% S. Boyd, et. al., "Convex Optimization of Graph Laplacian Eigenvalues"
% ICM'06 talk examples (www.stanford.edu/~boyd/cvx_opt_graph_lapl_eigs.html)
% Written for CVX by Almir Mutapcic 08/29/06
% (figures are generated)
%
% In this example we consider a graph described by the incidence matrix A.
% Each edge has a weight W_i, and we optimize various functions of the
% edge weights as described in the referenced paper; in particular,
%
% - the fastest distributed linear averaging (FDLA) problem (fdla.m)
% - the fastest mixing Markov chain (FMMC) problem (fmmc.m)
%
% Then we compare these solutions to the heuristics listed below:
%
% - maximum-degree heuristic (max_deg.m)
% - constant weights that yield fastest averaging (best_const.m)
% - Metropolis-Hastings heuristic (mh.m)

% randomly generate a graph with 50 nodes and 200 edges
% and make it pretty for plotting
n = 50; threshold = 0.2529;
rand('state',209);
xy = rand(n,2);

angle = 10*pi/180;
Rotate = [ cos(angle) sin(angle); -sin(angle) cos(angle) ];
xy = (Rotate*xy')';

Dist = zeros(n,n);
for i=1:(n-1);
  for j=i+1:n;
    Dist(i,j) = norm( xy(i,:) - xy(j,:) );
  end;
end;
Dist = Dist + Dist';
Ad = Dist < threshold;
Ad = Ad - eye(n);
m = sum(sum(Ad))/2;

% find the incidence matrix
A = zeros(n,m);
l = 0;
for i=1:(n-1);
  for j=i+1:n;
    if Ad(i,j)>0.5
      l = l + 1;
      A(i,l) =  1;
      A(j,l) = -1;
    end;
  end;
end;
A = sparse(A);

% Compute edge weights: some optimal, some based on heuristics
[n,m] = size(A);

[ w_fdla, rho_fdla ] = fdla(A);
[ w_fmmc, rho_fmmc ] = fmmc(A);
[ w_md,   rho_md   ] = max_deg(A);
[ w_bc,   rho_bc   ] = best_const(A);
[ w_mh,   rho_mh   ] = mh(A);

tau_fdla = 1/log(1/rho_fdla);
tau_fmmc = 1/log(1/rho_fmmc);
tau_md   = 1/log(1/rho_md);
tau_bc   = 1/log(1/rho_bc);
tau_mh   = 1/log(1/rho_mh);

eig_opt  = sort(eig(eye(n) - A * diag(w_fdla) * A'));
eig_fmmc = sort(eig(eye(n) - A * diag(w_fmmc) * A'));
eig_mh   = sort(eig(eye(n) - A * diag(w_mh)   * A'));
eig_md   = sort(eig(eye(n) - A * diag(w_md)   * A'));
eig_bc   = sort(eig(eye(n) - A * diag(w_bc)   * A'));

fprintf(1,'\nResults:\n');
fprintf(1,'FDLA weights:\t\t rho = %5.4f \t tau = %5.4f\n',rho_fdla,tau_fdla);
fprintf(1,'FMMC weights:\t\t rho = %5.4f \t tau = %5.4f\n',rho_fmmc,tau_fmmc);
fprintf(1,'M-H weights:\t\t rho = %5.4f \t tau = %5.4f\n',rho_mh,tau_mh);
fprintf(1,'MAX_DEG weights:\t rho = %5.4f \t tau = %5.4f\n',rho_md,tau_md);
fprintf(1,'BEST_CONST weights:\t rho = %5.4f \t tau = %5.4f\n',rho_bc,tau_bc);

% plot results
figure(1), clf
gplot(Ad,xy);
hold on;
plot(xy(:,1), xy(:,2), 'ko','LineWidth',4, 'MarkerSize',4);
axis([0.05 1.1 -0.1 0.95]);
title('Graph')
hold off;

figure(2), clf
v_fdla = [w_fdla; diag(eye(n) - A*diag(w_fdla)*A')];
[ifdla, jfdla, neg_fdla] = find( v_fdla.*(v_fdla < -0.001 ) );
v_fdla(ifdla) = [];
wbins = [-0.6:0.012:0.6];
hist(neg_fdla,wbins); hold on,
h = findobj(gca,'Type','patch');
set(h,'FaceColor','r')
hist(v_fdla,wbins); hold off,
axis([-0.6 0.6 0 12]);
xlabel('optimal FDLA weights');
ylabel('histogram');

figure(3), clf
xbins = (-1:0.015:1)';
ymax  = 6;
subplot(3,1,1)
hist(eig_md, xbins); hold on;
max_md = max(abs(eig_md(1:n-1)));
plot([-max_md -max_md],[0 ymax], 'b--');
plot([ max_md  max_md],[0 ymax], 'b--');
axis([-1 1 0 ymax]);
text(0,5,'MAX DEG');
title('Eigenvalue distributions')
subplot(3,1,2)
hist(eig_bc, xbins); hold on;
max_opt = max(abs(eig_bc(1:n-1)));
plot([-max_opt -max_opt],[0 ymax], 'b--');
plot([ max_opt  max_opt],[0 ymax], 'b--');
axis([-1 1 0 ymax]);
text(0,5,'BEST CONST');
subplot(3,1,3)
hist(eig_opt, xbins); hold on;
max_opt = max(abs(eig_opt(1:n-1)));
plot([-max_opt -max_opt],[0 ymax], 'b--');
plot([ max_opt  max_opt],[0 ymax], 'b--');
axis([-1 1 0 ymax]);
text(0,5,'FDLA');

figure(4), clf
xbins = (-1:0.015:1)';
ymax  = 6;
subplot(3,1,1)
hist(eig_md, xbins); hold on;
max_md = max(abs(eig_md(1:n-1)));
plot([-max_md -max_md],[0 ymax], 'b--');
plot([ max_md  max_md],[0 ymax], 'b--');
axis([-1 1 0 ymax]);
text(0,5,'MAX DEG');
title('Eigenvalue distributions')
subplot(3,1,2)
hist(eig_mh, xbins); hold on;
max_opt = max(abs(eig_mh(1:n-1)));
plot([-max_opt -max_opt],[0 ymax], 'b--');
plot([ max_opt  max_opt],[0 ymax], 'b--');
axis([-1 1 0 ymax]);
text(0,5,'MH');
subplot(3,1,3)
hist(eig_fmmc, xbins); hold on;
max_opt = max(abs(eig_fmmc(1:n-1)));
plot([-max_opt -max_opt],[0 ymax], 'b--');
plot([ max_opt  max_opt],[0 ymax], 'b--');
axis([-1 1 0 ymax]);
text(0,5,'FMMC');

figure(5), clf
v_fmmc = [w_fmmc; diag(eye(n) - A*diag(w_fmmc)*A')];
[ifmmc, jfmmc, nonzero_fmmc] = find( v_fmmc.*(v_fmmc > 0.001 ) );
hist(nonzero_fmmc,80);
axis([0 1 0 10]);
xlabel('optimal positive FMMC weights');
ylabel('histogram');

figure(6), clf
An = abs(A*diag(w_fmmc)*A');
An = (An - diag(diag(An))) > 0.0001;
gplot(An,xy,'b-'); hold on;
h = findobj(gca,'Type','line');
set(h,'LineWidth',2.5)
gplot(Ad,xy,'b:');
plot(xy(:,1), xy(:,2), 'ko','LineWidth',4, 'MarkerSize',4);
axis([0.05 1.1 -0.1 0.95]);
title('Subgraph with positive transition prob.')
hold off;
 
Calling sedumi: 2551 variables, 2350 equality constraints
------------------------------------------------------------
SeDuMi 1.21 by AdvOL, 2005-2008 and Jos F. Sturm, 1998-2003.
Alg = 2: xz-corrector, Adaptive Step-Differentiation, theta = 0.250, beta = 0.500
Split 1 free variables
eqs m = 2350, order n = 103, dim = 5003, blocks = 3
nnz(A) = 3203 + 0, nnz(ADA) = 3318750, nnz(L) = 1660550
 it :     b*y       gap    delta  rate   t/tP*  t/tD*   feas cg cg  prec
  0 :            2.96E-01 0.000
  1 :   3.60E+00 2.38E-02 0.000 0.0804 0.9900 0.9900  -0.22  1  1  8.8E-01
  2 :   1.12E+00 9.42E-03 0.000 0.3954 0.9000 0.9000   3.06  1  1  1.5E-01
  3 :   9.43E-01 3.44E-03 0.000 0.3655 0.9000 0.9000   1.86  1  1  4.1E-02
  4 :   9.27E-01 1.01E-03 0.000 0.2938 0.9000 0.9000   1.09  1  1  1.2E-02
  5 :   9.08E-01 3.29E-04 0.000 0.3250 0.9000 0.9000   1.06  1  1  3.9E-03
  6 :   9.04E-01 1.05E-04 0.000 0.3193 0.9000 0.9000   1.02  1  1  1.3E-03
  7 :   9.03E-01 2.58E-05 0.000 0.2460 0.9022 0.9000   1.00  1  1  3.4E-04
  8 :   9.02E-01 4.63E-06 0.000 0.1796 0.9074 0.9000   1.00  1  1  9.1E-05
  9 :   9.02E-01 5.57E-07 0.000 0.1201 0.9187 0.9000   1.00  1  1  2.3E-05
 10 :   9.02E-01 1.12E-07 0.000 0.2011 0.9160 0.9000   1.00  1  1  5.4E-06
 11 :   9.02E-01 2.56E-08 0.000 0.2284 0.9055 0.9000   1.00  1  1  1.3E-06
 12 :   9.02E-01 5.43E-09 0.000 0.2123 0.9054 0.9000   1.00  1  1  2.7E-07
 13 :   9.02E-01 1.49E-09 0.000 0.2756 0.9063 0.9000   1.00  1  1  7.6E-08
 14 :   9.02E-01 3.41E-10 0.000 0.2279 0.9027 0.9000   1.00  1  1  1.7E-08
 15 :   9.02E-01 6.99E-11 0.000 0.2051 0.9000 0.9031   1.00  1  2  3.6E-09

iter seconds digits       c*x               b*y
 15     20.0   Inf  9.0207867670e-01  9.0207869224e-01
|Ax-b| =   3.2e-09, [Ay-c]_+ =   1.5E-09, |x|=  1.1e+01, |y|=  1.2e+00

Detailed timing (sec)
   Pre          IPM          Post
3.300E-01    2.003E+01    1.000E-02    
Max-norms: ||b||=9.400000e-01, ||c|| = 1,
Cholesky |add|=0, |skip| = 0, ||L.L|| = 54.6293.
------------------------------------------------------------
Status: Solved
Optimal value (cvx_optval): +0.902079
 
Calling sedumi: 2801 variables, 2600 equality constraints
------------------------------------------------------------
SeDuMi 1.21 by AdvOL, 2005-2008 and Jos F. Sturm, 1998-2003.
Alg = 2: xz-corrector, Adaptive Step-Differentiation, theta = 0.250, beta = 0.500
Split 1 free variables
eqs m = 2600, order n = 353, dim = 5253, blocks = 3
nnz(A) = 3702 + 0, nnz(ADA) = 3918898, nnz(L) = 1961925
 it :     b*y       gap    delta  rate   t/tP*  t/tD*   feas cg cg  prec
  0 :            8.59E-02 0.000
  1 :   1.13E+00 5.63E-02 0.000 0.6559 0.9000 0.9000   5.60  1  1  1.7E+00
  2 :   1.06E+00 2.85E-02 0.000 0.5056 0.9000 0.9000   1.54  1  1  8.3E-01
  3 :   9.31E-01 8.13E-03 0.000 0.2855 0.9000 0.9000   1.80  1  1  1.6E-01
  4 :   9.53E-01 1.71E-03 0.000 0.2103 0.9000 0.9000   1.48  1  1  2.7E-02
  5 :   9.39E-01 4.28E-04 0.000 0.2503 0.9000 0.9000   1.12  1  1  6.5E-03
  6 :   9.29E-01 2.20E-04 0.000 0.5135 0.9000 0.9000   1.06  1  1  3.3E-03
  7 :   9.24E-01 1.24E-04 0.000 0.5648 0.9000 0.9000   1.04  1  1  1.9E-03
  8 :   9.24E-01 3.10E-05 0.383 0.2500 0.9000 0.0000   1.03  1  1  1.4E-03
  9 :   9.22E-01 7.28E-06 0.012 0.2347 0.9480 0.9000   1.02  1  1  4.9E-04
 10 :   9.18E-01 2.65E-06 0.000 0.3643 0.9000 0.9071   1.02  1  1  1.8E-04
 11 :   9.17E-01 1.39E-06 0.000 0.5231 0.9255 0.9000   1.01  1  1  9.1E-05
 12 :   9.16E-01 8.29E-07 0.000 0.5969 0.9468 0.9000   1.01  1  1  5.3E-05
 13 :   9.16E-01 4.24E-07 0.225 0.5120 0.0000 0.9000   1.00  1  1  2.9E-05
 14 :   9.15E-01 1.45E-07 0.000 0.3429 0.9000 0.9098   1.00  1  1  1.0E-05
 15 :   9.15E-01 5.76E-08 0.000 0.3957 0.8138 0.9000   1.00  1  1  4.2E-06
 16 :   9.15E-01 1.96E-08 0.000 0.3407 0.9000 0.8329   1.00  1  1  1.4E-06
 17 :   9.15E-01 6.23E-09 0.000 0.3176 0.8328 0.9000   1.00  1  1  4.5E-07
 18 :   9.15E-01 1.51E-09 0.000 0.2427 0.9000 0.9000   1.00  1  1  1.1E-07
 19 :   9.15E-01 1.17E-10 0.000 0.0774 0.9900 0.9900   1.00  1  2  8.5E-09

iter seconds digits       c*x               b*y
 19     35.0   Inf  9.1515175405e-01  9.1515175583e-01
|Ax-b| =   6.9e-09, [Ay-c]_+ =   4.9E-09, |x|=  1.1e+01, |y|=  1.2e+00

Detailed timing (sec)
   Pre          IPM          Post
4.200E-01    3.498E+01    2.000E-02    
Max-norms: ||b||=9.600000e-01, ||c|| = 1,
Cholesky |add|=0, |skip| = 0, ||L.L|| = 81.924.
------------------------------------------------------------
Status: Solved
Optimal value (cvx_optval): +0.915152

Results:
FDLA weights:		 rho = 0.9021 	 tau = 9.7037
FMMC weights:		 rho = 0.9152 	 tau = 11.2784
M-H weights:		 rho = 0.9489 	 tau = 19.0839
MAX_DEG weights:	 rho = 0.9706 	 tau = 33.5236
BEST_CONST weights:	 rho = 0.9470 	 tau = 18.3549