Thank for your great job in practical BNB implementation.
I wonder does the method of B&B can always find the "optimal value" of Convex Mixed Integer Nonlinear Programming? Is there any reference or prove to above statement?
I had to dig into the code to get this sucker to work (sans GUI). Below is a matlab file for a function of a single variable in FuncEval with nonlinears in constraintEval. Also, note that you will get an error when it's done. You can either ignore it, or comment out the very last line of BNB20.m . If your function requires separate inputs other than x, put those in the call to BNB20 after options.
%% BNB Call
% Initialize x0, Column Vector
x0 = [1];
% BNB variable types
% Continuous for xstatus(i)=0
% Integer for xstatus(i)= 1
% Fixed for xstatus(i)=2
xstat = [1];
% this problem has no inequality constraints
A = [];
b = [];
% this problem has no equality constraints
Aeq = [];
beq = [];
% Given Bounds in problem statement
lb = [ ...
0
];
ub = [ ...
31
];
% Set settings
setts = []; % Leave as default
% Set options
options = optimset('Display','iter','TolX',1e-4,'maxSQPiter',Inf)
%% Run the BNB Code!
[errmsg,Z,X,t,c,fail] = BNB20('FuncEval',x0, ...
xstat,lb,ub,A,b,Aeq,beq,'constraintEval',setts,options)
I'm trying to run a mixed integer nonlinear programming (MINLP) problem using BNB20. However, irrespective of the complexity of my algorithm, I'm always getting the following results. It would be great if anyone can explain this result.
*** searched 0 % of tree
*** Z : Inf
*** t : 0.0 secs
*** c : 0 cycles
*** fail : 0 cycles
Optimization terminated: first-order optimality measure less
than options.TolFun and maximum constraint violation is less
than options.TolCon.
Active inequalities (to within options.TolCon = 1e-006):
lower upper ineqlin ineqnonlin
11 31
12 32
13 33
34
35
36
37
38
39
40
41
42
43
44
45
??? Error using ==> lt
Matrix dimensions must agree.
Error in ==> BNB20 at 247
elseif all(abs(round(x(K))-x(K))<errx) & convflag>0
I have a problem with this algorithm I start it like that
[errmsg,Z,X,t,c,fail] = BNB20(fun,para',[0 0 1 1 1 1]',[150 125 3 2 11 1]',ub',[],[],[],[],'problemconstraints',[],options,dmodel,SampMin,weight,constraints)
yet one of the parameters is assigned a negative value during the operation of the algorithm
27 Sep 2008
liu bingbing
very good!
16 Sep 2008
Luo chengjin
I think it's good!
06 Sep 2008
Guo Xiaoyu
Thank you very mucn
05 Sep 2008
Sun Qiyue
I can not work it out? the example is not so good and I always get the error: fun caused error
09 Jul 2008
Yubin Bao
easy to use
08 Jun 2008
mark elec
Does the algorithm support for non-convex problem ???
10 Apr 2008
jacky LI
the example is not good enough!
30 Jun 2007
wang lingli
study
17 Apr 2007
Mojtaba Hajihasani
Your proposed method suffers from big problem, beacuse in sub-problem you have brought the problem into continues domain.
27 Dec 2006
asd sad
10 Nov 2006
Dingyu Xue
Further suggestions
1) better to allow function handles
2) fine-tuning of the argument and results for integer variables:
2a. in the starting part of the function, give xlb(xstat)=ceil(xlb(xstat)); xub(xstat)=floor(xub(xstat))
2b. at the end of the function, X(xstat)=round(X(xstat))
25 Aug 2006
Naresh Nandola
02 Aug 2006
Puthut Wibowo
Hmmm......
15 Nov 2005
Bess Bai
Thanks! is there the manual for bnb.m (matlab) file?
04 Sep 2005
kinky leung
Thanks!But is it free?
21 Mar 2005
shrinivas pradeep
sir,
can you please send the manual for bnb.m (matlab) file.
04 Mar 2005
Pradeep Shrinivas
Dear sir,
Your work is excellent, we are using your work for our project.
Regards,
Pradeep &Shrinivas
( BE (CSE)Final Year)
BCET Bangalore-560081
29 Dec 2004
Bakhtiyar Tokpayev
31 Jul 2004
vinod kumar
please provide pdf help files and pseudo codes and comments on each step.
29 Jun 2004
Andrew Odjo
02 Jul 2003
Atanas Chervenkov
01 Jul 2003
Nguyen Quoc Vo
03 Dec 2002
H hasper
19 May 2002
George Pau
Very good basic implementation of branch and bound.