preprocess_bilinear_bounds.m

matlab波形优化算法经常要用到的matlab toolbox工具箱:yalmip

function p = preprocess_bilinear_bounds(p)

if ~isempty(p.integer_variables)
    for i = 1:size(p.bilinears,1)
        if ismember(p.bilinears(i,2),p.integer_variables)
            if ismember(p.bilinears(i,3),p.integer_variables)
                p.integer_variables = [p.integer_variables p.bilinears(i,1)];
            end
        end
    end
    if p.K.f > 0
        for i = 1:p.K.f
            if all(p.F_struc(i,:) == fix(p.F_struc(i,:)))
                involved = find(p.F_struc(i,2:end));
                if nnz(ismember(involved,p.integer_variables)) == length(involved)-1
                    p.integer_variables = [p.integer_variables involved];
                end
            end
        end
        p.integer_variables  = unique(p.integer_variables );
    end
end

if ~isempty(p.binary_variables)
    for i = 1:size(p.bilinears,1)
        if ismember(p.bilinears(i,2),p.binary_variables)
            if ismember(p.bilinears(i,3),p.binary_variables)
                p.binary_variables = [p.binary_variables p.bilinears(i,1)];
            end
        end
    end
    for i = 1:p.K.f
        if all(p.F_struc(i,:) == fix(p.F_struc(i,:)))
            involved = find(p.F_struc(i,2:end));
            if nnz(ismember(involved,p.binary_variables)) == length(involved)-1
                p.binary_variables = [p.binary_variables involved];
            end
        end
    end
    p.binary_variables  = unique(p.binary_variables );

end

if isempty(p.ub)
    p.ub = repmat(inf,length(p.c),1);
end
if isempty(p.lb)
    p.lb = repmat(-inf,length(p.c),1);
end
if ~isempty(p.F_struc)
    [lb,ub,used_rows_eq,used_rows_lp] = findulb(p.F_struc,p.K);
    if ~isempty([used_rows_eq;used_rows_lp])
        lower_defined = find(~isinf(lb));
        if ~isempty(lower_defined)
            p.lb(lower_defined) = max(p.lb(lower_defined),lb(lower_defined));
        end
        upper_defined = find(~isinf(ub));
        if ~isempty(upper_defined)
            p.ub(upper_defined) = min(p.ub(upper_defined),ub(upper_defined));
        end
        % Remove linear bound inequalities
        if ~isempty(used_rows_lp)
            used_rows_lp = used_rows_lp(find(~any(p.F_struc(p.K.f+used_rows_lp,1+p.nonlinears),2)));
            not_used_rows = setdiff(1:p.K.l,used_rows_lp);
            for i = 1:length(p.KCut.l)
                p.KCut.l(i) = find(not_used_rows == p.KCut.l(i));
               % p.originalModel.KCut.l(i) = find(not_used_rows == p.originalModel.KCut.l(i) );
            end
            if ~isempty(used_rows_lp)
                p.F_struc(p.K.f+used_rows_lp,:)=[];
              %  p.originalModel.F_struc(p.originalModel.K.f+used_rows_lp,:)=[];
                p.K.l = p.K.l - length(used_rows_lp);
               % p.originalModel.K.l = p.originalModel.K.l - length(used_rows_lp);
            end
        end
        % Remove linear bound inequalities
        if ~isempty(used_rows_eq)
            used_rows_eq = used_rows_eq(find(~any(p.F_struc(used_rows_eq,1+p.nonlinears),2)));                       
            not_used_rows = setdiff(1:p.K.f,used_rows_eq);
            for i = 1:length(p.KCut.f)
                p.KCut.f(i) = find(not_used_rows==p.KCut.f(i));
              %  p.originalModel.KCut.f(i) = find(not_used_rows==p.originalModel.KCut.f(i));
            end
            if ~isempty(used_rows_eq)
                p.F_struc(used_rows_eq,:)=[];
              %  p.originalModel.F_struc(used_rows_eq,:)=[];
                p.K.f = p.K.f - length(used_rows_eq);
              %  p.originalModel.K.f = p.originalModel.K.f - length(used_rows_eq);
            end
        end
    end
end
p.lb(p.binary_variables) = max(0,p.lb(p.binary_variables));
p.ub(p.binary_variables) = min(1,p.ub(p.binary_variables));
p.lb(p.integer_variables) = ceil(p.lb(p.integer_variables));
p.ub(p.integer_variables) = floor(p.ub(p.integer_variables));
p = clean_bounds(p);
if ~isempty(p.bilinears)
    p = updatemonomialbounds(p);
end