Login
Register
Search
Home
Help
Search
Login
Register
Official Forum for TuringBot Software
General Category
Symbolic Regression General
Is it possible to find constants in a customized function?
« previous
next »
Print
Pages: [
1
]
Is it possible to find constants in a customized function?
3 Replies
1597 Views
xntang88
Newbie
6
Is it possible to find constants in a customized function?
«
on:
July 25, 2021, 09:35:37 am »
For example, is it possible to find the constants (a,b,c) in the following customized function?
z = (ax+by)^c, where x,y,z are variables.
Logged
admin
Administrator
Newbie
18
Re: Is it possible to find constants in a customized function?
«
Reply #1 on:
July 29, 2021, 10:00:42 pm »
Not at the moment, but many users have been requesting this feature: optimize z = (ax+by)^c where abc are arbitrary functions. This is a planned feature that will be implemented in the next versions of the program.
Logged
Jiminy
Newbie
1
Re: Is it possible to find constants in a customized function?
«
Reply #2 on:
August 06, 2021, 03:59:08 am »
yeah put me in for that. It's essential. Eureqa allowed multiple functions as well and IIRC I could specify constants as parameter free functions, e.g. f1(), f2().
Also on my wish list is a step([variableList]) function
Nice so far, although I may have a problem with wild processes remaining on my Windows 10 box - still investigating that.
Cheers
Jiminy
Logged
admin
Administrator
Newbie
18
Re: Is it possible to find constants in a customized function?
«
Reply #3 on:
September 20, 2021, 09:57:34 pm »
Multi-target optimization has been implemented in
TuringBot 1.96
.
In your example, to fit z = (ax+by)^c, you should now use
z = pow(f()*x+f()*y, c)
f() means a constant, ie a function with no input variables. You can also include terms which are functions your input variables, like y = x*f(x,z) + y*f(z) + f(x,z,y).
Logged
Print
Pages: [
1
]
« previous
next »