Changeset 34 for liacs

Timestamp:

Dec 16, 2009, 1:24:21 PM (15 years ago)

Author:

Rick van der Zwet

Message:

First pso attempt commit :-)

Location:

liacs/nc/laser-pulse-shaping

Files:

• Makefile (modified) (1 diff)
• pso.m (copied) (copied from liacs/nc/low-correlation/sa.m ) (2 diffs)

liacs/nc/laser-pulse-shaping/Makefile

@@ -1 @@
-mcs.out: mcs.m
-        @octave -q mcs.m
+pso.out: pso.m
+        @octave -q pso.m
     

liacs/nc/laser-pulse-shaping/pso.m

@@ -1 @@
-% Simulated Annealing low-autocorrelation program
+% Particle Swarm optimalisation
 % BSDLicence
 % Rick van der Zwet - 0433373 - <hvdzwet@liacs.nl>
-
-% Brute-force result of length 20
-% best_20 = [-1,1,1,-1,-1,-1,1,1,-1,1,-1,-1,-1,1,-1,1,1,1,1,1];
-% autocorrelation(best_20);
+% Modeled after http://en.wikipedia.org/wiki/Particle_swarm_optimization
 
 function number = randint(low,high)
@@ -24 +21 @@
 endfunction
 
-function [fitness,value] = sa(length, temperature, stopLimit)
-    best_fitness = 0;
-    s = initseq(length);
-    stopCounter = 0;
-
-    while (stopCounter < stopLimit)
-        % Generate new mutation
-        newseq = mutation(s);
-
-        % Better is always accept
-        fitness = autocorrelation(newseq);
-        if (fitness > best_fitness)
-            best_value = s;
-            best_fitness = fitness;
-            s = newseq;
-            fitness
-            disp(rot90(newseq,-1));
-            stopCounter = 0;
-        else
-            % Make the next 'move' less atractive
-            temperature =- 1;
-
-            stopCounter += 1;
-            % Accept on an certain probability
-            if (temperature < 0)
-                break;
-            else 
-                if(randint(0,temperature) > temperature / 4)
-                    s = newseq;
-                endif
-            endif
-        endif
-    endwhile
-    value = best_value;
-    fitness = best_fitness;
+% Iterate on multiple vectors
+function fitness = SHGa(v)
+    for i = 1:length(v(1,:))
+        fitness(i) = SHG(v(:,i));
+    endfor
 endfunction
 
-%% Basic variables
-iterations = [1:10:1000];
-repetitions = 20;
-length = 20;
-temperature = 1000;
+% Parameters
+iterations = 1;
+parameters = 3;
+local_swarm_size = 2;
+local_swarms = 5;
+wander = 0.5;
+
+% 'Influence' of the envirionment with regards to solutions
+% Trust the group global solution to be feasible
+c_social  = 0.4;
+% Trust the neighbor solution to be feasible
+c_cognitive = 0.4;
+% Trust the own best solution to be feasible
+c_ego = 0.2;
+
+% Initiate all particles
+flock_p = rand(parameters,local_swarm_size,local_swarms) .* (2 * pi);
+flock_v = zeros(size(flock_p));
+    
+
+% Global best placeholder
+g_best = zeros(parameters,1);
+g_fitness = 1;
+% at (:,x) lives the neighbor best of local_swarm 'x'
+n_best = zeros(parameters,local_swarms);
+n_fitness = ones(local_swarms);
+% at (:,p,x) leves the local best of particle 'p' in local_swarm 'x'
+l_best = zeros(parameters,local_swarm_size,local_swarms);
+l_fitness = ones(local_swarm_size, local_swarms);
 
 
-% Plot the stuff
-fitnesses = [];
-for iteration = iterations
-    fitness = [];
-    for rep = 1:repetitions
-        printf('Iter:%i, Rep:%i\n',iteration,rep);
-        [new_fitness, value] =  sa(length,iteration,temperature);
-        fitness = [fitness, new_fitness];
+% Code not optimised for performance, but for readablility
+for i = 1:iterations
+    for s = 1:local_swarms
+        fitness = SHGa(flock_p(:,:,s));
+        % See if we got any better local optimum
+        for p = 1:local_swarm_size
+            if fitness(p) < l_fitness(p,s)
+                l_fitness(p,s) = fitness(p);
+                l_best(:,p,s) = flock_p(:,p,s);
+            endif
+        endfor
 
-        % Little hack to display output nicely
-        % disp(rot90(value,-1));
+        % See if we got any better neighbor optimum
+        for p = 1:local_swarm_size
+            if l_fitness(p,s) < n_fitness(s)
+                n_fitness(s) = l_fitness(p,s);
+                n_best(:,s) = l_best(:,p,s);
+            endif
+        endfor
     endfor
-    
-    fitnesses = [fitnesses,mean(fitness)];
+
+    % See wether we have a new global optimum
+    for s = 1:local_swarms
+        if n_fitness(s) < g_fitness
+          g_fitness = n_fitness(s);
+          g_best = n_best(s);
+        endif
+    endfor
+
+    % Update particles to new value
+    r_cognitive = rand();
+    r_social = rand();
+    r_ego = rand();
+    for s = 1:local_swarms
+        for p = 1:local_swarm_size
+            flock_v(:,p,s) = (flock_v(:,p,s) .* wander) + \
+                (g_best .* (c_cognitive * r_cognitive)) + \
+                (n_best(s) .* (c_social *r_social)) + \
+                (l_best(p,s) .* (c_ego *r_ego));
+            s
+            p
+            flock_p(:,p,s)
+            flock_p(:,p,s) += flock_v(:,p,s);
+            flock_p(:,p,s)
+        endfor
+    endfor
 endfor
 
-plot(iterations,fitnesses);
-title(sprintf('Simulated Annealing on Low-Corretation set - repetitions %i',repetitions));
-ylabel('fitness');
-xlabel('iterations');
-grid on;
-legend(sprintf('Length %i',length));
-print("sa-fitness.eps","-depsc2");
 
+
+%plot(iterations,fitnesses);
+%title(sprintf('Particle Swarm Optimalisation on Laser-Pulse shaping problem - repetitions %i',repetitions));
+%ylabel('fitness');
+%xlabel('iterations');
+%grid on;
+%legend(sprintf('Length %i',length));
+%print("sa-fitness.eps","-depsc2");
+