Statistical Survey on alchemy mechanics

[gnox24]

During the last days i thought about a way to make use of the new alchemy equipment. I came up with the following workflow, which can work if certain conditions are met.

The Basic principle is:

1. Pressure cook an item
2. Put into Test Tubes
3. Put it into Alembic
4. Repeat at 1

Step 1 raises the purity for sure and is the driving factor of the process. Step 2 is required to (randomly) raise some purity values. Together with the first step, the mean value of the random is slightly shifted to positive values. This means, provided enough runs, purity should go up over time. The main problem is the instability, which has to be countered with an alembic.

And here comes crucial point:
Does the alembic provide enough reduction of instability, while removing less purity than the Pressure Cooker/Test Tubes create?

I wrote an test application, which implements the functionality of the alchemy processors (Test Tubes, Pressure Cooker, Alembic, Retort) and creates test runs with changing parameters. The basic principles work, but i lack reasonable input parameters (like change rates, instability increments, ...).

For this reason i started this thread and ask all fellow pilgrims to help me gathering data. Assuming that enough data can be aquired, i'll run statistical analyses on them and post the results here. What i need are test runs from all alchemy processor formated like this:

Code: Select all

("Alchemy Processor Name",
[purityValues before],
[purityValues after],
["stability before", "stability after"])

Here is an example:

Code: Select all

("Pressure Cooker",
[.2, .2, .35, .25],
[.19, .19, .35, .27],
["slightly unstable", "somewhat unstable"])

("Pressure Cooker",
[.2, .2, .35, .25],
[],
["slightly unstable", "destroyed"])


I would appreciate if you guys could use the pattern above, because it helps parsing the data automatically. If some equipment is destroyed like a stove or an alchemy processor, just write it down in plain text until i figure out a better way.

Every sample helps!

Thanks.

[n0kturnal]

Finally, someone to cut through all the whining and put some effort into unwrapping the alchemy system. The mystification of the system is by design, but hopefully together we can figure this out.

[Darkside]

I wish I had the resources to help gather data on the pricey alchemy stuff - but I think one problem in your guidelines is that I *think* any instability at all could cause the instrument to break - so if the pressure cooker in step 1 causes instability increase, then you chance losing your test tubes.

[gnox24]

You dont have to follow the pattern to produce samples, every use of alchemy equipment helps. I would suggest using the alembic after every step or dont use the other stuff at all.

And i forgott one thing: only do single ticks. For example: the pressure cooker tick at ~one hour.

[Avarice]

If the item you are using is 25/25/25/25 neutral, you start at step 2.

I personally do this though.

1. Test Tubes
2.Cooker
3.Retort
4. Alembic

And go straight to the Alembic if the instability of an item becomes greater than somewhat unstable.

[florenso]

my items says nothing about the stability yet.

and i also dont understand the humor part...

Code: Select all

[.2, .2, .35, .25]

why not for example

Code: Select all

[20,40;15,60;27,90;36,10]

i mean, that's the proper way to write it down

(i am not sure where you are from but in my county they use a "," they also do that in salem)

but thx for taking the lead for tackling this!
i will for sure give you information!

[lachlaan]

The formula they use assumes the alchemical composition to be in the format salt+mercury+sulphur+lead = 1. I mean, it'd be easy to do what I do when calculating it, and just divide the result of the addition by 10000, then doing whatever else is needed.

[gnox24]

I used the '.' because it is used like this in programing languages. The comma has usually some different meaning. My values are ranged between 0 and 1 and not between 0% and 100%. But you can use both

[MagicManICT]

International standards are a decimal if I remember correctly. It's also the English method (I'm not aware of any of the English countries--UK, US, AU, NZ--that use commas). In game, Java uses the International settings of your operating system for things such as decimal separators and such. That's why you see it like you do.

Items don't show anything about stability if they haven't gone through any alchemical processes yet.

[gnox24]

Now i got some data on the pressure cooker and the alembic. The other two will take a while, because i lack the skill and money to get them. Test Tubes should come in as soon as i have Perennial Philosophy at 10 and can accuire the new alchemy skill to build them.

Here are the results:


Pressure Cooker:
The pressure cooker increases the second last purity value by 1%. The lowest two are decreased about 1%, weighted by their value. The actual formulas are:

Code: Select all

purity values: [a, b, c, d], with 1 >= a >= b >= c >= d >= 0;

a = a
b = b + 0.01
c = c - 0.01 * c / (c + d)
d = d - 0.01 * d / (c + d)

mean absolute purity change = 0.005


I'm not sure how they behave if b id closer to a than 0.01 or c and d are to close to 0, But i'm sure the multiplier is capped to make it fit.

To make a statement about stability increments i ordered and valued the statbility values:

Code: Select all

('a bit' = 1, 'slightly' = 2, 'somewhat' = 3, 'clearly' = 4, 'very' = 5)

For an item which wasn't used before in an alchemy process, there appears to be a 50% chance getting either 'a bit' or 'slightly'. All successive steps appear to have a 50% chance to be raised about 2 levels (eg.: 'slightly' --> 'clearly' or 'a bit' --> 'somewhat').

I dont have sufficient data to make assumptions about loss of items/equipment, but i tell my results: Out of 80 test ticks with max staibility value 'clearly', i had two destroyed items. One at 'clearly' and one at 'somewhat'.


Alembic:
The alembic equals the purity values of an item by a fixed value, depending on the distance to 25%. One can say: The more pure an item is, the more it get's equalized.

Code: Select all

purity values before:      pB_i, with i from 1..4 and 0 <= pB_i <= 1
purity values after:       pA_i, with i from 1..4 and 0 <= pA_i <= 1

pA_i = pB_i + (0.25 - pB_i) / 16

mean absolute purity change  is between 0 (0% pure item) and 0.0234 (100% pure item)


I determined the factor 16 empirically, but it works for all results so far.

The stability decrements seem to behave like the pressure cooker increments. For every tick there is a 50% chance to be decreased by 2 points. But i need more data to verify the results. I'm also not sure how it behaves if the value is 1 (a bit) and cant be decreased further. Stay tuned to get better results.



Conclusions so far:

At the moment it looks like, that the Alembic can counter the pressure cookers stability increments. But it's only usefull for low purity items. Eg for a [15, 10, 40, 35] item the changes are [0.63, 0.94, -0.94, -0.62]. So it's maybe usefull at beginner levels, but you can barely afford the alchemy items at this point.

If it's really usefull at start, depends on the results for test tubes.