123 lines
3.4 KiB
Matlab
Executable File
123 lines
3.4 KiB
Matlab
Executable File
function y = vertcat(varargin)
|
|
%VERTCAT (overloaded)
|
|
|
|
prenargin = nargin;
|
|
% Fast exit
|
|
if prenargin<2
|
|
y=varargin{1};
|
|
return
|
|
end
|
|
|
|
% Get dimensions
|
|
n = zeros(prenargin,1);
|
|
m = zeros(prenargin,1);
|
|
for i = 1:prenargin
|
|
if isa(varargin{i},'blkvar')
|
|
varargin{i} = sdpvar(varargin{i});
|
|
end
|
|
[n(i),m(i)]=size(varargin{i});
|
|
end
|
|
|
|
% Keep only non-empty
|
|
keep_these = find((n.*m)~=0);
|
|
if length(keep_these)<length(n)
|
|
varargin = {varargin{keep_these}};
|
|
n = n(keep_these);
|
|
m = m(keep_these);
|
|
end;
|
|
|
|
% All widths should be equal
|
|
if any(m~=m(1))
|
|
error('All rows in the bracketed expression must have the same number of columns.');
|
|
end
|
|
|
|
nblocks = size(varargin,2);
|
|
isasdpvar = zeros(nblocks,1);
|
|
% Get dimensions
|
|
for i = 1:nblocks
|
|
isasdpvar(i) = isa(varargin{i},'sdpvar');
|
|
end
|
|
|
|
% Find all free variables used
|
|
all_lmi_variables = [];
|
|
for i = 1:nblocks
|
|
if isasdpvar(i)
|
|
all_lmi_variables = [all_lmi_variables varargin{i}.lmi_variables];
|
|
end
|
|
end
|
|
all_lmi_variables = uniquestripped(all_lmi_variables);
|
|
|
|
% Pick one of the sdpvar objects to build on...
|
|
y = varargin{min(find(isasdpvar))};
|
|
|
|
% Some indexation tricks...
|
|
basis_i = [];
|
|
basis_j = [];
|
|
basis_s = [];
|
|
m = m(1);
|
|
sumn=sum(n);
|
|
shift=[0;cumsum(n)];
|
|
for j = 1:nblocks
|
|
if isasdpvar(j)
|
|
if length(all_lmi_variables)==length(varargin{j}.lmi_variables) && all_lmi_variables(1)==varargin{j}.lmi_variables(1) && all_lmi_variables(end)==varargin{j}.lmi_variables(end)
|
|
% Avoid call to ismember and find
|
|
in_this = 1:length(all_lmi_variables);
|
|
else
|
|
members = ismembcYALMIP(all_lmi_variables,varargin{j}.lmi_variables);
|
|
in_this = find(members);
|
|
end
|
|
dummy = [1 1+in_this];
|
|
[i2,j2,s2] = find(varargin{j}.basis);
|
|
j2 = dummy(j2);
|
|
else
|
|
[i2,j2,s2] = find(varargin{j}(:));
|
|
end
|
|
if ~isempty(j2)
|
|
col = 1+floor((i2-1)/n(j));
|
|
row = i2-(col-1)*n(j);
|
|
basis_i = [basis_i;row(:)+shift(j)+sumn*(col(:)-1)];
|
|
basis_j = [basis_j;j2(:)];
|
|
basis_s = [basis_s;s2(:)];
|
|
end
|
|
end
|
|
basis = sparse(basis_i,basis_j,basis_s,sum(n)*m,1+length(all_lmi_variables));
|
|
|
|
y.dim(1) = sum(n);
|
|
y.dim(2) = m;
|
|
y.basis = basis;
|
|
y.lmi_variables = all_lmi_variables;
|
|
% Reset info about conic terms
|
|
y.conicinfo = [0 0];
|
|
y.extra.opname='';
|
|
y.extra.createTime = definecreationtime;
|
|
y = unfactor(y);
|
|
|
|
% Update the factors
|
|
% But first, check to see that factors exist in all terms, if not simply
|
|
% exit
|
|
for i = 1:length(varargin)
|
|
if isa(varargin{i},'sdpvar')
|
|
if length(varargin{i}.leftfactors)==0
|
|
y = flush(y);
|
|
return
|
|
end
|
|
end
|
|
end
|
|
doublehere = [];
|
|
for i = 1:length(varargin)
|
|
if isa(varargin{i},'sdpvar')
|
|
for j = 1:length(varargin{i}.leftfactors)
|
|
w = size(varargin{i}.leftfactors{j},2);
|
|
y.leftfactors{end+1} = [spalloc(sum(n(1:1:i-1)),w,0); varargin{i}.leftfactors{j}; spalloc(sum(n(i+1:1:end)),w,0)];
|
|
y.midfactors{end+1} = varargin{i}.midfactors{j};
|
|
y.rightfactors{end+1} = varargin{i}.rightfactors{j};
|
|
end
|
|
elseif isnumeric(varargin{i})
|
|
here = length(y.midfactors)+1;
|
|
doublehere = [doublehere here];
|
|
y.leftfactors{here} = [spalloc(sum(n(1:1:i-1)),size(varargin{i},1),0); speye(size(varargin{i},1)); spalloc(sum(n(i+1:1:end)),size(varargin{i},1),0)];
|
|
y.midfactors{here} = varargin{i};
|
|
y.rightfactors{here} = speye(size(varargin{i},2));
|
|
end
|
|
end
|
|
y = cleandoublefactors(y); |