Page 1 of 1

Weird saltpetre math

PostPosted: Fri Jan 25, 2019 2:42 am
by RobustAndRun
So I was curious about the effect of saltpetre on plants, specifically its effect on production stats
Code: Select all
myseed.adjust_production(-round(salt/100)-prob(salt%100))
myseed.adjust_potency(round(salt*0.5))


so i wrote a quick python script to test values of this, using an interpretation of .dm's prob() function

Spoiler:
Code: Select all
import random

def splash(salt): #calculates one addition of a given amount of saltpetre

    rng = random.randrange(1,101) #this block is just my recreation the prob() function
    if rng < salt%100:
        rng = 1
    else:
        rng = 0
       
    prod_adj = (-round(salt/100)-rng)
    pot_adj = round(salt*0.5)
   
    return (prod_adj, pot_adj)


for i in range(25,101,5): #i = the units tested, from 25 to 100 in steps of 5
    avgProdList = [] #fills up with the test values for production changes, and potency changes go below
    avgPotList = []
    for j in range(50): #run each one 50 times, idunno
        a = splash(i)
        avgProdList.append(a[0])
        avgPotList.append(a[1])
    if len(avgProdList) > 1 and len(avgPotList) > 1: #conditional to avoid div by 0, calculates the average of each, prints em
        avgProd = sum(avgProdList)/len(avgProdList)
        avgPot = sum(avgPotList)/len(avgPotList)
        print(i,"u: prod change = ",avgProd," | pot change = ",avgPot)
    else:
        pass

which in short just tests every 5u interval a bunch of times and prints the avg effect on production and potency, but what I found gets really weird around the 100u mark

test outputs:
Spoiler:
Code: Select all
25 u: prod change =  -0.08  | pot change =  12.0
30 u: prod change =  -0.32  | pot change =  15.0
35 u: prod change =  -0.26  | pot change =  18.0
40 u: prod change =  -0.38  | pot change =  20.0
45 u: prod change =  -0.4  | pot change =  22.0
50 u: prod change =  -0.44  | pot change =  25.0
55 u: prod change =  -1.5  | pot change =  28.0
60 u: prod change =  -1.56  | pot change =  30.0
65 u: prod change =  -1.6  | pot change =  32.0
70 u: prod change =  -1.62  | pot change =  35.0
75 u: prod change =  -1.76  | pot change =  38.0
80 u: prod change =  -1.78  | pot change =  40.0
85 u: prod change =  -1.78  | pot change =  42.0
90 u: prod change =  -1.96  | pot change =  45.0
95 u: prod change =  -1.98  | pot change =  48.0
100 u: prod change =  -1.0  | pot change =  50.0


do I have a poor understanding of the prob() math or is the original coding just bad??

Re: Weird saltpetre math

PostPosted: Fri Jan 25, 2019 3:20 am
by somerandomguy
100%100 is 0, so potloss with 100u is always round(100/100) [-1] - 0

It's just awful code

Re: Weird saltpetre math

PostPosted: Fri Jan 25, 2019 7:14 am
by oranges
sound about right lol

Re: Weird saltpetre math

PostPosted: Fri Jan 25, 2019 11:51 am
by 4dplanner
You got the python wrong lol

(Round is the floor operation in byond)

Re: Weird saltpetre math

PostPosted: Sat Jan 26, 2019 2:11 am
by RobustAndRun
4dplanner wrote:You got the python wrong lol

(Round is the floor operation in byond)


alright but what would have been the correct way of doing this?

Re: Weird saltpetre math

PostPosted: Sat Jan 26, 2019 3:50 pm
by 4dplanner
Code: Select all
import random
import math #for floor, can't remember if you need it

def splash(salt): #calculates one addition of a given amount of saltpetre

    rng = random.randrange(0,100) #we want rng < 100 to be a 100% chance, rng < 1 to be a 1% chance etc
    if rng < salt%100:
        rng = 1
    else:
        rng = 0
       
    prod_adj = (-math.floor(salt/100)-rng) #this is what was giving your weird result
    pot_adj = round(salt*0.5)
   
    return (prod_adj, pot_adj)


for i in range(25,101,5): #i = the units tested, from 25 to 100 in steps of 5
    avgProdList = [] #fills up with the test values for production changes, and potency changes go below
    avgPotList = []
    for j in range(5000): #gotta get that SAMPLE SIZE
        a = splash(i)
        avgProdList.append(a[0])
        avgPotList.append(a[1])
    if len(avgProdList) > 1 and len(avgPotList) > 1: #conditional to avoid div by 0, calculates the average of each, prints em
        avgProd = sum(avgProdList)/len(avgProdList)
        avgPot = sum(avgPotList)/len(avgPotList)
        print(i,"u: prod change = ",avgProd," | pot change = ",avgPot)
    else:
        pass

Re: Weird saltpetre math

PostPosted: Sun Jan 27, 2019 3:26 am
by RobustAndRun
4dplanner wrote:
Code: Select all
smart man stuff



aite cool thx, next time I play botanist I'll test this out in game

Re: Weird saltpetre math

PostPosted: Tue Jan 29, 2019 1:34 pm
by RobustAndRun
Alright update, tested it in game and found that 99u is indeed the best way to reduce production time, with 100u not doing anything outside of potency. Will make a new maffs function for this soon.

Update: I like the function y = round(-0.1 - 0.04x + 0.0005 x^2)
(y = prod change, x = saltp amount.)
Might add a prob into this for some RNG goodness.

This means that dumping gigantic amounts of saltpetre (above 50u at once, which is currently the ideal) will increase potency but have diminishing returns or even cause damage (90u+) to production speed.
This partially mirrors the effect of potassium nitrate irl, as a plant with an excessive amount of potassium ions in its soil/internal chemistry will have difficulty absorbing other nutrients such as nitrogen and magnesium.

Re: Weird saltpetre math

PostPosted: Sat Feb 02, 2019 10:09 pm
by Cobby
are you saying this crap is actually not crap

Re: Weird saltpetre math

PostPosted: Sun Feb 03, 2019 1:00 am
by 4dplanner
I've just tested and 100u definitely does work as intended.

Re: Weird saltpetre math

PostPosted: Sun Feb 03, 2019 1:01 am
by 4dplanner
Basically this doesn't need changing, it works fine