Interpolating a family of algebraic functions which tell you for how long you will be banned

NarcyChadlite

NarcyChadlite

Aesthete, scholar, Meghan's husband
Apr 11, 2021
764
Today I have been trying to discover a family of mathematical functions which spit out the number of days someone will be banned for based on their ban percentage.

Chapter 1. Building the function
The argument of the function is the ban percent (50% ban, 70% ban, 90% ban, 100% ban, you got the idea). For simplicity purposes, I am going to convert percentages into their decimal forms (namely, 0.5, 0.7, 0.9, 1). In this sense, the domain can only be the interval from 0 to 1 (you cannot be issued a 200% ban or a -10% ban). The range, on the other hand, is the entire real line. Mathematically, a 100% ban results in a permaban - ergo, an infinity of days.

f : [0; 1] -> ℝ

According to the nigger of neets.me, @Haruhi Suzumiya haruhifag, here is how the ban system works:
0% ban - of course, 0 days. It makes no sense to be banned for 0% ban, so the origin of the Cartesian system is included in the graph of the function. (Relation 1)
70% ban - 2 days ban
90% ban - 4 days ban
100% ban - permaban

So it makes sense to write the following mathematical relations (essentially, a system of functional equations):

f(0) = 0 (Condition 1, Relation 1)
f(0.7) = 2 (Condition 2)
f(0.9) = 4 (Condition 3)
lim{x->1}f(x)=inf (Condition 4, Relation 2)

2. Remarkable solutions

The first relation (f(0)=0) tells us an incredibly important aspect about the structure of our function. When x=0 (0% ban), f(x)=0 (banned for 0 days). If I wish to express my function as a fraction, this means that x MUST appear in the numerator. (Relation 2)

The second relation (the permaban case, where a user is banned for infinite days for a 100% ban) is also incredibly important.
At first, I thought the function must fundamentally be an exponential. But the condition made above calls for a vertical asymptote.

That's right, the line x=1 is a vertical asymptote to the graph of our function. So my function must be fundamentally rational! :D :D :D
Vertical asymptotes (pointing, in this case, towards positive infinity for the left limit) usually occur upon dividing by 0. This means that my function has the term (1-x) in the denominator! If x=1, 1-x =0, my function gets divided by 0 yielding positive infinity! :D (Relation 3)

From relations 2, and 3, our function must have this structure:

f(x) ~ x/(1-x)

Now, of course, this is not enough. Our function so far only satisfies conditions 1 and 2. It does have a vertical asymptote and it does include the origin point, but its graph does not cross the points (0.7, 2), (0.9, 4) (conditions 2 and 3), as my Desmos graph shows:
1642098531868


We shall further build upon this family of functions.

Chapter 3. Interpolation

Mathematically, graphing a function using only a system of points is known as interpolation (the graph must go through all specified points). This would have easily been achieved had our function been a polynomial. Unfortunately, that is not the case, since it is rational. There are 3 methods at the forefront of polynomial interpolation - Lagrange, Newton, and Neville. I have been trying using those interpolations myself, but to no avail. By crossing out the (1-x) term (and compensating for this by dividing by 1-x in the expression of the resulting polynomial), all three methods returned the same interesting result -
f(x) = -x^2 + x = -x(x - 1) = x(1 - x)

After the rearrangement made above it becomes quite obious that upon dividing by (1-x), the function turns into the linear function f(x)=x. It was to be expected. This method DOES NOT work.

After experimenting a bit with these functions, I came to the conclusion that an exponential might be the way to go.
Namely,
f(x) ~ x/(1-x) * e^(ax)
The base e was arbitrarily chosen (it could have been any other base and the constant a is adjusted based on the base). A is a really important coefficient. a=1 does not work by itself;
1642099036928

Now it might seem we got even further from our previous result! However, this expression gives us the freedom to manipulate the graph in multiple ways.

For simplicity purposes, I will purposefully try to structurally adjust this function and discover a constant a for which
2*f(0.7) = f(0.9) (law of proportionality)​

It's quite complicated to work out the correct values from the very beginning (namely, f(0.7)=2, f(0.9)=4). If I manage to solve an exponential equation and find the correct a for which that proportionality law holds true, I am almost done! :D :D :D

We could, then, write the function so as to exactly get f(0.7)=2 and f(0.9)=4:

f(x) = x/(1-x) * e^(a*x) * b
Where a and b are constants.

And THAT gentleman and gentleman (bcos femoids should rot in hell) is the FINAL form of our function. We now shall move onto calculating a and b.

Chapter 3. Exponential equation
(i wont bother posting the equations here, I used LaTeX instead):
1642100602005
1642100662897

And that is the value of a for which 2*f(0.7) = f(0.9). We have finally solved the proportionality law!!! Now we only have to solve for b and actually make the two values equal to 2, and 4 respectively!!! :D :D :D

Plotting the identified value for a into our function, we obtain that
f(0.7) ~ 0.234234971b = 2
f(0.9) ~ 0.468469942b = 4


We can therefore find a value for b:

b = 4.269217337...

Now here is what Desmos has to say:
1642100890238

It works!!! It fucking works!! :D :D :D
It respects all 4 necessary conditions. We can finally describe the function as
1642101044604

Which is lifefuel! Our mission here is done :)

Chapter 5. Final notes
I am not completely satisfied with this function in particular. It follows an unnatural concavity (notice how approximately f(x)=0.43 is an inflection point, the function becoming concave after starting out as convex). It does indeed follow our axiomatic rules, but it seems a bit farfetched. Nevertheless, disclosing a family of functions capable of telling us for how many days a user will be banned based on their ban percentage is truly an achievement on its own.

Chapter 6. Where next?
We shall further explore the behavior of trigonometric functions for our purposes.
The function f(x) = x/(1-x) * arcsin(ax) * b (with different values for a and b, of course) achieves quite the result (still, without crossing the points of interest):
1642101381725

A Taylor series might do the job too. But I have tried so far to stick to simple fundamental algebraic and trigonometric families of functions. The 3-step approach - finding remarkable solutions, interpolation/proportionality and fine-tuning the result with the coefficient b - seems quite rewarding.
Anyways, thank you for reading this!
You can now calculate for yourself the amount of days you will be banned for with the function above! :D :D :D

Tagging the brocels:
@Atila @Neetgod2 @Looksmax Refugee II @Optimus @elzde95 @kaang @NeverEndingWinter @Eren @MilkisterMooTwo @Alexanderr @Ritalincel @Lucillian @IGiveUp @VoDkA @Lain @chudur-budur @Copexodius Maximus
 
NarcyChadlite

NarcyChadlite

Aesthete, scholar, Meghan's husband
Apr 11, 2021
764
how the fuck am i supposed to tag you? i dont have your dumb quirky username copied all the time
 
Optimus

Optimus

NEET
Jan 8, 2022
561
This hurt my neet brain to read, although thanks for the tag. As Haruhi likes to say, "Rules are rules" :feelslol:
 
NarcyChadlite

NarcyChadlite

Aesthete, scholar, Meghan's husband
Apr 11, 2021
764
UPDATE: I HAVE FOUND A NEW FAMILY OF FUNCTIONS WITH A BETTER LOOKING CONCAVITY

Introducing

f(x) = x/(1-x) * arcsin(x(ax+b) * c

I had to manually fine tune values for a, b, c (c could easily be calculated once i find a and b, but the latter values are really difficult to calculate by hand. Not even Wolfram Mathematica is able to do the job. Matlab might). It's a pretty complicated inverse trigonometric equation. I managed to get rid of the arcsine by approximating a 14/27 term which subsequently appears (it's 2*0.7*0.1/(0.3*0.9)) with 0.5 and applying the half angle formula in a sine (basically sin(arcsin(x))=x and cos(arcsin(x))=+/-sqrt(1-x^2) but for our purposes we could have just chosen the positive sign)

Here's the graph and its behavior (notice how both the points are ALMOST crossed and how it follows a more natural progression)

1642134974473


WE HAVE A WINNER EVERYONE AND ITS THE ARCSINE FUNCTION
😍 😍 😍 😍 😍 😍 😍 😍 😍 😍 😍 😍
😍
1642135003822
😍
😍 😍 😍 😍 😍 😍 😍 😍 😍 😍 😍 😍
 
NarcyChadlite

NarcyChadlite

Aesthete, scholar, Meghan's husband
Apr 11, 2021
764
Holy shit nigger this was a stellar thread... I was on COCA when writing it.... Guess it really enhances the brains 🤣🤣🤣
 
NarcyChadlite

NarcyChadlite

Aesthete, scholar, Meghan's husband
Apr 11, 2021
764
Today I have been trying to discover a family of mathematical functions which spit out the number of days someone will be banned for based on their ban percentage.

Chapter 1. Building the function
The argument of the function is the ban percent (50% ban, 70% ban, 90% ban, 100% ban, you got the idea). For simplicity purposes, I am going to convert percentages into their decimal forms (namely, 0.5, 0.7, 0.9, 1). In this sense, the domain can only be the interval from 0 to 1 (you cannot be issued a 200% ban or a -10% ban). The range, on the other hand, is the entire real line. Mathematically, a 100% ban results in a permaban - ergo, an infinity of days.

f : [0; 1] -> ℝ

According to the nigger of neets.me, @Haruhi Suzumiya haruhifag, here is how the ban system works:
0% ban - of course, 0 days. It makes no sense to be banned for 0% ban, so the origin of the Cartesian system is included in the graph of the function. (Relation 1)
70% ban - 2 days ban
90% ban - 4 days ban
100% ban - permaban

So it makes sense to write the following mathematical relations (essentially, a system of functional equations):

f(0) = 0 (Condition 1, Relation 1)
f(0.7) = 2 (Condition 2)
f(0.9) = 4 (Condition 3)
lim{x->1}f(x)=inf (Condition 4, Relation 2)

2. Remarkable solutions

The first relation (f(0)=0) tells us an incredibly important aspect about the structure of our function. When x=0 (0% ban), f(x)=0 (banned for 0 days). If I wish to express my function as a fraction, this means that x MUST appear in the numerator. (Relation 2)

The second relation (the permaban case, where a user is banned for infinite days for a 100% ban) is also incredibly important.
At first, I thought the function must fundamentally be an exponential. But the condition made above calls for a vertical asymptote.

That's right, the line x=1 is a vertical asymptote to the graph of our function. So my function must be fundamentally rational! :D :D :D
Vertical asymptotes (pointing, in this case, towards positive infinity for the left limit) usually occur upon dividing by 0. This means that my function has the term (1-x) in the denominator! If x=1, 1-x =0, my function gets divided by 0 yielding positive infinity! :D (Relation 3)

From relations 2, and 3, our function must have this structure:

f(x) ~ x/(1-x)

Now, of course, this is not enough. Our function so far only satisfies conditions 1 and 2. It does have a vertical asymptote and it does include the origin point, but its graph does not cross the points (0.7, 2), (0.9, 4) (conditions 2 and 3), as my Desmos graph shows:
View attachment 30276

We shall further build upon this family of functions.

Chapter 3. Interpolation

Mathematically, graphing a function using only a system of points is known as interpolation (the graph must go through all specified points). This would have easily been achieved had our function been a polynomial. Unfortunately, that is not the case, since it is rational. There are 3 methods at the forefront of polynomial interpolation - Lagrange, Newton, and Neville. I have been trying using those interpolations myself, but to no avail. By crossing out the (1-x) term (and compensating for this by dividing by 1-x in the expression of the resulting polynomial), all three methods returned the same interesting result -
f(x) = -x^2 + x = -x(x - 1) = x(1 - x)

After the rearrangement made above it becomes quite obious that upon dividing by (1-x), the function turns into the linear function f(x)=x. It was to be expected. This method DOES NOT work.

After experimenting a bit with these functions, I came to the conclusion that an exponential might be the way to go.
Namely,
f(x) ~ x/(1-x) * e^(ax)
The base e was arbitrarily chosen (it could have been any other base and the constant a is adjusted based on the base). A is a really important coefficient. a=1 does not work by itself;
View attachment 30277
Now it might seem we got even further from our previous result! However, this expression gives us the freedom to manipulate the graph in multiple ways.

For simplicity purposes, I will purposefully try to structurally adjust this function and discover a constant a for which
2*f(0.7) = f(0.9) (law of proportionality)​

It's quite complicated to work out the correct values from the very beginning (namely, f(0.7)=2, f(0.9)=4). If I manage to solve an exponential equation and find the correct a for which that proportionality law holds true, I am almost done! :D :D :D

We could, then, write the function so as to exactly get f(0.7)=2 and f(0.9)=4:

f(x) = x/(1-x) * e^(a*x) * b
Where a and b are constants.

And THAT gentleman and gentleman (bcos femoids should rot in hell) is the FINAL form of our function. We now shall move onto calculating a and b.

Chapter 3. Exponential equation
(i wont bother posting the equations here, I used LaTeX instead):
View attachment 30278View attachment 30279
And that is the value of a for which 2*f(0.7) = f(0.9). We have finally solved the proportionality law!!! Now we only have to solve for b and actually make the two values equal to 2, and 4 respectively!!! :D :D :D

Plotting the identified value for a into our function, we obtain that
f(0.7) ~ 0.234234971b = 2
f(0.9) ~ 0.468469942b = 4


We can therefore find a value for b:

b = 4.269217337...

Now here is what Desmos has to say:
View attachment 30280
It works!!! It fucking works!! :D :D :D
It respects all 4 necessary conditions. We can finally describe the function as
View attachment 30281
Which is lifefuel! Our mission here is done :)

Chapter 5. Final notes
I am not completely satisfied with this function in particular. It follows an unnatural concavity (notice how approximately f(x)=0.43 is an inflection point, the function becoming concave after starting out as convex). It does indeed follow our axiomatic rules, but it seems a bit farfetched. Nevertheless, disclosing a family of functions capable of telling us for how many days a user will be banned based on their ban percentage is truly an achievement on its own.

Chapter 6. Where next?
We shall further explore the behavior of trigonometric functions for our purposes.
The function f(x) = x/(1-x) * arcsin(ax) * b (with different values for a and b, of course) achieves quite the result (still, without crossing the points of interest):
View attachment 30282
A Taylor series might do the job too. But I have tried so far to stick to simple fundamental algebraic and trigonometric families of functions. The 3-step approach - finding remarkable solutions, interpolation/proportionality and fine-tuning the result with the coefficient b - seems quite rewarding.
Anyways, thank you for reading this!
You can now calculate for yourself the amount of days you will be banned for with the function above! :D :D :D

Tagging the brocels:
@Atila @Neetgod2 @Looksmax Refugee II @Optimus @elzde95 @kaang @NeverEndingWinter @Eren @MilkisterMooTwo @Alexanderr @Ritalincel @Lucillian @IGiveUp @VoDkA @Lain @chudur-budur @Copexodius Maximus
bump
 
Activity
So far there's no one here

Similar threads

The_Hierophant
Replies
2
Views
219
anon1822fourthacc
anon1822fourthacc
K
Replies
13
Views
573
Kevin Logan
K
barcamgtow
Replies
8
Views
750
Polar-Z
Polar-Z
Lord_hierophantūs
Replies
3
Views
266
Lord_hierophantūs
Lord_hierophantūs
Top