benchmark silhouette.m

% Benchmark silhouette: Blocked vs Original implementation
addpath('inst');

dims = 10; k = 5; trials = 3;
sizes = [5 10 50 100 200 350 500 600 750 900 1000 1250 1500 2000 3000 4000 5000 6000 7000 8000 10000];

printf('\n%7s %10s %10s %8s %8s\n', 'N', 'Original', 'Blocked', 'Ratio', 'Winner');
printf('%s\n', repmat('-', 1, 50));

results = [];
for n = sizes
  if (n*n*8 > 8e9), printf('%7d  SKIPPED (>8GB)\n', n); continue; endif

  rand('seed', 12345); X = rand(n, dims); clust = randi(k, n, 1);
  clusterIDs = unique(clust); m = length(clusterIDs);

  % ORIGINAL: pdist + squareform + full matrix loop
  t_orig = [];
  for t = 1:(trials+1)
    tic;
    distMatrix = squareform(pdist(X, 'squaredeuclidean'));
    si_o = zeros(n, 1);
    for i = 1:n
      grp = {}; for j = 1:m, grp{clusterIDs(j)} = distMatrix(i, clust == clusterIDs(j)); endfor
      if length(grp{clust(i)}) == 1, si_o(i) = 1; continue; endif
      a_i = sum(grp{clust(i)}) / (length(grp{clust(i)}) - 1);
      others = clusterIDs(clusterIDs != clust(i));
      b_i = min(arrayfun(@(c) mean(grp{c}), others));
      si_o(i) = (b_i - a_i) / max(a_i, b_i);
    endfor
    t_orig(end+1) = toc;
  endfor

  % BLOCKED: row-by-row computation
  t_blk = [];
  for t = 1:(trials+1)
    tic; si_b = silhouette(X, clust, 'squaredeuclidean', 'DoNotPlot'); t_blk(end+1) = toc;
  endfor

  t1 = median(t_orig(2:end)); t2 = median(t_blk(2:end));
  assert(max(abs(si_o - si_b)) < 1e-10, 'Results mismatch!');

  ratio = t1/t2; winner = ifelse(ratio > 1, 'Blocked', 'Original');
  printf('%7d %9.4fs %9.4fs %7.2fx %8s\n', n, t1, t2, ratio, winner);
  results = [results; n t1 t2 ratio];
endfor

% Save CSV
csvwrite('benchmark_silhouette_results.csv', results);
printf('\nResults saved to benchmark_silhouette_results.csv\n');

% Plot
figure('Position', [100 100 900 500]);
loglog(results(:,1), results(:,2), 'b-o', 'LineWidth', 2, 'MarkerSize', 8, 'MarkerFaceColor', 'b');
hold on;
loglog(results(:,1), results(:,3), 'r-s', 'LineWidth', 2, 'MarkerSize', 8, 'MarkerFaceColor', 'r');
grid on; set(gca, 'FontSize', 12, 'LineWidth', 1);
xlabel('N (number of points)', 'FontSize', 14, 'FontWeight', 'bold');
ylabel('Time (seconds)', 'FontSize', 14, 'FontWeight', 'bold');
title('Silhouette Performance: Original vs Blocked', 'FontSize', 16, 'FontWeight', 'bold');
legend({'Original (pdist+squareform)', 'Blocked (row-by-row)'}, 'Location', 'northwest', 'FontSize', 12);
set(gca, 'XTick', [10 100 1000 10000], 'XTickLabel', {'10', '100', '1K', '10K'});
print('-dpng', '-r150', 'benchmark_silhouette_plot.png');
printf('Plot saved to benchmark_silhouette_plot.png\n');

% Summary
if all(results(:,4) > 1)
  printf('\n✓ Blocked is faster for ALL sizes tested - no threshold needed\n');
else
  idx = find(results(:,4) < 1, 1, 'last');
  if ~isempty(idx), printf('\n⚡ Crossover near N=%d\n', results(idx,1)); endif
endif