|
|
Re: generation of random multiset permutation with restrictions
Posted:
Aug 23, 2012 6:22 AM
|
|
"Bruno Luong" <b.luong@fogale.findmycountry> wrote in message <k103kl$g6d$1@newscl01ah.mathworks.com>...
> I will change the generate random permutation with restrictio. D will be different than my first post, but the goal of generaion is respected.
>
> a = [1,1,1,2,2,3,3,3]
> s = {[3],[1,2],[1,2,3],[1,2,3],[1,2,3],[1,2,3],[2,3],[2,3]};
>
> %% Covert cell of sets to boolean array
> m = max(a(:));
> n = length(a);
> j = arrayfun(@(j) j+zeros(size(s{j})), 1:n, 'unif', 0);
> sbool = accumarray([s{:}; j{:}]',1, [m n])
> clear m n j
>
> %% Generate 100 vectors p, a permutation of a, and respecting the set-constraints
> % p is ranged in 100 x 8 arrays, working with a and sbool
> [m n] = size(sbool); % 3 x 8
> [~, ia] = sort(rand(100,n),2);
> p = a(ia);
> c = m*(0:n-1)';
> pp = permute(p, [3 2 1]);
> D = 1-sbool(bsxfun(@plus,pp,c));
> for k=1:size(p,1) % 100
> [i cost] = assignmentoptimal(D(:,:,k));
> if any(i==0) || cost>0
> error('No solution exists');
> end
> p(k,:) = p(k,i);
> end
> disp(p)
>
>
> %% Some p
> p1 = [3,1,1,1,2,2,3,3];
> p2 = [1,3,1,1,2,2,3,3];
> p = [p1; p2]
> clear p1 p2
>
> % Check the validity of p
> % working with a and sbool
> [m n] = size(sbool);
> OK = @(p) all(sbool(bsxfun(@plus,p,m*(0:n-1))) & bsxfun(@eq,sort(p,2),sort(a)),2);
> OK(p)
>
> % Bruno
Thank you, this is good method.
|
|
