Random key from Ruby Hash (even faster)

blackrat (Paul McKibbin) — Wed, 02 May 2007 03:00:58 GMT

I came across baby's code and then aiosup's from 2 years ago, and realized that I like the simplicity of one and the approach of the other. Kudos to both authors, and I found a way to combine the two approaches into one neat mechanism. More of my ruby here,



class Hash

  @keys_not_used = nil



  def random_key

          @keys_not_used = self.dup if (!@keys_not_used or @keys_not_used.size == 0)

	  key = @keys_not_used.keys[rand(@keys_not_used.size)]

	  @keys_not_used.delete(key)

          key

  end



end

Particle Swarm Optimization

bugmenot (BugMeNot) — Fri, 19 Jan 2007 00:14:49 GMT

Particle Swarm Optimization written in Python, more infos and a pretty printer here : http://www.biais.org/blog/index.php/2007/01/14/13-metaheuristic-particle-swarm-optimization-pso-in-python



# Particle swarm optimization

# Maxime Biais : 



from random import uniform

 

class PSO:

    def __init__(self, pop_size, min, max, phi, phi2, lr, maxiter, func):

        self.func = func

        self.pop = []

        # 0: position, 1: velocity, 2: fitness

        self.min = min

        self.max = max

        for i in xrange(pop_size):

            self.pop.append([uniform(self.min, self.max), 

                                   uniform(-1, 1), 0])

        self.evaluate()

        self.gdest = self.pop[0]

        self.pdest = self.pop[0]

        self.phi = phi

        self.phi2 = phi2

        self.lr = lr

        self.maxiter = maxiter

    

    def update_velocity(self):

        for i in self.pop:

            i[1] = self.lr * i[1] + uniform(0, self.phi) \

                    * (self.pdest[0] - i[0]) + uniform(0, self.phi2) \

                    * (self.gdest[0] - i[0])

 

    def evaluate(self):

        for i in self.pop:

            i[2] = self.func(i[0])

 

    def move(self):

        for i in self.pop:

            i[0] += i[1]

 

    def __cmp_by_fitness(self, a, b):

        return cmp(a[2], b[2])

    

    def run(self, update_func=False):

        for i in xrange(self.maxiter):

            if update_func:

                update_func()

            self.update_velocity()

            self.move()

            self.evaluate()

            self.pop.sort(self.__cmp_by_fitness, reverse=0)

            self.pdest = self.pop[0]

            if self.pdest[2] < self.gdest[2]:

                self.gdest = self.pdest

DZone Snippets: optimization code

Random key from Ruby Hash (even faster)

Particle Swarm Optimization