Cannot debug the optimizer

Abhishek BANERJEE

2013 2월 19

0 답변

조회 수: 5 (30일)

0 개 추천

Hi All,

I have written the following code

% Optimize EMBI Global Diversified BM
clear all;
clc;
[~,~,raw] = xlsread('U:\Matlab\SolverError.xlsx');
countryArray = raw(2:end,3:3);
isinArray =raw(2:end,1:1);
qtdPerformance = raw(2:end,17:17);
isinWeight = raw(2:end,13:13);
countryFilter = unique(countryArray);
A = zeros(size(isinArray));
v(1:120) = -1; %%min number of securities
A = diag(v);
pad = zeros(120,size(isinArray,1)-120); %%Atleast first 120 securities have non-zero value
A =  horzcat(A,pad);
a = zeros(size(A),1);
c = cell2mat(qtdPerformance);
for i = 0:size(countryFilter)-1
   evalStmt = sprintf('x%s = double(strcmp(countryArray,countryFilter(%s)));',num2str(i+1),num2str(i+1));   
   evalStmt2 = sprintf('A = vertcat(A,x%s'');',num2str(i+1));   
   countryWeight = sum(cell2mat(isinWeight(strcmp(countryArray,countryFilter(i+1))))) ;
   if countryWeight > 0.035 
       targetWeight = countryWeight + 0.05;
   else
       targetWeight = countryWeight + 0.03;
   end
   evalStmt3 = sprintf('a(%s) =  targetWeight;',num2str(i+120+1));   
   %%disp(evalStmt);
   eval(evalStmt);
   %%disp(evalStmt2);
   eval(evalStmt2);
   %%disp(evalStmt3);
   eval(evalStmt3);
end %%end for loop
disp(evalStmt);
A = vertcat(A,ones(1,size(isinArray)));
a(size(A,1))=1; %%sum of all countries should be 100 percent
lb = zeros(size(A,1),1);
ub = 0.02 * ones(size(A,1),1);
[xsol,fval,flag,output,lagMult] = linprog(c,A,a,[],[],lb,ub);

to optimize the weights of instruments in a portfolio. I get the following error after 37 iterations

Exiting: One or more of the residuals, duality gap, or total relative error has grown 100000 times greater than its minimum value so far: the dual appears to be infeasible and the primal unbounded since the primal objective < -1e+10 and the dual objective < 1e+6.

The linear program I am trying to solve is

I have a list of securities sorted by performance. I want to use this list to create an optimal portfolio.

sum of all weights < = 100% each security between 0 and 2% Atleast first 120 securities have non-zero value Atleast first 120 securities should have non zero weights Securities belonging to one country are restricted by upper limit and positive numbers.

Can someone please help me out and let me know how should I go about debugging this ?

Regards, Abhishek