This is a research code which solves Smoluchowski equation for coalescing particles. It also can login into a gmail account (login data to be obviously provided by the user) and send an email with one of the generated output pdf files as an attachment.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 | #! /usr/bin/env python
import scipy as s
import pylab as p
import scipy.integrate as si
from scipy import stats #I need this module for the linear fit
import smtplib
from email.MIMEMultipart import MIMEMultipart
from email.MIMEBase import MIMEBase
from email.MIMEText import MIMEText
from email import Encoders
import os
def mail(to, subject, text, attach):
msg = MIMEMultipart()
msg['From'] = gmail_user
msg['To'] = to
msg['Subject'] = subject
msg.attach(MIMEText(text))
part = MIMEBase('application', 'octet-stream')
part.set_payload(open(attach, 'rb').read())
Encoders.encode_base64(part)
part.add_header('Content-Disposition',
'attachment; filename="%s"' % os.path.basename(attach))
msg.attach(part)
mailServer = smtplib.SMTP("smtp.gmail.com", 587)
mailServer.ehlo()
mailServer.starttls()
mailServer.ehlo()
mailServer.login(gmail_user, gmail_pwd)
mailServer.sendmail(gmail_user, to, msg.as_string())
# Should be mailServer.quit(), but that crashes...
mailServer.close()
def Brow_ker_cont_optim(Vlist):
kern_mat=2.*k_B*T_0/(3.*mu)*(Vlist[:,s.newaxis]**(1./3.)+\
Vlist[s.newaxis,:]**(1./3.))**2./ \
(Vlist[:,s.newaxis]**(1./3.)*Vlist[s.newaxis,:]**(1./3.))
return kern_mat
def coupling_optim_garrick(y,t):
creation=s.zeros(n_bin)
destruction=s.zeros(n_bin)
#now I try to rewrite this in a more optimized way
destruction = -s.dot(s.transpose(kernel),y)*y #much more concise way to express\
#the destruction of k-mers
for k in xrange(n_bin):
kyn = (kernel*f_garrick[:,:,k])*y[:,s.newaxis]*y[s.newaxis,:]
creation[k] = s.sum(kyn)
creation=0.5*creation
out=creation+destruction
return out
#Now I work with the function for espressing smoluchowski equation when a uniform grid is used
def coupling_optim(y,t):
creation=s.zeros(n_bin)
destruction=s.zeros(n_bin)
#now I try to rewrite this in a more optimized way
destruction = -s.dot(s.transpose(kernel),y)*y #much more concise way to express\
#the destruction of k-mers
kyn = kernel*y[:,s.newaxis]*y[s.newaxis,:]
for k in xrange(n_bin):
creation[k] = s.sum(kyn[s.arange(k),k-s.arange(k)-1])
creation=0.5*creation
out=creation+destruction
return out
#Now I go for the optimal optimization of the chi_{i,j,k} coefficients used by Garrick for
# dealing with a non-uniform grid.
def mycount_garrick(V):
f=s.zeros((n_bin, n_bin, n_bin))
Vsum=V[:,s.newaxis]+V[s.newaxis,:] # matrix with the sum of the volumes in the bins
for k in xrange(1,(n_bin-1)):
f[:,:,k]=s.where((Vsum<=V[k+1]) & (Vsum>=V[k]), (V[k+1]-Vsum)/(V[k+1]-V[k]),\
f[:,:,k] )
f[:,:,k]=s.where((Vsum<=V[k]) & (Vsum>=V[k-1]),(Vsum-V[k-1])/(V[k]-V[k-1]),\
f[:,:,k])
return f
def total_concentration(number_mat, box_volume):
number= s.sum(number_mat,axis=1)*box_volume
return number
def total_mass_conservation(number_mat, vol_grid,box_volume):
ini_mass=s.dot(number_mat[0,:],vol_grid)*box_volume
fin_mass=s.dot(number_mat[-1,:],vol_grid)*box_volume
mass_conservation=(ini_mass-fin_mass)/ini_mass
results=s.array([ini_mass,fin_mass,mass_conservation])
return results
def fitting_stat(x,y):
slope, intercept, r, prob2, see = stats.linregress(x,y)
if (len(x)>2):
see=see*s.sqrt(len(x)/(len(x)-2.))
mx = x.mean()
sx2 = ((x-mx)**2).sum()
sd_intercept = see * s.sqrt(1./len(x) + mx*mx/sx2)
sd_slope = see * s.sqrt(1./sx2)
results=s.zeros(5)
results[0]=slope
results[1]=intercept
results[2]=r
if (len(x)>2):
results[3]=sd_slope
results[4]=sd_intercept
return results
#Now a list of the physical parameters needed to carry out the calculation
n_mon=5000 #total number of monomers
initial_density=0.01 #monomer density in the box
box_vol=n_mon/initial_density #volume of the box containing the monomers
r_mon=0.5 #radius of each monomer
v_mono=4./3.*s.pi*r_mon**3. #volume of each monomer
beta=1. #cluster-monomer 1/tau
k_B=1. #in these units
T_0=0.5 #temperature of the system
m_mon=1. #monomer mass in these units
sigma=1. #monomer diameter
mu=(m_mon*beta)/(3.*s.pi*sigma) # fluid viscosity
t=s.linspace(0.,1000.,1001) # choose time grid for time evolution
#Specify the bin structure you want to use
linear =1 #linear ==1---> use a linear bin structure and solve smoluchowski equation in standard form
#linear !1---> use a non-linear (log-spaced) bin structure and solve smoluchowski equation
# using the splitting operator
k_max=1000 #maximum number of monomers I consider in a k_mer
n_bin=200 # to be used only if a non-linear bin structure is used
send_email=1 #tells whether you want to send an email with an attached
# file
# variables for sending emails
gmail_user = "someaccount@gmail.com" #modify this using the account name and password of
#your own gmail account
gmail_pwd = "password"
mailto = "robert.h.schingler@nasa.gov"
mailtitle = "The job you submitted is done"
mailtext = "This is an automatically-generated email, please do not reply"
if (linear==1):
mailattachment="evolution_number_of_monomers_linear_binning.pdf"
else:
mailattachment="evolution_number_of_monomers_nonlinear_binning.pdf"
if (linear == 1):
k_list=s.linspace(1., k_max, k_max) #list of number I use to label each bin, i.e. size of the
#corresponding monomer
vol_grid=k_list*v_mono #volume of the particle in the k-th bin
n_bin=len(k_list) #overwrite the number of bins
elif (linear !=1):
k_list=s.logspace(s.log10(1.), s.log10(k_max),n_bin)
vol_grid=k_list*v_mono #volume of the particle in the k-th bin (this time the volume list is
#nonlinear)
if (linear !=1): #calculate the splitting operator on the non-uniform grid
f_garrick=mycount_garrick(vol_grid) #I calculate the splitting operator on the grid
#generate initial condition [monodisperse aerosol]
y0=s.zeros(n_bin)
y0[0]=initial_density #initial state (monodisperse aerosol)
#Generate the kernel matrix
kernel=Brow_ker_cont_optim(vol_grid)
if (linear==1):
solution = si.odeint(coupling_optim, y0, \
t,printmessg=1,rtol=1e-10,atol=1e-10)
elif (linear!=1):
solution = si.odeint(coupling_optim_garrick, y0, \
t,printmessg=1,rtol=1e-10,atol=1e-10)
total_monomers=total_concentration(solution, box_vol)
#now save the total number of monomers and the time in two separate files
if (linear==1):
p.save("number_monomers_linear_binning.dat", total_monomers)
elif (linear !=1):
p.save("number_monomers_nonlinear_binning.dat", total_monomers)
p.save("time.dat", t)
#check the quality of the simulation by testing mass conservation
mass_tests=total_mass_conservation(solution, vol_grid,box_vol)
print "initial and final total mass in the box are", mass_tests[0], mass_tests[1] ,"respectively"
print "mass is conserved up to ", mass_tests[2]*100., "percent"
#finally, perform some basic statistical analysis (fit decay of total number of clusters to a power-law)
sel_late=s.where(t>800.)
results=fitting_stat(s.log(t[sel_late]),s.log(total_monomers[sel_late]))
power_decay=results[0]
print "the exponent of the power-law decay [the theoretical value is -1] is, ", power_decay
#finally, plot the results
fig = p.figure()
axes = fig.gca()
axes.plot(t,total_monomers,"ro",linewidth=2.)
p.xlabel('Time')
p.ylabel('Total number of monomers')
p.title('Evolution of the total number of monomers')
p.grid(True)
if (linear ==1 ):
fig_name="evolution_number_of_monomers_linear_binning.pdf"
elif (linear !=1):
fig_name="evolution_number_of_monomers_nonlinear_binning.pdf"
p.savefig(fig_name)
p.clf()
#Now send one of the generated plots automatically by email
#NB: this will not work unless one really inputs its email login details
mail(mailto,mailtitle, mailtext,mailattachment)
print "Calculation ended."
|