I find that the "bounds" takes no roles in "proc model" or programming. The following is an example. I want to get the positive solution. But it gives the negative one. Dose anybody has any suggestion?
Notes: (1)You can impose the initial value of x=5. But in general, you do not know the solution and so it is very difficutlt to give a close initial values.
(2)You can replace x by exp(x). But it makes the computing more difficult and sometimes can not get the solution.
data aaaa;
b2=1; b1=-4; b3=-5;
run;
proc model data=aaaa out=ab_1 noprint;
bounds x>0;
dependent x 1;
eq.eq1=b2*x**2+b1*x+b3;
solve x /converge=1e-6 maxiter=200;
run;
Message was edited by: HHBB