added second plot - pairwise distances

plot1.m

@@ -7,10 +7,14 @@ simHists = simHists(endsWith({simHists.name}, 'miSimHist.mat'));
 assert(length(simHists) == length(simInits), "input data availability mismatch");
 
 % Initialize plotting data
-Cfinal = NaN(12, 1);
+nRuns = length(simHists);
-n = NaN(12, 1);
+Cfinal = NaN(nRuns, 1);
-doubleIntegrator = NaN(12, 1);
+n = NaN(nRuns, 1);
-numObjective = NaN(12, 1);
+doubleIntegrator = NaN(nRuns, 1);
+numObjective = NaN(nRuns, 1);
+positions = cell(6, nRuns);
+commsRadius = NaN(nRuns, 1);
+collisionRadius = NaN(nRuns, 1);
 
 % Aggregate relevant data
 for ii = 1:length(simHists)
@@ -26,11 +30,18 @@ for ii = 1:length(simHists)
     n(ii) = init.numAgents;
     doubleIntegrator(ii) = init.useDoubleIntegrator;
     numObjective(ii) = size(init.objectivePos, 1);
+    commsRadius(ii) = unique(init.comRange);
+    collisionRadius(ii) = unique(init.collisionRadius);
     for jj = 1:length(hist.out.agent)
         alphaDist(jj, ii) = hist.out.agent(jj).sensor.alphaDist;
+        positions{jj, ii} = hist.out.agent(jj).pos;
+        assert(hist.out.agent(jj).commsRadius == commsRadius(ii));
+        assert(hist.out.agent(jj).collisionRadius == collisionRadius(ii));
     end
 end
 
+commsRadius = unique(commsRadius); assert(isscalar(commsRadius));
+collisionRadius = unique(collisionRadius); assert(isscalar(collisionRadius));
 sensors = unique(alphaDist(1, :));
 
 config = [];
@@ -62,8 +73,8 @@ for ii = 1:length(simHists)
 end
 
 close all;
-f = figure;
+f1 = figure;
-x = axes; grid(x, "on");
+x1 = axes;
 
 n_unique = sort(unique(n));
 C = [];
@@ -77,5 +88,68 @@ ylabel("Final coverage (fraction of maximum)");
 title("Final performance of parameterizations");
 legend(["$AI\alpha$"; "$AI\beta$"; "$AII\alpha$"; "$BI\beta$"], "Interpreter", "latex");
 grid("on");
-keyboard
 
+f2 = figure;
+x2 = axes;
+
+% Compute pairwise distances between agents in each column of positions
+% cell array
+% Compute pairwise distances between agents
+maxPairs = nchoosek(6, 2); % 15 pairs for max 6 agents
+pairDist = cell(maxPairs, nRuns);
+for ii = 1:nRuns
+    pp = 0;
+    for jj = 1:n(ii)-1
+        for kk = jj+1:n(ii)
+            pp = pp + 1;
+            pairDist{pp, ii} = vecnorm(positions{jj, ii} - positions{kk, ii}, 2, 2);
+        end
+    end
+end
+
+% Compute mean, min, max pairwise distance across all pairs and timesteps per run
+meanPairDist = NaN(nRuns, 1);
+minPairDist = NaN(nRuns, 1);
+maxPairDist = NaN(nRuns, 1);
+for ii = 1:nRuns
+    nPairs = nchoosek(n(ii), 2);
+    D = vertcat(pairDist{1:nPairs, ii});
+    meanPairDist(ii) = mean(D);
+    minPairDist(ii) = min(D);
+    maxPairDist(ii) = max(D);
+end
+
+% Group pairwise distance stats by n-value (same layout as bar plot)
+nConfigs = nRuns / length(n_unique);
+meanD = NaN(length(n_unique), nConfigs);
+minD = NaN(length(n_unique), nConfigs);
+maxD = NaN(length(n_unique), nConfigs);
+for ii = 1:length(n_unique)
+    idx = find(n == n_unique(ii));
+    meanD(ii, :) = meanPairDist(idx)';
+    minD(ii, :) = minPairDist(idx)';
+    maxD(ii, :) = maxPairDist(idx)';
+end
+
+% Plot whiskers (min to max) with mean markers
+nGroups = length(n_unique);
+barWidth = 0.8;
+groupWidth = barWidth / nConfigs;
+hold(x2, 'on');
+for jj = 1:nConfigs
+    xPos = (1:nGroups) + (jj - (nConfigs + 1) / 2) * groupWidth;
+    errorbar(x2, xPos, meanD(:, jj), meanD(:, jj) - minD(:, jj), maxD(:, jj) - meanD(:, jj), ...
+        'o', 'LineWidth', 1.5, 'MarkerSize', 6, 'CapSize', 10);
+end
+hold(x2, 'off');
+set(x2, 'XTick', 1:nGroups, 'XTickLabel', string(n_unique));
+xlabel(x2, "Number of agents");
+ylabel(x2, "Pairwise distance");
+title(x2, "Pairwise Agent Distances (min/mean/max)");
+legend(x2, ["$AI\alpha$"; "$AI\beta$"; "$AII\alpha$"; "$BI\beta$"], "Interpreter", "latex");
+grid(x2, "on");
+
+yline(collisionRadius, 'r--', "Label", "Collision Radius", "LabelHorizontalAlignment", "left", "HandleVisibility", "off");
+yline(commsRadius, 'r--', "Label", "Communications Radius", "LabelHorizontalAlignment", "left", "HandleVisibility", "off");
+
+ylim([0, commsRadius + 5]);

test/results.m

@@ -10,7 +10,7 @@ classdef results < matlab.unittest.TestCase
         %% Diagnostic Parameters
         % No effect on simulation dynamics
         makeVideo = false; % disable video writing for big performance increase
-        makePlots = false; % disable plotting for big performance increase (also disables video)
+        makePlots = true; % disable plotting for big performance increase (also disables video)
         plotCommsGeometry = false; % disable plotting communications geometries
 
         %% Scenario Reinitialization
@@ -54,17 +54,6 @@ classdef results < matlab.unittest.TestCase
         end
     end
 
-    methods (TestMethodSetup)
-        % % Generate a random domain for each test
-        % function tc = setDomain(tc)
-        %     tc.testClass.domain = rectangularPrism;
-        %     % random integer-dimensioned cubic domain
-        %     tc.testClass.domain = tc.testClass.domain.initializeRandom(REGION_TYPE.DOMAIN, "Domain", tc.minDimension);
-        %     % Random bivariate normal PDF objective
-        %     tc.testClass.domain.objective = tc.testClass.domain.objective.initializeRandomMvnpdf(tc.testClass.domain, tc.discretizationStep, tc.protectedRange);
-        % end
-    end
-
     methods (Static, Access = private)
         function c = makeConfigs()
             rng(results.seed);