Looking for a multivariate function with certain properties

Click For Summary

Discussion Overview

The discussion revolves around the exploration of a multivariate function that can model the effectiveness of combined traits in a roguelike game. The focus is on procedural generation and the mathematical representation of emergent properties, particularly how different parameters interact to create desirable or undesirable outcomes.

Discussion Character

  • Exploratory
  • Mathematical reasoning

Main Points Raised

  • One participant proposes the idea of a function that combines parameters to yield effectiveness, envisioning a 3D landscape with maxima representing optimal combinations of traits.
  • There is a suggestion that the function could involve sinusoidal combinations or Fourier transforms to create multiple maxima in a defined domain.
  • Another participant mentions the Fibonacci sequence as a potential basis for the function, where the proximity of combined traits to Fibonacci numbers could influence the output value.
  • A later reply introduces the idea of using a sum of exponential decay functions to model the desired terrain.
  • One participant expresses a reconsideration of the complexity of the problem, suggesting that simpler sinusoidal functions with varying phases might suffice to achieve the desired effects through interference.

Areas of Agreement / Disagreement

Participants do not appear to reach a consensus on the best approach to define the function. Multiple competing ideas and models are presented, indicating ongoing exploration and uncertainty regarding the optimal solution.

Contextual Notes

Participants express uncertainty about the mathematical framework required, with references to analysis and the complexity of the function potentially complicating the problem. There are also varying interpretations of what constitutes the desired "function" versus a "terrain."

Asteropaeus
Messages
143
Reaction score
36
As a undergrad future scientist, in my free time I like to do some programming to get some practice as my education programme doesn't include it, but it is a useful skill to have.

I am programming some type of roguelike game, so I can get away from graphics and focus on procedural generation, math systems, AI and stuff like that. It is never going to amount to anything but be entertainment and practice for myself.

So I am thinking about having the ability to design or evolve creatures or weapons. Some parts are going to be added together and the game is going to calculate the effectiveness.
So there are going to be some parameters. They can be the degrees of freedom. And a function is going to give the effectiveness given these parameters.

Now, the function can be to add or multiply the numbers and the higher the result, the better. But this is boring and one-dimensional. If traits or emergent properties are going to be combined, some should have synergy while others should be bad together.

When I imagine this as a 3d function with 2 independent variables, I can see a flat landscape with a number of round symmetrical hills. Ideally, all will be of slightly different sizes and heights. Every maxima of this hill is where the behavior the function describes is very efficient. Say, these represent two parts of say a weapon or a creature that are combined to give the final properties of the weapon/creature. The high values are going to be special rare solutions. Every part has one inherent value that is going to be one of the independent variables. All these maxima are going to be solutions the player would try to converge towards.

I have no idea how this function would look like algebraically. Maybe some silly combination of sinusoid functions. There must be some Fourier transform type of thing that can give a number of maxima in a certain domain.

Another thing I was thinking about is the Fibonacci sequence. If I take two random numbers and add them, the better these two numbers match a Fibonacci number, the higher the y value is. I can see how this can create a landscape like the one I described.

Am I on the right track? What would be the correct way to find a solution to a problem like this?

I guess a problem like this falls within the branch of analysis? Not sure so I posted this in general math. Feel free to move.
 
Last edited:
Mathematics news on Phys.org
Ok, so I figured out I can use a sum different e^-(x^2+y^2).
 
No. I think I am just looking at this in a way that's making it too complex.

I think I can just as well just use sinoid functions with different phases that have constructive or destructive interference. Then take a square and normalize it and see what the answer is.
 

Similar threads

  • · Replies 6 ·
Replies
6
Views
4K
  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 1 ·
Replies
1
Views
3K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 1 ·
Replies
1
Views
1K
  • · Replies 19 ·
Replies
19
Views
3K
  • · Replies 5 ·
Replies
5
Views
2K
  • · Replies 26 ·
Replies
26
Views
3K
  • · Replies 6 ·
Replies
6
Views
2K