In [ ]:

    

from vpython import *
from numpy import arange, array, newaxis, square, sum, sqrt
from math import pi, exp, sin, cos
import math


    

In [ ]:

    

win=500

Natoms = 1000  # change this to have more or fewer atoms

# Typical values
L = 2. # container is a cube L on a side
gray = (0.7,0.7,0.7) # color of edges of container
Matom = 4E-3/6E23 # helium mass
Ratom = 0.03 # wildly exaggerated size of helium atom
k = 1.4E-23 # Boltzmann constant
T = 300. # around room temperature
dt = 1E-5


    

In [ ]:

    

scene = canvas(title="Gas", width=win, height=win, x=0, y=0,
                center=vec(L/2.,L/2.,L/2.))

deltav = 100. # binning for v histogram
vdist = graph(x=0, y=win, ymax = Natoms*deltav/1000.,
             width=win, height=0.6*win, xtitle='v', ytitle='dN')
theory = gcurve(color=color.cyan)

dv = 10.
for v in arange(0.,3001.+dv,dv): # theoretical prediction
    theory.plot(pos=(v,
        (deltav/dv)*Natoms*4.*pi*((Matom/(2.*pi*k*T))**1.5)
                     *exp((-0.5*Matom*v**2)/(k*T))*v**2*dv))

# observation = ghistogram(bins=arange(0.,3000.,deltav),
#                         accumulate=1, average=1, color=color.red)


    

In [ ]:

    

gray = vec(0.7,0.7,0.7)
gray


    

In [ ]:

    

xaxis = curve(pos=[vec(0,0,0), vec(L,0,0)], color=gray)
yaxis = curve(pos=[vec(0,0,0), vec(0,L,0)], color=gray)
zaxis = curve(pos=[vec(0,0,0), vec(0,0,L)], color=gray)
xaxis2 = curve(pos=[vec(L,L,L), vec(0,L,L), vec(0,0,L), vec(L,0,L)], color=gray)
yaxis2 = curve(pos=[vec(L,L,L), vec(L,0,L), vec(L,0,0), vec(L,L,0)], color=gray)
zaxis2 = curve(pos=[vec(L,L,L), vec(L,L,0), vec(0,L,0), vec(0,L,L)], color=gray)


    

In [ ]:

    

Atoms = []
colors = [color.red, color.green, color.blue,
          color.yellow, color.cyan, color.magenta]
poslist = []
plist = []
mlist = []
rlist = []


    

In [ ]:

    

for i in range(Natoms):
    Lmin = 1.1*Ratom
    Lmax = L-Lmin
    x = Lmin+(Lmax-Lmin)*random()
    y = Lmin+(Lmax-Lmin)*random()
    z = Lmin+(Lmax-Lmin)*random()
    r = Ratom
    Atoms = Atoms+[sphere(pos=vec(x,y,z), radius=r, color=colors[i % 6])]
    mass = Matom*r**3/Ratom**3
    pavg = sqrt(2.*mass*1.5*k*T) # average kinetic energy p**2/(2mass) = (3/2)kT
    theta = pi*random()
    phi = 2*pi*random()
    px = pavg*sin(theta)*cos(phi)
    py = pavg*sin(theta)*sin(phi)
    pz = pavg*cos(theta)
    poslist.append((x,y,z))
    plist.append((px,py,pz))
    mlist.append(mass)
    rlist.append(r)


    

In [ ]:

    

pos = array(poslist)
p = array(plist)
m = array(mlist)
m.shape = (Natoms,1)
radius = array(rlist)

pos = pos+(p/m)*(dt/2.) # initial half-step


    

In [ ]:

    

while True:
    rate(100)
#     observation.plot(data=mag(p/m))

    # Update all positions
    pos = pos+(p/m)*dt

    r = pos-pos[:,newaxis] # all pairs of atom-to-atom vectors
    rmag = sqrt(sum(square(r),-1)) # atom-to-atom scalar distances
    hit = np.less_equal(rmag,radius+radius[:,None])-np.identity(Natoms)
    hitlist = np.sort(np.nonzero(hit.flat)[0]).tolist() # i,j encoded as i*Natoms+j

    # If any collisions took place:
    for ij in hitlist:
        i, j = divmod(ij,Natoms) # decode atom pair
        hitlist.remove(j*Natoms+i) # remove symmetric j,i pair from list
        ptot = p[i]+p[j]
        mi = m[i,0]
        mj = m[j,0]
        vi = p[i]/mi
        vj = p[j]/mj
        ri = Atoms[i].radius
        rj = Atoms[j].radius
#         a = mag(vj-vi)**2
        a = (vj-vi).dot(vj-vi)
        if a < 1e-3:
            continue # exactly same velocities
        b = 2*dot(pos[i]-pos[j],vj-vi)
        c = (pos[i]-pos[j]).dot(pos[i]-pos[j]) - (ri+rj)**2
        d = b**2-4.*a*c
        if d < 0:
            continue # something wrong; ignore this rare case
        deltat = (-b+math.sqrt(d))/(2.*a) # t-deltat is when they made contact
        pos[i] = pos[i]-(p[i]/mi)*deltat # back up to contact configuration
        pos[j] = pos[j]-(p[j]/mj)*deltat
        mtot = mi+mj
        pcmi = p[i]-ptot*mi/mtot # transform momenta to cm frame
        pcmj = p[j]-ptot*mj/mtot
        rrel = math.sqrt((pos[j]-pos[i]).dot(pos[j]-pos[i]))
        pcmi = pcmi-2*dot(pcmi,rrel)*rrel # bounce in cm frame
        pcmj = pcmj-2*dot(pcmj,rrel)*rrel
        p[i] = pcmi+ptot*mi/mtot # transform momenta back to lab frame
        p[j] = pcmj+ptot*mj/mtot
        pos[i] = pos[i]+(p[i]/mi)*deltat # move forward deltat in time
        pos[j] = pos[j]+(p[j]/mj)*deltat
 
    # Bounce off walls
    outside = np.less_equal(pos,Ratom) # walls closest to origin
    p1 = p*outside
    p = p-p1+abs(p1) # force p component inward
    outside = np.greater_equal(pos,L-Ratom) # walls farther from origin
    p1 = p*outside
    p = p-p1-abs(p1) # force p component inward

    # Update positions of canvas objects
    for i in range(Natoms):
        Atoms[i].pos = vec(*pos[i])


    

In [ ]: