UTC)

There is, but I don't know where it applies. Ztobor 14:00, 7 November 2010 (UTC)

This formula can't be correct; no matter where I apply or don't apply rounding, I get slightly higher or slightly lower values than the experience I actually get in the game. Also, unless this one formula gets special treatment out of all the formulas in the game, the Pokémon games round down after every division or any other operation resulting in a non-integer, so if this were literally it, the level factor could never be anything between 0 and 1. Dragonfree 14:37, 7 November 2010 (UTC)

I think it multiplies by the defeated Pokémon's level + 2 first. It's been correct every time I've checked it, at any rate. Ztobor 03:48, 9 November 2010 (UTC)

Could you include some of your calculations and data? It could speed up finding out what's going on here. Dragonfree 23:00, 13 November 2010 (UTC)

Okay, take your first battle with the other starters - Level 5 vs. Level 5 (a level multiplier of 1) gives you 43 experience points. The starters' base yield is 28, which is multiplied by 1.5 because it's a trainer battle, giving you 42. No other multipliers apply, so you add 1 to 42 which gives you 43. Ztobor 22:57, 15 November 2010 (UTC)

For that matter, could you report some of your own data? You might be applying the wrong formula. I'll set up a report section on the talk page for it.

Base EXP

The page currently says that the Pokemon that gives the most base EXP is Blissey with 608 - only every other page on this wiki and on other sites says that it's only 255. (Tabunne, for the record, has a yield of 390.) It also says that the fifth-gen starters give the least at 28; this time, the NUMBER is right, but Magikarp still apparently only gives 20. What gives? --HeroicJay 22:51, 25 October 2010 (UTC)

Those yields are still from before Generation V. The information on other pages is correct for Generations IV and prior. I haven't had the time nor the patience to change them all yet. Ztobor 00:43, 26 October 2010 (UTC)

Magikarp's yield is now 40, BTW. Just so you know Ztobor 00:44, 26 October 2010 (UTC)

Report anomalies in the new Exp. gain formula

As far as we know right now, this is the new formula for experience used in Gen.5:

If you think this formula might not apply to a certain type of battle, append your case to the bottom of this post with the following data:

• The Pokémon that was defeated, and its level.
• The Pokémon that were involved in the battle, and their levels.
• Whether the battle was a Trainer battle or not.
• Whether the Pokémon involved were holding any items.

Your support is much appreciated. Ztobor 23:01, 15 November 2010 (UTC)

ERRATIC behaviour in Mathematica

So, I went to type in the EXP per level into Wolfram Mathematica 6, in an attempt to come up with both better graphs and interesting phenomonons. The non-piecewise functions turned out fine. However, the erratic and fluct. formulas kept coming up with strange behaviour in their plots, which I was able to filter out. However, look at this plot for the erratic formula:

This is the derivitive of the function of the Erratic experience group. Look at the range 68-98. Any ideas as to what is causing this? I had typed the formula in direstly as was shown on the article. --TruePikachu 20:37, 20 November 2010 (UTC)

*Facepalm* Just zoomed in, seems to be caused by the Floor() function and the erratic nature of that portion of the growth curve, which is why it didn't show up for Fluct. The formulas weren't designed for this kind of manipulation...time to weite an interger-based derivitave function to better model this...

Anyway, any ideas as to wether or not I should undergo this project? I can easily cut out the inputted formula (It is harder to include it rather than exclude it). The GenI formulas are rather well established, being elementry curves, but the GenIII curves I can do lots of research into, and add to the article under a different heading. For example, I can easily generate a graph of the 'changing multiplier' for these two functions, or superimpose all parts of the piecewise function (so someone could see what happens if one part of it is used for the entire function). --TruePikachu 20:47, 20 November 2010 (UTC)

I already created a changing multiplier graph. Third one down from the top. I also created an "EXP per level" graph (shown here on the talk page), but for lack of a place to put it, haven't put it onto the main page yet. Ztobor 03:19, 22 November 2010 (UTC)

Reshiram and Zekrom don't gain experience?

For some reason, people are editing this onto the article, even though it isn't true, at least from what I've seen of the playthroughs. Why do people think that it doesn't yield experience? Ztobor 00:59, 30 November 2010 (UTC)

What that means is when you defeat one you don't gain experience. I wonder if this is true or not. --Landfish7 01:03, 30 November 2010 (UTC)

When facing the wild version mascot before facing N, defeating it yields no Exp, as when the battlt ends, it'll just be there in the overworld as if nothing happened. Facing the one N has will yield Exp though. Shiramu Kuromu 03:18, 12 December 2010 (UTC)

Question

If there was a Pokémon that would change it's Exp group upon evolution, what would happen? Shiramu Kuromu 03:18, 12 December 2010 (UTC)

In Generation III at least, the Pokémon would change level so that it would be consistent with the what the new group's experience formula would dictate.

For example, if you hacked a Pikachu (1000000 at level 100) so that it would evolve into Gyarados (1250000 at level 100) at level 20, the Pikachu would only need 8000 exp. points to reach level 20. So suppose that the Pikachu receives 8000 exp. points, reaches level 20, and is allowed to evolve. But since Gyarados is in a different experience group, the Pokémon would actually drop down to level 18 (for Gyarados's group, 8000 points is only sufficient for level 18) immediately after evolution is complete. Meaning that when you check your Pokémon menu right after evolution, the Pokémon will appear as level 18, with 8000 exp. points.

This has apparently been observed with a Generation III ROM, but I don't know exactly what would in other Generations. Ultraflame 04:55, 12 December 2010 (UTC)

Just realized that a potential question might arise with my example: Would the newly-evolved Gyarados attempt to learn Bite (a level 20 move), even though it's technically level 18 after evolution? Unfortunately I don't have the answer to that particular question. Ultraflame 05:06, 12 December 2010 (UTC)