$x^2$ potential


In [1]:
import pickle
import numpy as np
import pycuda.gpuarray as gpuarray
from scipy.special import hyp1f1
import scipy.fftpack as fftpack
import pylab as plt
import time

#-------------------------------------------------------------------------------------
from pywignercuda_path import SetPyWignerCUDA_Path
SetPyWignerCUDA_Path()
from GPU_Wigner2D_FFT import *

In [2]:
%matplotlib inline

Settings


In [3]:
class frame( GPU_Wigner2D_FFT ):
    def __init__ (self):
        X_gridDIM = 512
        P_gridDIM = 512
        
        X_amplitude  = 15
        P_amplitude  = 10
        
        kappa = 1.
        dt= 0.005
        
        timeSteps =    1000
        skipFrames =     200
        
        mass = 1.      
        # Diffusion parameter 
        D_Theta  = 0.0143
        D_Lambda = 0.000    
        
        # Damping parameters (implies another source of diffusion as well)
        self.dampingFunction = 'CaldeiraLeggett'
        gammaDamping = 0.07 #8*10**(-6)
        epsilon       = 0.01;
        
         #Gross Pitaevskii coefficient
        self.grossPitaevskiiCoefficient = 0.0
        
        # Potential and derivative of potential
        self.omega = 0
        X2_constant = 0.5*mass*self.omega**2
        
        potentialString  = '{0}*pow(x,2)'.format(X2_constant)

        dPotentialString = '2*{0}*x'.format(X2_constant)
        
        kinematicString  = '0.5*p*p' #.format(mass=mass)
        
        self.fp_Damping_String = ' p*p/sqrt( p*p + {epsilon}  ) '.format( epsilon=epsilon )
        
        self.SetTimeTrack( dt, timeSteps, skipFrames,
        fileName = '/home/rcabrera/DATA/Wigner2D/X2/Free.hdf5' )
        
        GPU_Wigner2D_FFT.__init__(self,
            X_gridDIM,P_gridDIM,X_amplitude,P_amplitude,
            kappa,mass,D_Theta,D_Lambda,gammaDamping,potentialString,dPotentialString,kinematicString)
    
    
    def Set_Initial_Condition_HarmonicOscillator(self):
        """
        Sets   self.PsiInitial_XP with the Wigner function of the Harmonic oscillator  
        """
        self.x_init = -5
        self.p_init = 2.
        n=0
        omega = 1
        self.W_init = self.Wigner_HarmonicOscillator(n, omega , self.x_init, self.p_init)

Run


In [4]:
instance = frame()
print '							'
print ' 	Wigner2D propagator with damping	'
print '							'

instance.Set_Initial_Condition_HarmonicOscillator ()

%time instance.Run( )


							
 	Wigner2D propagator with damping	
							
 X_gridDIM =  512    P_gridDIM =  512
 dx =  0.05859375  dp =  0.0390625
 dLambda =  0.209439510239  dTheta =  0.314159265359
  
         GPU memory Total        5.24945068359 GB
         GPU memory Free         4.92453765869 GB
         GPU memory Free  post gpu loading  4.89328765869 GB
 ------------------------------------------------------------------------------- 
     Split Operator Propagator  GPU with damping                                 
 ------------------------------------------------------------------------------- 
 progress  0 %
 progress  19 %
 progress  39 %
 progress  59 %
 progress  79 %
 progress  99 %
CPU times: user 8.71 s, sys: 5.98 s, total: 14.7 s
Wall time: 15.9 s
Out[4]:
0

Plots


In [5]:
print 'Potential'
fig, ax = plt.subplots(figsize=(10, 3))
ax.plot( instance.X_range,   instance.Potential(0,instance.X_range) )
#ax.set_xlim(-10,10)
ax.set_ylim(-1,60)
ax.set_xlabel('x')
ax.set_ylabel('V')
ax.grid('on')


Potential

In [6]:
def PlotWignerFrame( W_input , x_plotRange,p_plotRange):
    W = W_input.copy()
    W = fftpack.fftshift(W.real)    
    
    dp    = instance.dP
    p_min = -instance.P_amplitude
    p_max =  instance.P_amplitude - dp   
    
    #p_min = -dp*instance.P_gridDIM/2.
    #p_max =  dp*instance.P_gridDIM/2. - dp    
    
    x_min = -instance.X_amplitude
    x_max =  instance.X_amplitude - instance.dX
    
    global_max = 0.17          #  Maximum value used to select the color range
    global_min = -0.31         # 
        
    print 'min = ', np.min( W ), ' max = ', np.max( W )
    print 'final time =', instance.timeRange[-1] ,'a.u.  =',\
    instance.timeRange[-1]*( 2.418884326505*10.**(-17) ) , ' s '
    
    print 'normalization = ', np.sum( W )*instance.dX*dp

    zero_position =  abs( global_min) / (abs( global_max) + abs(global_min)) 
    wigner_cdict = {'red' 	: 	((0., 0., 0.),
							(zero_position, 1., 1.), 
							(1., 1., 1.)),
					'green' :	((0., 0., 0.),
							(zero_position, 1., 1.),
							(1., 0., 0.)),
					'blue'	:	((0., 1., 1.),
							(zero_position, 1., 1.),
							(1., 0., 0.)) }
    wigner_cmap = matplotlib.colors.LinearSegmentedColormap('wigner_colormap', wigner_cdict, 256)

    fig, ax = plt.subplots(figsize=(12, 5))

    cax = ax.imshow( W ,origin='lower',interpolation='none',\
    extent=[ x_min , x_max, p_min, p_max], vmin= global_min, vmax=global_max, cmap=wigner_cmap)

    ax.contour(instance.Hamiltonian ,
                np.arange(0, 10, 1 ),origin='lower',extent=[x_min,x_max,p_min,p_max],
               linewidths=0.25,colors='k')
    
    axis_font = {'size':'24'}
    
    ax.set_xlabel(r'$x$',**axis_font)
    ax.set_ylabel(r'$p$',**axis_font)
    
    ax.set_xlim((x_plotRange[0] , x_plotRange[1] ))
    ax.set_ylim((p_plotRange[0] , p_plotRange[1] ))
    ax.set_aspect(1.)
    #ax.grid('on')
    cbar = fig.colorbar(cax, ticks=[-0.3, -0.2,-0.1, 0, 0.1, 0.2 , 0.3])
    matplotlib.rcParams.update({'font.size': 18})
    return fig

In [7]:
plot_init = PlotWignerFrame( instance.W_init.real , (-12.,12) ,(-5,5)  )


min =  1.83247742086e-236  max =  0.318169063665
final time = 5.0 a.u.  = 1.20944216325e-16  s 
normalization =  1.0

In [8]:
plot_init = PlotWignerFrame( instance.W_end , (-12.,12) ,(-5,5)  )


min =  -3.80615540457e-08  max =  0.257950430789
final time = 5.0 a.u.  = 1.20944216325e-16  s 
normalization =  1.0

In [9]:
def PlotMarginals():
    
    W = fftpack.fftshift( instance.W_end )
    
    dp    = instance.dP
    p_min = -instance.P_amplitude
    p_max =  instance.P_amplitude - dp   
        
    W0 = fftpack.fftshift(instance.W_init  )
    
    marginal_x_init = np.sum(  W0 , axis=0 )*dp
    marginal_p_init = np.sum(  W0 , axis=1 )*instance.dX

    marginal_x = np.sum(  W, axis=0 )*dp
    marginal_p = np.sum(  W, axis=1 )*instance.dX


    x_min = -instance.X_amplitude
    x_max = instance.X_amplitude - instance.dX 
    #.......................................... Marginal in position

    plt.figure(figsize=(10,10))
    plt.subplot(211)

    plt.plot(instance.X_range, marginal_x_init, '-',label='initial')
    plt.plot(instance.X_range, marginal_x,  label='final')
    #plt.axis([x_min, 0*x_max, -0.01,6])
    plt.xlabel('x')
    plt.ylabel('Prob')

    plt.legend(loc='upper right', shadow=True)

    #..........................................  Marginal in momentum

    print 'p = ', np.sum( marginal_p*instance.P_range )*dp,\
    '->', np.sum( W*instance.P )*instance.dX*dp
    print 'x = ', np.sum( W0*instance.X )*instance.dX*dp, \
    '->',np.sum( W*instance.X )*instance.dX*dp
    
    rangeP = np.linspace( p_min, p_max, instance.P_gridDIM )
    
    plt.subplot(212)
    plt.plot(rangeP, marginal_p_init ,'-', label='initial')
    plt.plot(rangeP, marginal_p  , label='final')
    plt.axis([p_min, p_max, -0.01, 1])
    plt.xlabel('p')
    plt.ylabel('Prob')

    plt.legend(loc='upper right', shadow=True)

In [10]:
PlotMarginals()


p =  0.994881866633 -> -8.77476570163
x =  9.99999999997 -> -6.48217421318

First Order Ehrenfest


In [11]:
fig, ax = plt.subplots(figsize=(10, 4))

ax.plot( instance.timeRange ,  np.gradient(instance.X_average, instance.dt) , '-',
        label = '$\\frac{d}{dt} \\langle x \\rangle $' ,color = 'red', linewidth=1.5)

ax.plot( instance.timeRange ,  instance.P_average/instance.mass , '--' ,
        label='$\\frac{1}{m}\\langle p \\rangle$', linewidth=1.5 )


#ax.set_xlim(0,3.5)
#ax.set_ylim(-1.,1.2)
ax.legend(bbox_to_anchor=(1.05, 1), loc=2, prop={'size':22})
ax.set_xlabel('t')
ax.grid();



In [12]:
fig, ax = plt.subplots(figsize=(10, 4))

ax.plot( instance.timeRange ,  np.gradient( instance.P_average , instance.dt) ,'-' , 
        label =  '$\\frac{d}{dt} \\langle p \\rangle $' ,color = 'r' , linewidth=1.5)

ax.plot( instance.timeRange , 
        - instance.dPotentialdX_average  -2*instance.gammaDamping*instance.P_average  , '--' ,
        label = '$ \\langle \\frac{d}{dx}V  \\rangle - 2 \\gamma \\langle p \\rangle $' ,linewidth=1.5)


ax.legend(bbox_to_anchor=(1.05, 1), loc=2, prop={'size':22})
#ax.set_ylim(-0.8,0.8)
ax.set_xlabel('t')
#ax.set_ylabel(' ')
ax.grid();


Second Order Ehrenfest Theorems


In [13]:
fig, ax = plt.subplots(figsize=(10, 4))

ax.plot( instance.timeRange , np.gradient( instance.X2_average , instance.dt) , '-',
        label='$\\frac{d}{dt}  \\langle x^2  \\rangle$' , color = 'red', linewidth=1.5)

ax.plot( instance.timeRange , \
        2*instance.XP_average/instance.mass, '--',label = '$\\frac{2}{m}  \\langle xp  \\rangle$',linewidth=1.5 )


#ax.set_xlim(0,3.5)
#ax.set_ylim(-1.,1.2)
ax.legend(bbox_to_anchor=(1.05, 1), loc=2, prop={'size':24})
ax.set_xlabel('t')
ax.grid();



In [14]:
fig, ax = plt.subplots(figsize=(12, 6))

ax.plot( instance.timeRange , np.gradient( instance.P2_average , instance.dt)  , '-',
        label = '$\\frac{d}{dt}  \\langle p^2  \\rangle$',
        color = 'red', linewidth=1.5)

ax.plot( instance.timeRange , \
        -2*instance.PdPotentialdX_average\
        +2.*instance.D_Theta \
        -4*instance.gammaDamping*instance.P2_average
        , '--',
        label = '$-  \\langle p\\frac{dV}{dx} +\\frac{dV}{dx} p \\rangle + 2 D_{\\theta}- 4\\gamma \\langle p^2 \\rangle $',
        linewidth=1.5 )


#ax.set_xlim(-0.2,26)
#ax.set_ylim(-1.,1.2)
ax.legend(bbox_to_anchor=(1.05, 1), loc=5, prop={'size':22})
ax.set_xlabel('t')
ax.grid();



In [15]:
fig, ax = plt.subplots(figsize=(12, 6))

ax.plot( instance.timeRange ,  2*np.gradient( instance.XP_average , instance.dt) ,
        '-' ,label = '$\\frac{d}{dt} \\langle xp+px  \\rangle$' , color = 'r' , linewidth=1.5 )

ax.plot( instance.timeRange , \
          2*instance.P2_average/instance.mass  \
         -2*instance.XdPotentialdX_average   \
         -4*instance.gammaDamping*instance.XP_average
         , '--' , 
         label = '$\\frac{2}{m}  \\langle p^2  \\rangle - 2  \\langle x \\frac{d}{dx}V  \\rangle - 2 \\gamma  \\langle xp + px  \\rangle  $'
        ,linewidth=1.5)


ax.legend(bbox_to_anchor=(1.05, 1), loc=5, prop={'size':22})
#ax.set_ylim(- 12 , 7)
ax.set_xlabel('t')
ax.set_ylabel(' ')
ax.grid();



In [16]:
fig, ax = plt.subplots(figsize=(12, 6))

ax.plot( instance.timeRange ,
np.sqrt(instance.X2_average - instance.X_average**2)*np.sqrt(instance.P2_average - instance.P_average**2)
        , '-' , label = '$\\Delta x \\Delta p$' , linewidth=1.)


ax.legend(bbox_to_anchor=(1.05, 1), loc=5, prop={'size':22})
#ax.set_ylim(-50, 0)
ax.set_xlabel('t')
#ax.set_xlim(0,41000)
ax.set_ylabel(' ')
ax.grid();



In [17]:
fig, ax = plt.subplots(figsize=(12, 6))

ax.plot( instance.timeRange , instance.Hamiltonian_average 
        , '-' , label = '$Energy$' , linewidth=1.)


ax.legend(bbox_to_anchor=(1.05, 1), loc=5, prop={'size':22})
#ax.set_ylim(3.48 , 3.52)
ax.set_xlabel('t')
ax.set_ylabel(' ')
ax.grid();



In [18]:
fig, ax = plt.subplots(figsize=(12, 6))

ax.plot( instance.timeRange , \
          np.sqrt(instance.P2_average - instance.P_average**2) \
         , '-' , label = '$p^2 $',linewidth=2.)

#ax.plot( instance.timeRange , instance.X3_average - 2*gamma*instance.P_average , '-' ,
#        label = '$-F-2\gamma <P>$' ,linewidth=2.)

ax.legend(bbox_to_anchor=(1.05, 1), loc=2, prop={'size':22})
#ax.set_ylim(- 12 , 7)
ax.set_xlabel('t')
ax.set_ylabel(' ')
ax.grid();



In [19]:
fig, ax = plt.subplots(figsize=(12, 6))

ax.plot( instance.timeRange , \
          np.sqrt(instance.X2_average - instance.X_average**2) \
         , '-' , label = '$x^2 $',linewidth=2.)

#ax.plot( instance.timeRange , instance.X3_average - 2*gamma*instance.P_average , '-' ,
#        label = '$-F-2\gamma <P>$' ,linewidth=2.)

ax.legend(bbox_to_anchor=(1.05, 1), loc=2, prop={'size':22})
#ax.set_ylim(- 12 , 7)
ax.set_xlabel('t')
ax.set_ylabel(' ')
ax.grid();


Classical trajectory


In [20]:
def dVdx(x):
    return instance.dPotential(0,x)

In [21]:
xp_init = np.array([ instance.x_init, instance.p_init  ])

trajectory = instance.SymplecticPropagator(
    instance.dt, instance.timeSteps, xp_init, instance.gammaDamping ).T

In [22]:
x_min = -instance.X_amplitude
x_max = instance.X_amplitude - instance.dX 

p_min = -instance.dP*instance.P_gridDIM/2.
p_max =  instance.dP*instance.P_gridDIM/2. -instance.dP 

print ' Quantum Ehrenfest trajectory vs classical trajectory'
print ' final time = ', instance.dt*instance.timeSteps
fig, ax = plt.subplots(figsize=(10, 10))

ax.plot( instance.X_average ,  instance.P_average , '-' ,color = 'g', 
        label = 'quantum (<X>,<P>)', linewidth=1.5 )
ax.set_xlabel('X')
ax.set_ylabel('P')
ax.set_aspect(1.5)

ax.set_xlim(-4.,4.)
ax.set_ylim(-4.,4.)

ax.plot( trajectory[0] ,  trajectory[1] , '--' , color='r', 
        label = 'classical (X,P)', linewidth=1. )

ax.contour(instance.Hamiltonian ,
                np.arange(-45, 100, 2 ),origin='lower',extent=[x_min,x_max,p_min,p_max],
               linewidths=0.25,colors='k')

ax.legend(bbox_to_anchor=(1.05, 1), loc=2)
#ax.grid();
ax.set_aspect(1.)


 Quantum Ehrenfest trajectory vs classical trajectory
 final time =  5.0

In [23]:
def NumpyPropertyOnly (obj):
    """
    Class that will contain only picklable properties
    """
    obj_ = dict()
    for prop_name in dir(obj) :
        prop = getattr(obj, prop_name)
        if isinstance(prop, np.ndarray) : obj_[prop_name] = prop
    return obj_

In [24]:
#pickle.dump( NumpyPropertyOnly(instance) , open( "X2.pickle", "wb" ) )

Loading saved File


In [25]:
W = instance.WignerFunctionFromFile(1*instance.skipFrames)

In [26]:
instance.PlotWignerFrame( W.real , 
                         plotRange=((-6.,6) ,(-6,6)),
                         global_color=(-0.2, 0.2),
                         energy_Levels=(0, 20, 1), aspectRatio=1.);


min =  -5.44084584972e-09  max =  0.350870096
normalization =  1.0

In [26]: