Algorithm Optimisation Championship. - page 17

New comment
 
Everyone understands what a surface and vertices are. But what the input vector space is, personally, I don't understand. And what I should be looking for on it.
 
First there was crossing genes and mutations, binary chromosomes, evolution, then equations, coefficients, input space vector, multidimensional space, hundreds of optimisable parameters... I'm in a mess in my head:) Can we simplify the representation with a common, everybody understandable analogy from the real, physical world?
 
Реter Konow:
Everyone understands what a surface and vertices are. But what the input vector space is, personally, I don't understand. And what I am supposed to look for on it.
The dimensions in which the error surface is enclosed are the input coefficients. Usually the Z-axis is used to plot the value of the error. XY is the two input coefficients for the example. There can be as many as 1000 of them. The vector of coefficients for a particular observation specifies the coordinates of a point on the surface.
 
Реter Konow:
First there was crossing genes and mutations, binary chromosomes, evolution, then equations, coefficients, input space vector, multidimensional space, hundreds of optimisable parameters... I'm in a mess in my head:) Can we simplify the idea with a common, understandable analogy from the real, physical world?
Why don't you like my example? Without any analogies.
 
Thank you for the clarification. Now I understand it all. Especially the error vertex, which is put on the Z-axis. :)
 
Here's your example with the surface, as opposed to the explanation, I really liked it.
 
Реter Konow:
First there was crossing of genes and mutations, binary chromosomes, evolution, then equations, coefficients, input space vector, multidimensional space, hundreds of optimizable parameters... I'm in a mess in my head:) Can we simplify the representation with a common, everybody understandable analogy from the real, physical world?
A very simple analogy from the real world - a man and a woman. What should a man do to win a woman? Men solve such tasks on a daily basis and do not even think about it from this perspective. If men thought of it as an optimization problem, they would die lonely. Sometimes you don't need to know too much about a problem to solve it.
I now risk to be pelted with rotten tomatoes, but I just wanted to show a real example, where it is not known what exactly needs to be done to solve the problem.
 
Andrey Dik:

...

What is matan?

Mathematical analysis.
 
Because of my way of thinking, I am not able to find a solution to a vaguely worded problem, or to a problem that I do not understand. However, in any case I am not going to withdraw from the championship. Without claiming any place, I want to give my own, alternative approach to find peaks of function. If there are no objections, I will start implementing it.
 
Yuri Evseenkov:
Why don't you like my example? Without any analogies.

It is not adapted for GA.

It's like this:

MathAbs(34a+43b+16c+30d+23e-6268); - look for the minimum

1...101112131415161718192021222324...133
New comment
Reason: