I have long symbolic function in Maxima, let say
fn(x,y):=<<some long equation using x and y>>
I would like to calculate polynomial approximation of this function, let say
fn_poly(x,y)
within known range of x and y and with maximum error e
I know, that there is a funcionality in Maxima, e.g. plsquares, but it needs a matrix on input and I have only function fn(x,y). I don't know how to generate this matrix from my function. genmatrix creates matrix not usable by plsquares.
Is this possible in Maxima?
Make list of lists and transform it to matrix.
load(plsquares);
f(x,y):=x^2+y^3;
mat:makelist(makelist([X,Y,f(X,Y)],X,1,10,2),Y,1,10,2);
-> [[[1,1,2],[3,1,10],[5,1,26],[7,1,50],[9,1,82]],[[1,3,28],[3,3,36],[5,3,52],[7,3,76],[9,3,108]],[[1,5,126],[3,5,134],[5,5,150],[7,5,174],[9,5,206]],[[1,7,344],[3,7,352],[5,7,368],[7,7,392],[9,7,424]],[[1,9,730],[3,9,738],[5,9,754],[7,9,778],[9,9,810]]]
mat2:[];
for i:1 thru length(mat) do mat2:append(mat2,mat[i]);
mat3:funmake('matrix,mat2);
-> matrix([1,1,2],[3,1,10],[5,1,26],[7,1,50],[9,1,82],[1,3,28],[3,3,36],[5,3,52],[7,3,76],[9,3,108],[1,5,126],[3,5,134],[5,5,150],[7,5,174],[9,5,206],[1,7,344],[3,7,352],[5,7,368],[7,7,392],[9,7,424],[1,9,730],[3,9,738],[5,9,754],[7,9,778],[9,9,810])
ZZ:rhs(plsquares(mat3,[X,Y,Z],Z,3,3));
-> Determination Coefficient for Z = 1.0
-> Y^3+X^2