it's written on section B.1 about characteristics of TinyGP, and point
1 said that the terminal is a user-definable number of floating point
variables (named x1 to xn).
is it mean that TinyGP only use variables (one of the terminal set
mentioned on table 3.1) as the terminal?
is it mean TinyGP can't use constant values and 0-arity functions
(also mentioned on table 3.1) as the terminal?
if it is so, is this possible to modify TinyGP so it can use all of
the terminal set mentioned on table 3.1?
Tiny GP can use variables and random constants. The first line of the
input data file tells the program how many variables your problem has,
then it tells it how many random constants GP can use, and then it
provides two values that represent the range (min,max) from which the
random constants are drawn. (The final element if the first line is
the number of fitness cases (examples). These follow one after the
other in the remaining lines of the file).
Of course, one could modify TinyGP so that it handles 0-arity
functions. The places where you would need to make changes are the
evaluation procedure (run), the initialisation (grow) and the printing
routing (print_indiv).
I hope this helps.
Riccardo
On 14 May, 07:11, Lyani <lyani.sihomb...@gmail.com> wrote:
> it's written on section B.1 about characteristics of TinyGP, and point
> 1 said that the terminal is a user-definable number of floating point
> variables (named x1 to xn).
> is it mean that TinyGP only use variables (one of the terminal set
> mentioned on table 3.1) as the terminal?
> is it mean TinyGP can't use constant values and 0-arity functions
> (also mentioned on table 3.1) as the terminal?
> if it is so, is this possible to modify TinyGP so it can use all of
> the terminal set mentioned on table 3.1?
algorithm 3.1 (page 25) tells about Interpreter for GP, is this
algorithm have any relation with the evaluation procedure (run)?
because it is written (by a comment) that this method is interpreter.
is this algorithm will hep me to make changes the evaluation procedure
(run)?
thank you for helping me learn more clearly,
Lyani
> Tiny GP can use variables and random constants. The first line of the
> input data file tells the program how many variables your problem has,
> then it tells it how many random constants GP can use, and then it
> provides two values that represent the range (min,max) from which the
> random constants are drawn. (The final element if the first line is
> the number of fitness cases (examples). These follow one after the
> other in the remaining lines of the file).
> Of course, one could modify TinyGP so that it handles 0-arity
> functions. The places where you would need to make changes are the
> evaluation procedure (run), the initialisation (grow) and the printing
> routing (print_indiv).
> I hope this helps.
> Riccardo
> On 14 May, 07:11, Lyani <lyani.sihomb...@gmail.com> wrote:
> > hello,
> > it's written on section B.1 about characteristics of TinyGP, and point
> > 1 said that the terminal is a user-definable number of floating point
> > variables (named x1 to xn).
> > is it mean that TinyGP only use variables (one of the terminal set
> > mentioned on table 3.1) as the terminal?
> > is it mean TinyGP can't use constant values and 0-arity functions
> > (also mentioned on table 3.1) as the terminal?
> > if it is so, is this possible to modify TinyGP so it can use all of
> > the terminal set mentioned on table 3.1?
the run procedure implements a recursive interpreter. It treats every
primitive as a macro, so the primitive must itself invoke the
interpreter (run) if and when necessary. So, it is essentially the
same algorithm as the one you refer to. However, for efficiency
reasons, we keep the code that implements each primitive within the
run procedure. Also, instead of passing expressions via the stack, we
make use of an external variable which constantly points to the
expression that run needs to evaluate. The pointer is updated at the
beginning of the run procedure, so that when run is invoked again
(recursively) it knows where the expression that it needs to interpret
is located.
I hope this helps.
Riccardo
On Jun 4, 11:27 am, Lyani <lyani.sihomb...@gmail.com> wrote:
> algorithm 3.1 (page 25) tells about Interpreter for GP, is this
> algorithm have any relation with the evaluation procedure (run)?
> because it is written (by a comment) that this method is interpreter.
> is this algorithm will hep me to make changes the evaluation procedure
> (run)?
> thank you for helping me learn more clearly,
> Lyani
> > Tiny GP can use variables and random constants. The first line of the
> > input data file tells the program how many variables your problem has,
> > then it tells it how many random constants GP can use, and then it
> > provides two values that represent the range (min,max) from which the
> > random constants are drawn. (The final element if the first line is
> > the number of fitness cases (examples). These follow one after the
> > other in the remaining lines of the file).
> > Of course, one could modify TinyGP so that it handles 0-arity
> > functions. The places where you would need to make changes are the
> > evaluation procedure (run), the initialisation (grow) and the printing
> > routing (print_indiv).
> > I hope this helps.
> > Riccardo
> > On 14 May, 07:11, Lyani <lyani.sihomb...@gmail.com> wrote:
> > > hello,
> > > it's written on section B.1 about characteristics of TinyGP, and point
> > > 1 said that the terminal is a user-definable number of floating point
> > > variables (named x1 to xn).
> > > is it mean that TinyGP only use variables (one of the terminal set
> > > mentioned on table 3.1) as the terminal?
> > > is it mean TinyGP can't use constant values and 0-arity functions
> > > (also mentioned on table 3.1) as the terminal?
> > > if it is so, is this possible to modify TinyGP so it can use all of
> > > the terminal set mentioned on table 3.1?
|
