Dynamic-Calibration/utils/YALMIP-master/operators/sumabsk.m

74 lines
2.8 KiB
Mathematica
Raw Permalink Normal View History

2019-12-18 11:25:45 +00:00
function varargout = sumabsk(varargin)
% SUMABSK Returns sum of k largest (by magnitude) (eigen-)values.
%
% s = SUMABSK(X,k)
%
% For a real vector X, SUMABSK returns the sum of the k largest (by magnitude) elements.
%
% For an Hermitian matrix X, SUMABSK returns the sum of the k largest (by magnitude) eigen-values.
%
% See also SUMK
% ***************************************************
% This file defines a nonlinear operator for YALMIP
%
% It can take three different inputs
% For double inputs, it returns standard double values
% For sdpvar inputs, it genreates a an internal variable
% When first input is 'model' it generates the epigraph
%
% ***************************************************
switch class(varargin{1})
case 'double' % What is the numerical value of this argument (needed for displays etc)
if nargin == 1
error('sumabsk needs two arguments');
else
X = varargin{1};
[n,m] = size(X);
if (min(n,m) > 1 & ~ishermitian(X)) | (n~=m & ~isreal(X))
error('sumabsk can only be applied on real vectors and Hermitian matrices');
else
k = min(length(X),varargin{2});
if min(n,m)==1
sorted = sort(abs(X));
else
sorted = sort(abs(eig(X)));
end
varargout{1} = sum(sorted(max(1,end-k+1):end));
end
end
case 'sdpvar' % Overloaded operator for SDPVAR objects. Pass on args and save them.
X = varargin{1};
[n,m] = size(X);
if (min(n,m) > 1 & ~ishermitian(X)) | (n~=m & ~isreal(X))
error('sumabsk can only be applied on real vectors and Hermitian matrices');
else
if nargin < 2
error('sumabsk needs two arguments');
else
varargout{1} = yalmip('define',mfilename,varargin{:});
end
end
case 'char' % YALMIP sends 'model' when it wants the epigraph or hypograph
if isequal(varargin{1},'graph')
t = varargin{2}; % Second arg is the extended operator variable
X = varargin{3}; % Third arg and above are the args user used when defining t.
k = min(varargin{4},length(X));
[n,m] = size(X);
Z = sdpvar(n,m);
s = sdpvar(1,1);
if min(n,m)==1
varargout{1} = (t-k*s-sum(Z) >= 0) + (Z >= 0) + (Z+s >= X >= -Z-s);
else
varargout{1} = (t-k*s-trace(Z) >= 0) + (Z >= 0) + (Z+s*eye(n) >= X >= -Z-s*eye(n));
end
varargout{2} = struct('convexity','convex','monotonicity','none','definiteness','none','model','graph');
varargout{3} = X;
else
end
otherwise
end