I Best way to fit three functions

  • I
  • Thread starter Thread starter LCSphysicist
  • Start date Start date
  • Tags Tags
    Fit Functions
LCSphysicist
Messages
644
Reaction score
162
TL;DR Summary
I have basically three set of data, each set corresponding to a function, so that the data should fit the function, and each function is dependent of 4 parameters. How to find the best 4 parameters that fit it?
So I have $$f(x,y,z,t,n) = 0,g(x,y,z,t,n) = 0,h(x,y,z,t,n) = 0 $$ and i need to find the best ##[x,y,z,t]## that fit the data, where n is the variable. Now, the amount of data for each function is pretty low (2 pair for f (that is, two (n,f)), 3 pair for g and another 3 pair for h)

The main problem here is: the functions are highly non-linear. So i have no idea how can i write a program to find the best x,y,z,t!

I have thought of try to simultaneously minimize the residues, for example, for f:
$$R(f) = \sqrt{(f(x,y,z,t,n_1) - y_1)^2+(f(x,y,z,t,n_2) - y_2)^2}$$
(where y is the f obtained on thee data at n_1 (the pair i have cited above))

But this minimization subject to the condition that ##R(g), R(h)## (here, three terms) also be minimum. So maybe i could use Lagrangian method:
$$dR(h) = \lambda dR(g) + \mu dR(h)$$
But this is going to be extremelly massive and tedius to write a code, and the code will take days to run. As i said, the functions are non-linear.

Any method suggestion? or program suggestion?
 
Mathematics news on Phys.org
Extremely massive and tedious is probably the best you can get. If it was one occurrence, I'd advise WolframAlpha or plot the 3 functions and visually pick the best fit (further improved with Newton's approximation method).

For now I'd advise to combine the requirements for all the data points on f, g and h and rewrite them as ##function - data = 0## and then the square of the left hand side must be minimal for all data points simultaneously: so take the derivative with respect to ##n## of these squares all added up and approximate where this formula in ##n## equals 0 using Newton's approximation method or some better root finding algorithm.

Wiki on Newtons method
 
Last edited:
You'll need to decide what you consider "best fit". Write that down mathematically as fitquality(x,y,z,t), then feed it in a program for minimization.

The sum of squared residues for all data points (data point minus the applicable function) sounds like a good start.
 
  • Like
Likes Maarten Havinga
Insights auto threads is broken atm, so I'm manually creating these for new Insight articles. In Dirac’s Principles of Quantum Mechanics published in 1930 he introduced a “convenient notation” he referred to as a “delta function” which he treated as a continuum analog to the discrete Kronecker delta. The Kronecker delta is simply the indexed components of the identity operator in matrix algebra Source: https://www.physicsforums.com/insights/what-exactly-is-diracs-delta-function/ by...
Fermat's Last Theorem has long been one of the most famous mathematical problems, and is now one of the most famous theorems. It simply states that the equation $$ a^n+b^n=c^n $$ has no solutions with positive integers if ##n>2.## It was named after Pierre de Fermat (1607-1665). The problem itself stems from the book Arithmetica by Diophantus of Alexandria. It gained popularity because Fermat noted in his copy "Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos, et...
I'm interested to know whether the equation $$1 = 2 - \frac{1}{2 - \frac{1}{2 - \cdots}}$$ is true or not. It can be shown easily that if the continued fraction converges, it cannot converge to anything else than 1. It seems that if the continued fraction converges, the convergence is very slow. The apparent slowness of the convergence makes it difficult to estimate the presence of true convergence numerically. At the moment I don't know whether this converges or not.
Back
Top