From Werner KRAUTH

Revision as of 23:46, 29 January 2011 Werner (Talk | contribs) ← Previous diff		Current revision Werner (Talk | contribs)
Line 4:		Line 4:
	[http://arxiv.org/abs/cond-mat/9612183v1 Electronic version (arXiv)]		[http://arxiv.org/abs/cond-mat/9612183v1 Electronic version (arXiv)]
		+
		+	=Further insights=
		+	In the original version, we formulated the algorithm as an exchange between two copies of a configuration. The "pocket" version of the algorithm, that I also used, with R. Moessner, in a [[Krauth_Moessner_2003|2003 paper on dimer models]], is much easier. This pocket version that is implemented in the below Python program.
		+
		+	=Python version of the cluster algorithm=
		+
		+	###========+=========+=========+=========+=========+=========+=========+=
		+	## PROGRAM : pocket_disks.py
		+	## PURPOSE : implement SMAC algorithm 2.18 (hard-sphere-cluster)
		+	## for four disks in a square of sides 1
		+	## LANGUAGE: Python 2.5
		+	##========+=========+=========+=========+=========+=========+=========+
		+	from random import uniform as ran, choice
		+	def box_it(x):
		+	x[0]=x[0]%1.
		+	if x[0] < 0 : x[0]=x[0]+1
		+	x[1]=x[1]%1.
		+	if x[1] < 0 : x[1]=x[1]+1
		+	return x
		+	def dist(x,y):
		+	d_x= abs(x[0]-y[0])%1
		+	d_x = min(d_x,1-d_x)
		+	d_y= abs(x[1]-y[1])%1
		+	d_y = min(d_y,1-d_y)
		+	return d_x**2 + d_y**2
		+	def T(x,Pivot):
		+	x=(2*Pivot[0]-x[0],2*Pivot[1]-x[1])
		+	return x
		+	#
		+	# Program starts here
		+	#
		+	Others=[(0.25,0.25),(0.25,0.75),(0.75,0.25),(0.75,0.75)]
		+	sigma_sq=0.15**2
		+	for iter in range(10000):
		+	a = choice(Others)
		+	Others.remove(a)
		+	Pocket = [a]
		+	Pivot=(ran(0,1),ran(0,1))
		+	while Pocket != []:
		+	a = choice(Pocket)
		+	Pocket.remove(a)
		+	a = T(a,Pivot)
		+	for b in Others[:]: # "Others[:]" is a copy of "Others"
		+	if dist(a,b) < 4*sigma_sq:
		+	Others.remove(b)
		+	Pocket.append(b)
		+	Others.append(a)
		+	print Others, ' ending config '
		+
		+
	[[Category:1995]] [[Category:Publication]] [[Category:Algorithm]]		[[Category:1995]] [[Category:Publication]] [[Category:Algorithm]]

---

Current revision

C. Dress, W. Krauth Cluster Algorithm for hard spheres and related systems Journal of Physics A: Math. Gen. 28, L597 (1995)

Abstract: In this paper, we present a cluster algorithm for the simulation of hard spheres and related systems. In this algorithm, a copy of the configuration is rotated with respect to a randomly chosen pivot point. The two systems are then superposed, and clusters of overlapping spheres in the joint system are isolated. Each of these clusters can be flipped independently, a process which generates non-local moves in the original configuration. A generalization of this algorithm (which works perfectly well at small density) can be successfully made to work at densities around the solid-liquid transition point in the two-dimensional hard-sphere system.

Electronic version (arXiv)

Further insights

In the original version, we formulated the algorithm as an exchange between two copies of a configuration. The "pocket" version of the algorithm, that I also used, with R. Moessner, in a 2003 paper on dimer models, is much easier. This pocket version that is implemented in the below Python program.

Python version of the cluster algorithm

###========+=========+=========+=========+=========+=========+=========+=
## PROGRAM : pocket_disks.py
## PURPOSE : implement SMAC algorithm 2.18 (hard-sphere-cluster) 
##           for four disks in a square of sides 1
## LANGUAGE: Python 2.5
##========+=========+=========+=========+=========+=========+=========+
from random import uniform as ran, choice
def box_it(x):
   x[0]=x[0]%1.
   if x[0] < 0 : x[0]=x[0]+1
   x[1]=x[1]%1.
   if x[1] < 0 : x[1]=x[1]+1
   return x
def dist(x,y):
   d_x= abs(x[0]-y[0])%1
   d_x = min(d_x,1-d_x)
   d_y= abs(x[1]-y[1])%1
   d_y = min(d_y,1-d_y)
   return  d_x**2 + d_y**2
def T(x,Pivot):
   x=(2*Pivot[0]-x[0],2*Pivot[1]-x[1])
   return x
#
# Program starts here
#
Others=[(0.25,0.25),(0.25,0.75),(0.75,0.25),(0.75,0.75)]
sigma_sq=0.15**2
for iter in range(10000):
   a = choice(Others)
   Others.remove(a)
   Pocket = [a]
   Pivot=(ran(0,1),ran(0,1))
   while Pocket != []:
      a = choice(Pocket)
      Pocket.remove(a)
      a = T(a,Pivot)
      for b in Others[:]:  # "Others[:]" is a copy of "Others"
         if dist(a,b) < 4*sigma_sq:
            Others.remove(b)
            Pocket.append(b)
      Others.append(a)
print Others, ' ending config '