hone math

[Bianca666]

hey so if we have a weapon like
Def Level 16
Def Level 1
Def. Level 3
Def Level 5

how do we do the math for the chances of 37 38 39 and 40 hone on this? Such as 1 in 16,000 hone for 39 etc

[???????]

1 in 200k

[Bianca666]

1 in 200k

I asked for math not for a random figure.

[???????]

Since no one math check'ed me (dissapointment)

If we assume:

p(A) 15 ~ 30 = 1/16
p(B) 1~ 5  = 1/5
p(C) 1~ 5  = 1/5
p(D) 1 ~ 5 = 1/5

we convert:

p(A) = 0,0625
p(B) = 0,2
p(C) = 0,2
p(D) = 0,2

we apply formula:

p(A)*p(B)*p(C)*p(D) 

aka 0,0625*0,2*0,2*0,2 = 0,0005

we transform into fraction to multiply and later convert decimal

(0,0005*100)/(1*100)

(0,05*100)/(100*100)

(5/5)/(10000/5)

1/2000 

aka 1 in 2000

You can also do in your calculator the following: (1/16)*(1/5)*(1/5)*(1/5) and get an insta result but we dont always have our calculators with us right? (my prof told me so)

[Nimy]

Since no one math check'ed me (dissapointment)

If we assume:

p(A) 15 ~ 30 = 1/16
p(B) 1~ 5  = 1/5
p(C) 1~ 5  = 1/5
p(D) 1 ~ 5 = 1/5

we convert:

p(A) = 0,0625
p(B) = 0,2
p(C) = 0,2
p(D) = 0,2

we apply formula:

p(A)*p(B)*p(C)*p(D)

aka 0,0625*0,2*0,2*0,2 = 0,0005

we transform into fraction to multiply and later convert decimal

(0,0005*100)/(1*100)

(0,05*100)/(100*100)

(5/5)/(10000/5)

1/2000

aka 1 in 2000

You can also do in your calculator the following: (1/16)*(1/5)*(1/5)*(1/5) and get an insta result but we dont always have our calculators with us right? (my prof told me so)

You are implying that every stat has the same chance of appearing. I believe thats wrong.

[Bianca666]

You are implying that every stat has the same chance of appearing. I believe thats wrong.

this is correct, there are internalized weighted ratio the higher you go up the range of numbers toward the best hone the less chance I think, let me see if i can pull the C++ for this from sources

[Bianca666]

Code: [Select]

class arg_addon<DOUBLE_POINT> : public addon_handler
{
public:
virtual int GenerateParam(int datatype,addon_data & data,int arg_num)
{
ASSERT(arg_num == 2);
data.arg[0] = abase::RandNormal(data.arg[0] , data.arg[1]);
return 1;
}

virtual int ApplyAtGeneration(const addon_data & data, void * essence,size_t size,prerequisition * require)
{
return 0;
}
inline void Check(const addon_data & data)
{
//由于精炼属性的存在，此检查作废
//ASSERT(addon_manager::GetArgCount(data.id) == 1);
}
};


template <>public class for DOUBLE_POINT structured rand_prop addons (honeable addons with lower and upper cap aka 2 args)

we can see RandNormal is applied during generation

and if we go to SpecRand   we see from this snippet our routine for RandNormal

Code: [Select]

struct SpecRand
{
int * IndexList;
int   IdxCap;
int   IdxIndex;
float * RandList;
int RandCap;
int  RandIndex;
SpecRand():IndexList(0),IdxCap(0),IdxIndex(0),RandList(0),RandCap(0),RandIndex(0){}
int RandSelect(int num)
{
if(!IndexList) return 0;
if(IdxIndex >= IdxCap) return 0;
if(num <=0) return 0;
int idx =IndexList[IdxIndex++];
if(idx >= num) idx = num - 1;
return idx;
}

int RandNormal(int lower, int upper)
{
if(!RandList) return lower;
if(RandIndex >= RandCap) return lower;
float r = RandList[RandIndex ++];
if(r <0) r = 0.f;
if(r >=1.0f) r = 0.9999999f;
return (int)((upper - lower + 1)* r + lower);
}

float Rand(float lower, float upper)
{
if(!RandList) return lower;
if(RandIndex >= RandCap) return lower;
float r = RandList[RandIndex ++];
if(r <0) r = 0.f;
if(r >=1.0f) r = 0.9999999f;
return (float)((upper - lower)* r + lower);
}
};
If any dev want to trace further into PW source code for the definitions of all these vars to make this into a proper formula to use would be nice. For example I don't know what RandList or RandIndex are I think they are vars that store the data of which value between upper and lower is selected but don't actually know.

but so assuming we get answers to that, the actual formula is (int)((upper - lower + 1)* r + lower)

[???????]

You are implying that every stat has the same chance of appearing. I believe thats wrong.

You are correct, the calc was done assuming in the first line all chances are equal, the lower and upper ones can have diff chances or be messed with, iirc low botton and highest up was =~equal, only having variations in the middle* (in nirvana times, i may be wrong because it was also another private server so chances couldve been altered*).

But since we dont have acess to the real %'s related to epw's r8 it is what it is until inf tells us (which wont do anything but give fuel for players to complain about unfair rng so i doubt he will), or Bianca cracking the code again and getting banned again lmao.

But if we figure it out i ll gladly try to work out the numbers

[???????]

- Rank 8 Weapon rerolling now only give Attack Level and Defense Level and a few other class related stat, like accuracy on Archer or interval on Assassin. These are the average probabilities for Attack Level and Defense Level:

Value: 40 - 0.0006% = 1 /  154320.99
Value: 39 - 0.0041% = 1 /   24366.47
Value: 38 - 0.0248% = 1 /    4025.76
Value: 37 - 0.0863% = 1 /    1159.02
Value: 36 - 0.2726% = 1 /     366.77
Value: 35 - 0.6237% = 1 /     160.34
Value: 34 - 1.3068% = 1 /      76.52
Value: 33 - 2.3287% = 1 /      42.94
Value: 32 - 3.9037% = 1 /      25.62
Value: 31 - 6.5579% = 1 /      15.25
Value: 30 - 9.5270% = 1 /      10.50
Value: 29 - 3.9058% = 1 /       7.19
Value: 28 - 7.8249% = 1 /       5.61
Value: 27 - 2.3038% = 1 /       4.48
Value: 26 - 1.2618% = 1 /       3.20
Value: 25 - 3.1514% = 1 /       3.02
Value: 24 - 5.6708% = 1 /       2.80
Value: 23 - 5.7485% = 1 /       2.19

These chances can vary a bit between classes that have 3 unique stats and those that have 2 unique stats, but they are similar. Chances of interval on Assassin/Duskblade/Warrior are around the same, though it's now much easier to get interval+useful stat. Note that you can only get 40 max now. The unique stat attack and defense levels have been changed from 25 to 20. This is to keep balance the same. If it remained 25 then people could easily get 43 Atk or Def and it would affect balance.
- New weapons can't be honed. You can still hone old weapons.

I found this by Burdette, its related to the crafting not hone, but one could reverse engineer this to try to get numbers out of it.

[Whomper400]

Bianca where is your post for demon psy no 100% silence?Is it fixed

[Bianca666]

Bianca where is your post for demon psy no 100% silence?Is it fixed

it is not fixed

[Yuko]

I found this by Burdette, its related to the crafting not hone, but one could reverse engineer this to try to get numbers out of it.

Do you still play the game by any chance sir ZE_CAPETA? 

[Gomez]

got 38 def lv with 100 pe 

[???????]

Do you still play the game by any chance sir ZE_CAPETA?  

Eh, we all play the game in some way, i always defined "playing" as being conected to the game in some form, there are active players and there are people that just come and go.

But how about instead of epw how abt we play a game only us 2? I still remember how hot u look like winky wonk~