comp.soft-sys.matlab - The MathWorks calculation and visualization package.
Peter, Thanks a lot, Best Nick On Thu, 16 Sep 2004 14:14:40 +0000 (UTC), Peter Spellucci wrote: > >In article <gqthrnb8du8v@legacy>, > XXXX@XXXXX.COM (Nick) writes: > >Hello, > >I receive a strange result when I try to use the lsqlin function: > >"Optimization terminated: relative function value changing by less > > than sqrt(OPTIONS.TolFun), and rate of progress is slow." > >I played around with the 'TolFun' but always end up receiving the same > >exitflag. > >Does it mean the regression is not valid ? > >How could I change it and make it work ? > >Many thanks, > >Best > > > >Nick > > >this can have two reasons: >your quadratic function is not uniformly convex of the Hessian extremely >illconditioned. then in minimizing in a terribly flat valley the code >does not have progress. hence test cond(C). >next , there may be redundant active constraints and a cycling caused by that. >check the output point for the active constraints and test rank of the >matrix of gradients (including active bounds). >if you detect rank deficiency, try to delete the redundant constraints >(this is not so easy, there exists software for that, but not in the toolbox) >hth >peter
Hello, Can neone pls tell me how to get the number of function evaluations done when I use 'lsqlin', its there in 'fmincon' but most probably not in lsqlin ! thanks
I am trying to solve a Ax=b problem, A is 2x2, x and b is 2x1 Specifically, I have A = Mx, b = RHS , and I used some constraint. I used the function : lsqlin(Mx,RHS,A,b,Aeq,beq,lb,ub); I got the following warning message : Warning: Large-scale method can handle bound constraints only; using medium-scale method instead. I am not sure what this means. What should I modify to make things right ?
Hi, I keep encounter a general question using fmincon. I expect to get a exitflag 1 indicating a sufficient small first-order optimality measure. However, most of the time, when the algorithm terminates, I get exitflag 5 and a pretty big first-order optimality measure. To my understanding, exitflag 5 means there is not much decrease can be obtained along the search direction by a reasonable step. Should this mean a small optimality measure as well? This really bothers me. Any help will be great! thanks, Ming