def lattice(data, funcG, funcW, n): #the only option - we need Gkw and Wqnu for self energy in the next iteration
#data.get_Gkw(funcG) #gets Gkw from G0 and Sigma
def func(var, data):
mu = var[0]
dt = data[0]
#print "func call! mu: ", mu, " n: ",dt.ns['up']
n= data[1]
dt.mus['up'] = mu
dt.mus['down'] = dt.mus['up']
dt.get_Gkw_direct(funcG) #gets Gkw from w, mu, epsilon and Sigma and X
dt.get_G_loc() #gets G_loc from Gkw
dt.get_n_from_G_loc()
#print "funcvalue: ",-abs(n - dt.ns['up'])
return 1.0-abs(n - dt.ns['up'])
if not MASTER_SLAVE_ARCHITECTURE: mpi.barrier()
varbest, funcvalue, iterations = amoeba(var=[data.mus['up']],
scale=[0.01],
func=func,
data = [data, n],
itmax=30,
ftolerance=1e-2,
xtolerance=1e-2)
if mpi.is_master_node():
print "mu best: ", varbest
print "-abs(diff n - data.n): ", funcvalue
print "iterations used: ", iterations
data.get_Gtildekw() #gets Gkw-G_loc
data.get_Wqnu_from_func(funcW) #gets Wqnu from P and J
data.get_W_loc() #gets W_loc from Wqnu
data.get_Wtildeqnu() #gets Wqnu-W_loc,
def search_for_mu(get_mu, set_mu, get_n, n, ph_symmetry, accepted_mu_range=[-20.0,20.0]):
print "getters: search_for_mu: n: ",n,", ph_symmetry",ph_symmetry, "accepted mu_range: ",accepted_mu_range
if (n is None) or ((n==0.5) and ph_symmetry):
print "no mu search to be performed! it is your duty to set the chemical potential to U/2. mu =",get_mu()
print 'n on the lattice : ', get_n()
else:
def func(var, data=None):
mu = var[0]
set_mu(mu)
actual_n = get_n()
val = 1.0-abs(n - actual_n)
print "amoeba func call: mu: %.2f desired n: %.2f actual n: %.2f val = "%(mu,n,actual_n),val
if val != val: return -1e+6
else: return val
print "about to do mu search:"
guesses = [get_mu(), 0.0, -0.1, -0.3, -0.4, -0.5, -0.7, 0.3, 0.5, 0.7]
found = False
for l in range(len(guesses)):
varbest, funcvalue, iterations = amoeba(var=[guesses[l]],
scale=[0.01],
func=func,
data = None,
itmax=30,
ftolerance=1e-2,
xtolerance=1e-2,
known_max = 1.0,
known_max_accr = 5e-5)
if (varbest[0]>accepted_mu_range[0] and varbest[0]<accepted_mu_range[1]) and (abs(funcvalue-1.0)<1e-2): #change the bounds for large doping
found = True
func(varbest)
break
if l+1 == len(guesses):
print "mu search FAILED: doing a scan..."
mu_grid = numpy.linspace(-1.0,0.3,50)
func_values = [func(var=[mu]) for mu in mu_grid]
print "func_values: "
for i in range(len(mu_grid)):
print "mu: ",mu_grid[i], " 1-abs(n-n): ", func_values[i]
mui_max = numpy.argmax(func_values)
print "using mu: ", mu_grid[mui_max]
set_mu(mu_grid[mui_max])
get_n()
if found:
print "guesses tried: ", l
print "mu best: ", varbest
print "1-abs(diff n - data.n): ", funcvalue
print "iterations used: ", iterations
f.write("%endblock PAO.Basis\n")
f.close()
atoms = ListOfAtoms([
Atom('Ga', (0.0, 0.0, 0.0), magmom=0),
Atom('Ga', (2.0, 0.0, 0.0), magmom=0)
])
a = Siesta(executable="/Users/ag/bin/siesta-xlf")
a.SetOption("%include", "base.fdf")
energy = a.run(atoms)
print rcs, rcp, energy
return -energy
URLbase = "http://fisica.ehu.es/ag/siesta-psffiles/"
# create a work subdirectory
orig_dir = os.getcwd()
dir = "simplex_work"
if os.path.isdir(dir): # does dir exist?
shutil.rmtree(dir) # yes, remove old directory
os.mkdir(dir) # make dir directory
os.chdir(dir) # move to dir
urllib.urlretrieve(URLbase + "Ga.lda.psf", "Ga.psf")
print amoeba.amoeba([4.25, 4.25], [2.5, 2.5], afunc, xtolerance=5.0e-2)
def main():
# global variables
global screen, pd, master, w, h, dcBright
## Graphics Initialization
w = 1200
h = 250
screen = pygame.display.set_mode((w, h))
pygame.display.set_caption("Amoebas")
pd = pygame.draw
drawmode = 'stick'
dcBright = randColorBright()
master = cake.c_master(friction=.92, gravity_glob=[0., 200.], slip=.8)
amoebas = []
amoeba0 = amoeba.amoeba(master, [100, h-80])
amoeba0.circle_res = 14
amoeba0.circle_radius = 50.
amoeba0.node_mass = 1.0
amoeba0.treading = 3
amoeba0.treadk = 100.
amoeba0.tread_damp = .5
amoeba0.musclek = 100
amoeba0.muscle_period = .2*2*pi
amoeba0.muscle_amp = 50.
amoeba0.muscle_damp = .5
amoeba1 = amoeba.amoeba(master, [100, h-80])
amoeba1.circle_res = 20
amoeba1.circle_radius = 50.
amoeba1.node_mass = 1.0
amoeba1.treading = 2
amoeba1.treadk = 200.
amoeba1.tread_damp = .5
amoeba1.musclek = 400
amoeba1.muscle_period = 1.2
amoeba1.muscle_amp = 60.
amoeba1.muscle_damp = .9
amoeba2 = amoeba.amoeba(master, [150, h-70])
amoeba2.circle_res = 14
amoeba2.circle_radius = 20.
amoeba2.node_mass = 1.
amoeba2.treading = 2
amoeba2.treadk = 100.
amoeba2.tread_damp = .9
amoeba2.musclek = 400.
amoeba2.muscle_period = .7
amoeba2.muscle_amp = 30.
amoeba2.muscle_damp = amoeba2.tread_damp
amoeba3 = amoeba.amoeba(master, [100, h-80])
amoeba3.circle_res = 8
amoeba3.circle_radius = 50.
amoeba3.node_mass = 1.0
amoeba3.treading = 2
amoeba3.treadk = 200.
amoeba3.tread_damp = .5
amoeba3.musclek = 400
amoeba3.muscle_period = .2*2*pi
amoeba3.muscle_amp = 60.
amoeba3.muscle_damp = .9
amoeba4 = amoeba.amoeba(master, [100, h-80])
amoeba4.circle_res = 10
amoeba4.circle_radius = 50.
amoeba4.node_mass = 1.0
amoeba4.treading = 2
amoeba4.treadk = 200.
amoeba4.tread_damp = .5
amoeba4.musclek = 400
amoeba4.muscle_period = .18*2*pi
amoeba4.muscle_amp = 90.
amoeba4.muscle_damp = .9
amoeba5 = amoeba.amoeba(master, [100, h-180])
amoeba5.circle_res = 16
amoeba5.circle_radius = 30.
amoeba5.node_mass = 1.0
amoeba5.treading = 2
amoeba5.treadk = 600.
amoeba5.tread_damp = .5
amoeba5.musclek = 40
amoeba5.muscle_period = .18*2*pi
amoeba5.muscle_amp = 20.
amoeba5.muscle_damp = .2
amoebas.append(amoeba5)
amoebas.append(amoeba0)
amoebas.append(amoeba1)
amoebas.append(amoeba2)
amoebas.append(amoeba3)
amoebas.append(amoeba4)
activeID = 0
amoebas[activeID].assemble()
# terrain
ground_wall = master.make_wall([(-200, h-20), (w+200, h-10)])
slope_wall = master.make_wall([(200, h-5), (400, h-50)])
master.make_wall([(400, h-50), (500, h)])
timescale = 1
timestep = time.time()
time_last = time.time()
time_start = time.time()
last_render = 0.
lagged = False
frame = -1
while True:
frame += 1
## Event handler
for event in pygame.event.get():
if event.type == pygame.QUIT:
pygame.quit(); sys.exit();
elif event.type == pygame.KEYDOWN:
if event.key == pygame.K_ESCAPE or event.unicode == 'q':
pygame.quit(); sys.exit();
elif event.key == pygame.K_v:
dcBright = randColorBright()
if drawmode == 'stick':
drawmode = 'solid'
elif drawmode == 'solid':
drawmode = 'stick'
elif event.unicode in ['1', '2', '3', '4', '5', '6', '7', '8', '9']:
if int(event.unicode) in range(1,len(amoebas)+1):
dcBright = randColorBright()
amoebas[activeID].destroy()
activeID = int(event.unicode)-1
amoebas[activeID].assemble()
print "amoeba ID:\t", activeID
elif event.key == pygame.K_RIGHT:
amoebas[activeID].control.right = True
elif event.key == pygame.K_LEFT:
amoebas[activeID].control.left = True
elif event.key == pygame.K_DOWN:
amoebas[activeID].control.down = True
elif event.key == pygame.K_UP:
amoebas[activeID].control.up = True
elif event.key == pygame.K_SPACE:
amoebas[activeID].control.poof = True
elif event.key == pygame.K_a:
amoebas[activeID].control.stick = True
# elif event.key == pygame.K_s:
# master.slip = .99
elif event.key == pygame.K_d:
amoebas[activeID].control.fat = True
elif event.type == pygame.KEYUP:
if event.key == pygame.K_RIGHT:
amoebas[activeID].control.right = False
elif event.key == pygame.K_LEFT:
amoebas[activeID].control.left = False
elif event.key == pygame.K_DOWN:
amoebas[activeID].control.down = False
elif event.key == pygame.K_UP:
amoebas[activeID].control.up = False
elif event.key == pygame.K_SPACE:
amoebas[activeID].control.poof = False
elif event.key == pygame.K_a:
amoebas[activeID].control.stick = False
# elif event.key == pygame.K_s:
# master.slip = .8
elif event.key == pygame.K_d:
amoebas[activeID].control.fat = False
elif event.type == pygame.MOUSEBUTTONDOWN:
amoebas[activeID].destroy()
activeID = (activeID + 1) % len(amoebas)
amoebas[activeID].assemble()
dcBright = randColorBright()
print "amoeba ID:\t", activeID
## Time Steppage
timestep = time.time() - time_last
time_last = time.time()
if master.update((time.time()-time_start)*timescale, max=.033) == False:
lagged = True
# print "simtime\t" + str(master.simtime)
# print "time offset\t" + str((time.time() - time_start) - master.simtime)
# print "frame\t" + str(frame)
# print "step\t" + str(timestep)
## Update Amoeba
amoebas[activeID].update(master.timestep)
# Win Test
if amoebas[activeID].circle[0].pos[0] >= w-15:
amoebas[activeID].destroy()
activeID = (activeID + 1) % len(amoebas)
amoebas[activeID].assemble()
dcBright = randColorBright()
print "amoeba ID:\t", activeID
## Render
# fps frames per second
since_render = time.time() - last_render
fps = 60.
dectime = time.time() - int(time.time())
if since_render >= 1/fps:
screen.fill((255, 255, 255))
render(drawmode, lagged)
pygame.display.flip()
lagged = False
last_render = time.time()
outname = outdir + '/' + sourceName + '-' + order
fp = open(outname + '-gtmap3.txt', 'w')
for iT in range(len(tList)):
t = tList[iT]
for iG in range(len(gList)):
g = gList[iG]
#p0 = [0.00010, 0.869, 5.0, 35.0, xStar[0], median(xStar), xStar[-1]]
p0 = [0.00010, 0.5, 0.0, 25.0, xStar[0], median(xStar), xStar[-1]]
scl = [
0.2 * p0[0], 0.1 * p0[1], 15.0, 25.0, p0[0] * 2.0, p0[0] * 2.0,
p0[0] * 2.0
]
ymodel = protoModels[:, iT, iG]
aresult = amoeba.amoeba(p0, scl, negchisq, data=(xmodel, ymodel))
p0 = aresult[0]
scl = list(array(scl) / 2.0)
aresult = amoeba.amoeba(p0, scl, negchisq, data=(xmodel, ymodel))
p0 = aresult[0]
print -aresult[1],
print t, g,
for thing in aresult[0]:
print thing,
print ''
if t >= 8.5 and t <= 28. and g >= 3.5 and g <= 5.5:
print >> fp, -aresult[1],
print >> fp, t, g,
for thing in aresult[0]:
print >> fp, thing,
def lattice(data, frozen_boson, n, ph_symmetry, accepted_mu_range=[-2.0,2.0]):
def get_n(dt):
dt.get_Gkw_direct() #gets Gkw from w, mu, epsilon and Sigma and X
dt.get_Fkw_direct() #gets Fkw from w, mu, epsilon and Sigma and X
dt.get_G_loc() #gets G_loc from Gkw
dt.get_n_from_G_loc()
if mpi.is_master_node(): print "supercond_hubbard: lattice"
if (n is None) or ((n==0.5) and ph_symmetry):
if n==0.5: #otherwise - nothing to be done
data.mus['up'] = 0
if 'down' in data.fermionic_struct.keys(): data.mus['down'] = data.mus['up']
get_n(data)
else:
def func(var, data):
mu = var[0]
dt = data[0]
#print "func call! mu: ", mu, " n: ",dt.ns['up']
n= data[1]
dt.mus['up'] = mu
if 'down' in dt.fermionic_struct.keys(): dt.mus['down'] = dt.mus['up']
get_n(dt) #print "funcvalue: ",-abs(n - dt.ns['up'])
val = 1.0-abs(n - dt.ns['up'])
if mpi.is_master_node(): print "amoeba func call: val = ",val
if val != val: return -1e+6
else: return val
if not MASTER_SLAVE_ARCHITECTURE: mpi.barrier()
if mpi.is_master_node(): print "about to do mu search:"
guesses = [data.mus['up'], 0.0, -0.1, -0.3, -0.4, -0.5, -0.7, 0.3, 0.5, 0.7]
found = False
for l in range(len(guesses)):
varbest, funcvalue, iterations = amoeba(var=[guesses[l]],
scale=[0.01],
func=func,
data = [data, n],
itmax=30,
ftolerance=1e-2,
xtolerance=1e-2,
known_max = 1.0,
known_max_accr = 5e-5)
if (varbest[0]>accepted_mu_range[0] and varbest[0]<accepted_mu_range[1]) and (abs(funcvalue-1.0)<1e-2): #change the bounds for large doping
found = True
func(varbest, [data, n])
break
if l+1 == len(guesses):
if mpi.is_master_node(): print "mu search FAILED: doing a scan..."
mu_grid = numpy.linspace(-1.0,0.3,50)
func_values = [func(var=[mu], data=[data,n]) for mu in mu_grid]
if mpi.is_master_node():
print "func_values: "
for i in range(len(mu_grid)):
print "mu: ",mu_grid[i], " 1-abs(n-n): ", func_values[i]
mui_max = numpy.argmax(func_values)
if mpi.is_master_node(): print "using mu: ", mu_grid[mui_max]
data.mus['up'] = mu_grid[mui_max]
if 'down' in data.fermionic_struct.keys(): data.mus['down'] = data.mus['up']
get_n(data)
if mpi.is_master_node() and found:
print "guesses tried: ", l
print "mu best: ", varbest
print "1-abs(diff n - data.n): ", funcvalue
print "iterations used: ", iterations
data.get_Gtildekw() #gets Gkw-G_loc
if not frozen_boson:
data.get_Wqnu_from_func(func = dict.fromkeys(data.bosonic_struct.keys(), dyson.scalar.W_from_P_and_J)) #gets Wqnu from P and J
data.get_W_loc() #gets W_loc from Wqnu, used in local bubbles
data.get_Wtildeqnu()
f.write("1 1\n")
f.write("%10.5f\n1.0\n" % (rcp,) )
f.write("%endblock PAO.Basis\n")
f.close()
atoms = ListOfAtoms([Atom('Ga', (0.0,0.0,0.0),magmom=0),
Atom('Ga', (2.0,0.0,0.0),magmom=0)])
a = Siesta(executable="/Users/ag/bin/siesta-xlf")
a.SetOption("%include","base.fdf")
energy = a.run(atoms)
print rcs, rcp, energy
return -energy
URLbase = "http://fisica.ehu.es/ag/siesta-psffiles/"
# create a work subdirectory
orig_dir = os.getcwd()
dir = "simplex_work"
if os.path.isdir(dir): # does dir exist?
shutil.rmtree(dir) # yes, remove old directory
os.mkdir(dir) # make dir directory
os.chdir(dir) # move to dir
urllib.urlretrieve(URLbase + "Ga.lda.psf", "Ga.psf")
print amoeba.amoeba([4.25,4.25],[2.5,2.5],afunc,xtolerance=5.0e-2)
def optimize(
runpath, resultspath, inputtemplatepath, vararr, runcommand, ftolerance=1.0e-4, xtolerance=1.0e-4, itmax=500
):
amoebalog = open(resultspath + "/amoeba.log", "w+")
optimizelog = open(resultspath + "/optimize.log", "w+")
optimizelog.write("runpath " + runpath + "\n")
optimizelog.write("resultspath " + resultspath + "\n")
optimizelog.write("inputtemplatepath " + inputtemplatepath + "\n")
optimizelog.write("Variables:\n")
inputtree = etree.parse(inputtemplatepath)
inputdscr = "Input file generated by script optimize.py \n"
inputdscr = inputdscr + "runpath: %s\n" % (runpath)
inputdscr = inputdscr + "resultspath: %s\n" % (resultspath)
inputdscr = inputdscr + "input template from path: %s\n" % (inputtemplatepath)
inputdscr = inputdscr + "optimization varibales:\n"
for var in vararr:
optimizelog.write("xpath: " + var[0] + "\n")
optimizelog.write("guess: " + str(var[1]) + "\n")
optimizelog.write("scale: " + str(var[2]) + "\n")
inputdscr = inputdscr + "xpath: " + str(var[0]) + "\n"
inputdscr = inputdscr + "guess: " + str(var[1]) + "\n"
inputdscr = inputdscr + "scale: " + str(var[2]) + "\n"
optimizelog.flush()
setinputxml.setByXpath(inputtree, "/input/keywords", inputdscr)
inputtree.write(runpath + "/input.xml", pretty_print="true")
guess = []
scale = []
data = []
data.append(runpath)
data.append(runcommand)
data.append([])
for var in vararr:
data[-1].append(var[0]) # build xpath array
guess.append(var[1]) # build array of initial guess values for the variables
scale.append(var[2])
simplex, fvalue, iteration = amoeba.amoeba(guess, scale, energy, ftolerance, xtolerance, itmax, data, amoebalog)
amoebalog.write("%d\t" % (iteration))
amoebalog.write("%1.10f\t" % (fvalue))
for value in simplex:
amoebalog.write("%1.10f\t" % (value))
amoebalog.write("\n")
amoebalog.flush()
optimizelog.write("\nOptimization converged after %d iterations of amoeba.py\n" % (iteration))
optimizelog.write("\nOptimized variables found:\n")
for i, var in enumerate(vararr):
optimizelog.write("xpath: " + var[0] + "\n")
optimizelog.write("optimized value: %1.0f\n\n" % (simplex[i]))
optimizelog.flush()
optimizelog.close()
amoebalog.close()