You are on page 1of 24

# Part 4: Diffusion and Hops

From Discrete to Continuous Hopping on a Lattice Notation: Even and Odd From one step to many An example Continuous Langevin equation An integration Gaussian Properties Generating Function A probability Discrete A generating Function A probability calculation we get an answer! fourier transform formulation binomial theorem Higher Dimension continued probability density One dimension Current Diffusion equation Higher Dimension You cannot go back
Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1.6 9/11/09 Leo Kadanoff

1

Hopping: From Discrete to Continuous
We are going to be spending some time talking about the physics of a particle moving in a solid. Often this motion occurs as a set of discrete hops. The particle gets stuck someplace, sits for a while, acquires some energy from around it, hops free, gets caught in some trap, and then sits for a while. I’m going to describe two mathematical idealizations of this motion: discrete hopping on a lattice and continuous random motion. One point is to see the difference between the two different topologies represented by a continuous and a discrete system. One often approximates one by the other and lots of modern physics and math is devoted to ﬁguring out what is gained and lost by going up and back. There is a ﬁne tradition to this. Boltzmann, one of the inventors of statistical mechanics, liked to do discrete calculations. So he often represented things which are quite continuous, like the energy of a classical particle by discrete approximations, A little later, Planck and Einstein had to ﬁgure out the quantum theory of radiation, which had been thought to be continuous, in terms of discrete photons. So we shall compare continuous and discrete theories of hopping.

Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1.6 9/11/09 Leo Kadanoff

2

σj =1 or σj = -1.. However. One way of doing this can be labeled by giving the position r=(n1. In one step of motion one progress from X(t) to X(t+ τ) = X(t)+ aσj.n2. the position is X(t). If we include all possible values of these integers. We could do this in any number of dimensions. We assume that we start at zero. the particular lattice generated is called is called the simple hypercubic lattice. Thus. We show a picture of this lattice in two dimensions. where n is an integer. but of course the average its square is non-zero and is given by < σj2 > =1.. so that our times t =j τ.)a where the n’s are integers. where σj is picked at random from among the two possible nearest neighbor hops along the lattice. At any given time.Hopping On a Lattice A lattice is a group of sites arranged in a regular pattern. This section is devoted to developing the concept of a random walk. We do this to emphasize that geometric problems can often be expressed in algebraic form and vice versa. which lies on one of the a lattice sites. < σj >=0. It is not accidental that we express the random walk in the same language as the Ising model...6 9/11/09 Leo Kadanoff 3 . we shall approach it is the simplest possible way by ﬁrst working it all out in one dimension and then stating results for higher dimensions A random walk is a stepping through space in which the successive steps occur at times t=M τ. x=an. Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1.

We shall also use the dimensional variables for time t=M τ and for space X(t)=na. the number of steps taken and n. the displacement from the origin after M steps. The general formula is M n= k=1 σk Notice that n is even if M is even and odd if M is odd.6 9/11/09 Leo Kadanoff 4 . usually x. When we need a nonrandom spacial variable. We start from n=0 at M=0. We use a capital X to remind ourselves that it is a random variable. calculation. we shall use a lower case letter. Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1. M.Notation: Even and Odd We represent the walk by two integers. detailed. We shall have to keep track of this property in our later.

6 9/11/09 Leo Kadanoff 5 . cover much ground. but the typical magnetization is only proportional to the square root of that number. the maximum magnetization is proportional to the number of spins. much smaller than maximum deviations. on each step the walker goes left as much as right.1 However.From one step to many steps We start from the two statements that <σj >=0 and that <σj σk >=δj.2 € Our statement is the same that in a zero ﬁeld uncoupled Ising system. Thus. which is the typical end-to-end distance of this random walk. since < X (t )2 >= a 2 ∑ < σ j σ k >= a 2 ∑ δ j . and thus as a result the average displacement of the entire walk is zero < X (t ) >= a∑ < σ j >= 0 j =1 M iv.k =1 M M iv. Typical ﬂuctuations are much. in net. In that case we would have had a distance aM. a random walk does not.k = a 2M € j .k On the average. Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1. We can see this fact by noting that the root mean square average of X2 is a √M.k =1 j . This distance is is much smaller than the maximum distance which would be covered were all the steps to go in the same direction. of course the mean squared displacement is not zero.

Consider a random walk that consists of M= one million steps steps. That is quite a difference! Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1. On the other hand the typical end to end distance of such a walk is M 1/2 centimeters or ten meters.An example: The difference between maximum length and RMS length (root mean square length) needs to be emphasized by an example.6 9/11/09 Leo Kadanoff 6 . If each step is one centimeter long the maximum distance traveled is M centimeters. or 10 thousand meters or 10 kilometers.

c.5 has the solution t X(t) = X(s) + s du η(u) 7 Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1.4 We would like to have a simple differential equation which appropriately reﬂects this physics. except that we have to ﬁx the value of the constant. The simplest form of differential equation would be a form described as a Langevin equation.3 iv. dX/dt = η(t) iv. to agree with what we did before.5 where η(t) is a random function of time that we shall pick to be uncorrelated at different times and have the properties <η(t)> =0 and <η(t) η(s)> =c δ(t-s)δj. Equation iv.k This looks like what we did in the random hopping on the lattice.6 9/11/09 Leo Kadanoff .Continuous Random Walks: Langevin Equation The physics of this random motion is well represented by the two equations <X(t)-X(s)> = 0 <[X(t)-X(s)]2> = a2 |t-s|/τ iv.

€ so that a comparison with our previous result indicates that we should choose € c= a2 /τ ` iv. that as before <R(t)-R(s)> =0. A new result is that for t>s.t X(t) = X(s) + s du η(u) Using this result.8 8 . we can calculate. we ﬁnd: (what happens for s>t?) < [X (t ) − X (s )]2 > = = = More speciﬁcally.7 so that < [X (t ) − X (s )]2 > = a 2 | t − s | /τ Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1.6 9/11/09 Leo Kadanoff ∫ du ∫ dv < η (u )η (v ) > s s t t ∫ du ∫ dv c δ(u − v ) ∫ du c = (t − s )c s s t s t t < [X (t )]2 >= c t iv.6 iv.

Its average is zero and its variance is a2t/τ . As an extra we can exhibit the generating function for this walk: € < exp[iqX (t )] >= e −q 2a2t /(2τ ) € Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1. of many pieces which are uncorrelated with one another. According to the central limit theorem. such a sum or integral is a Gaussian random variable.Gaussian Properties of Continuous Random Walk In the continuous case.9 < ρ(X (t ) = x ) >=   e 2 2πa t  Now we have said everything there is to say about the continuous random walk. we know everything there is to know about it. it has a probability distribution  τ 1/2 − x 2τ /(2a2t ) iv. X(t) is composed of a sum. Hence. Consequently. or rather integral.6 9/11/09 Leo Kadanoff 9 .

Z. Thus the ﬁnal result is <exp(iqX(t))>= [cos(qa)]M. If one knew the value of < exp(iqX(t))> it would be easy to calculate all averages by differentions with respect to q and even the probability distribution of X(t) by. The same approach works for the random walk. iv. a Fourier transform technique. the cosine given here. a generating function from which we can determine all the statistical properties. We start from the simplest such problem. Take t=τ so that there is but one step in the motion and < exp(iqX(τ))>= < exp(iqa σ1(t))>= [exp(iqa) + exp(-iqa)]/2 = cos qa Since X(t) is a sum of M identical terms.9 As we know from our earlier work such a generating function enables us to calculate. the traditional approach to the problem is to calculate the partition ﬁunction.In our calculation of the statistical mechanics of the Ising model. as we shall see. It is best to start on this calculation by a somewhat indirect route: The spirit of the calculation is that it is simple to ﬁnd the average of exp(iqX(t)) where q is a parameter which we can vary and X(t) is our random walk-result. the exponential is a product of M terms of the form < exp(iqa σj(t))> and each term in the product has exact the same value.6 9/11/09 Leo Kadanoff . as for example <exp(iqX(t))>= [cos(qa)]M at q=0 <1>=1 d/dq { <exp(iqX(t))>= [cos(qa)]M } at q=0 (d/dq)2 { <exp(iqX(t))>= [cos(qa)]M } at q=0 <iX(t)>=Ma sin qa =0 -<[X(t)]2>=-Ma2 d/dq sin qa = -Ma2 10 A generating function: discrete case Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1. or of any other problem involving the statistics of large numbers of particles.

6 9/11/09 Leo Kadanoff 11 .3) < f (X ) >= ∑ f (na)ρn This is an important deﬁnition. We can use this probability to calculate the value of any function of X in the form (see equation i. ρn which is the probability X will take on the value na. So ρ tells us everything there is to know about the problem.m. The formula that we shall use is only slightly more complex. and the probability of it having the value na is ρn .A Probability We wish to describe our result in terms of a probability. € Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1. then the average is the sum over n times the value of the function at that n times the probability of having that value.10 so that the average is once more the sum of the function times the probability. The formula for the average is then < f (X (mτ )) >= ∑ f (na)ρn . It says that if X takes on the values na. We deﬁne X as being time dependent: the probability that X(t) has the value na at time mτ is then deﬁned to be ρn.m n n € iv.

. Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1. -M Thus there are a total of M possible sites which could be occupied..-M+2. since we have had.. M-2. just before -M+2.. For M being even the possible sites are M..-1. .M iv. so we ﬁnd € n I’m going to spend some time on this result. M-2.... For M being odd the possible sites are M..To calculate probability Start from < f (X ) >= ∑ f (na)ρn n choose the function to be an exponential.. ..6 9/11/09 Leo Kadanoff < [cos q ]M >= ∑ exp(iqn )ρn .... It gives the probability that after a random walk of M steps a walker will ﬁnd itself at a site n=M-2j steps of its starting point. by this time. let’s say what it is. -M. .1.t /τ n We have already evaluated the left hand side of this expression.0 . Is it true then that an average probability of occupation among these sites is 1/M? or maybe 1/ (2M) What’s a typical occupation? What is that going to be? Recall that the RMS distance covered is . The pattern is € slightly strange. First. considerable experience dealing with exponentials € < exp(iqX (t ) / a) >= ∑ exp(iqn )ρn .. .11 12 ..

We simply use the binomial theorem n M! M (a + b) = aM bM −j € j!(M − j)! j=0 < [cos q ]M >= ∑ exp(iqn )ρn .11 to expand the left hand side as eiq + e−iq M M! eiqM −i2q j ( ) =∑ e M 2 j j!(M − j)! 2 to get. A formula..We get an answer! n Here M is the number of steps.M = 2−M j!(M − j)! ρn.. M! ρM−2 j. We do not even have to do a Fourier transform to evaluate the probability.13b We are a bunch of theorists.M M! = M +n M −n 2−M ( 2 )!( 2 )! iv..M iv.13a iv.6 9/11/09 Leo Kadanoff 13 . We have an answer.12 iv.. Now what can we do with that answer? What can we ﬁgure out from here? Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1.

6 9/11/09 Leo Kadanoff 14 .p 2π To obtain π −π dq −inq e ρp. Therefore one can also write π/2 dq −inq e [cos q]M ρn.M eiqp = ρn.Fourier transform ∞ ρp.M = 2π p=−∞ ∞ π −π dq −inq e [cos q]M 2π Because n=M (mod 2).M eiqp = [cos q]M p=−∞ iv.M = 2 −π/2 2π We have come across this kind of integral before. Now what? Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1. the integrand at q is just the same as the integrand at q+π.11 To invert the Fourier transform use π −π dq i(p−n)q e = δn.

Result From Binomial Theorem M! ρM−2 j. n close to -M.M M! = M +n M −n 2−M ( 2 )!( 2 )! We are a bunch of theorists.6 9/11/09 Leo Kadanoff 15 . Now what can we do with that answer? What can we ﬁgure out from here? sterling approximation. n close to M. A formula. n=0. M! ≈ M M e-M (2 π M )1/2 what happens for n=M.M = 2−M j!(M − j)! ρn. We have an answer. for large M. n close to zero? Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1. n=-M.

. but different signs for example a(1. a. If all hopping occurs from one site to a nearest neighbor site. the coordinate hops through one of the nearest neighbor vectors..)/2.. Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1. .)/2 or they are all odd. We use a lattice constant.. There is no ambiguity about how things very close to one another behave. -1. n3.. have components which behave entirely independently of one another. m3. where each of the m’s have magnitude one.n2. and exactly the same as the one-dimensional coordinate we have treated up to now. -4.... This particular choice makes the entire coordinate.-1. They are either at the same point or different points So now I would like to talk about a random walk in a d-dimensional system by considering a system on a simple lattice constructed as in the picture.3. x = a(0.). There are two possible kinds of assignments for the n’s: Either they are all even e..)/2 . ξα . The n’s are integers.m2..g. The lattice sites are given by x = a(n1.Higher Dimensions Reason for putting many different calculations on a lattice is that the lattice provides a simplicity and control not available in a continuum system.)/2.. for example x = a(1..6 9/11/09 Leo Kadanoff 16 . . which one being choosen at random. We then choose to describe the system by saying that in each step. x.-1. . which is twice as big as the one shown in the picture. . the hops are through one of 2d vectors of the form ξα= a(m1..2. .. because things cannot get very close.

mit.edu/.ﬁles./CourseHome/index.6 9/11/09 Leo Kadanoff 17 ..RANDOM WALKS dspace.htm http://particlezoo..png Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1.wordpress.com/2008/09/randomwalk.

M τ = α=1 ρ1 α . We have d ρd an.continued We denote the probability density of this d-dimensional case by a superscript d and the one for the previous one-dimensional case of by a superscript 1...6 9/11/09 Leo Kadanoff € .Higher Dimensions. As an additional difference.... After a while. we can jump directly to the answer for the continuum case. viz r2 = x2 +y2 +. The one-dimensional equation answer in eq iv..9 Here the bold faced quantities are vectors. we shall focus entirely upon the higher dimensional case and therefore drop the superscripts. the d-dimensional object will be a function of space and time rather than n and M. Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1.M n However.9 was  τ 1/2 − x 2τ /(2a2t ) < ρ(X (t ) = x ) >=   e 2 2πa t  so that the answer in d dimensions must be €  τ d /2 −r 2τ /(2a2t ) < ρ(R(t ) = r) >=   e 2 2πa t  18 iv. If the probability distribution for the discrete case is simply the product of the onedimensional distributions so must be the continuum distribution..

€ in separate exponents to get r2 = x2 +y2 +. Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1..... the result come from what is called a diffusion process and is typical of long-wavelenth phenomena in a wide variety of systems. coming from adding the x2 and y2 and.Higher dimensional probability density: again one dimension  τ  < ρ(X (t ) = x ) >=   2 2πa t  1/2 d dimensions is a product of ones  τ d /2 −r 2τ /(2a2t ) < ρ(R(t ) = r) >=   e 2 2πa t  e − x 2τ /(2a2t ) Notice that the result is a rotationally invariant quantity.6 9/11/09 Leo Kadanoff 19 . This is an elegant result coming from the fact that the hopping has produced a result which is independent of the lattice sitting under it. In fact. ... We woulod get a very similar result independent of the type of lattice underneath.

M / ∂n = 0 Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1. and the conservation law is written as conservation law ∂ρ(r . Diffusion has a second element: locality.M +1 conservation law This equation says that the change of probability over one time step is produced by the ﬂow of probability in from the left minus the ﬂow out to the right. On a one dimension lattice.6 9/11/09 Leo Kadanoff 20 € .t ) = 0 ∂t The time derivative of the density is produced by a divergence of the curent ﬂowing into a point.M / ∂M + ∂In . the rule takes the simpler form: ρn . The local amount of Q. changes because things ﬂow into and out of a region of space.M ≈ ∂ρn . called ρ. € Here.M − In +1/2.M€ − ρn .t ) + ∇ ⋅ j (r . The ﬂow is called a current.M = In −1/2. In our case the Q in question is the total probability of ﬁnding the diffusing particle someplace. j. dQ/dt=0.Diffusion Process Diffusion here is a result of a conservation law: a global statement that the total amount of something is unchanged by the time development of the system.M / ∂M The conservation law then takes the form ∂ρn . I shall assume that the initial probabilities is sufﬁciently smooth so that even and odd n-values have rather similar occupation probabilities so that we can get away with statement like ρn . n. I’m going to visualize a situation once more in which we have a discrete time coordinate M and a discrete space coordinate.M +1 − ρn .

m − ρn +1.The Current We do not have a full statement of the of what is happening until we can specify the current.6 9/11/09 Leo Kadanoff 21 € .m = −(∂ρn . and σn+1 is equal to -1. we write the difference in terms of a derivative.M / ∂ n 2 / 2 which can be written in dimensional form as € ∂ρ ∂2ρ diffusion equation =λ ∂t ∂ x2 with the diffusion coefﬁcient being given by λ=a2/(2 τ). Since there is a probability 1/2 for each of the σ-events the value of I is In +1/2. On the other hand it is given a contribution -1 when the site at n +1 is occupied at time M. An approximate deﬁnition of a current in a conservation law is called a constitutive equation. Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1. getting € In .m = (ρn . We now write this down.m ) / 2 constitutive equation Once again.M is given a contribution +1 when the site at n is occupied at time M. Our hopping model says that a In+1/2. and σn is equal to +1.M / ∂M = ∂2ρn .m / ∂n ) / 2 constitutive equation This can then be combined with the conservation law to give the diffusion equation € ∂ρn .

t ) € Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1.t ) so that the diffusion equation becomes € ∂t ρ(r.Higher Dimension In higher dimensions.t ) = λ∇2ρ(r. the current is proportional to the gradient of the current j (r .6 9/11/09 Leo Kadanoff 22 .t ) = −λ∇ρ(r .

This general principle is much used in the context of quantum ﬁeld theory and condensed matter physics. Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1. named for two contemporary physicists. In fact this is part of a general principle which permits only slow changes as a result of a conservation law. The idea is connected with the construction of the kind of particle known as a Nambu-Goldstone boson.Diffusion Equation This equation is one of several equations describing the slow transport of physical quantities from one part of the system to another. my Chicago colleague Yoichiro Nambu and the MIT theorist Jeffrey Goldstone. the conservation law guarantees that the rate of change in time will also be slow.6 9/11/09 Leo Kadanoff 23 . When there is slow variation is space.

What happens for t= -2? Boltzmann noted that equations of classical mechanics make sense if t is replaced by -t. Small rapidly varying errors at t=0 will produce small errors for positive t and huge errors for negative t. Plot solution for t=0. Where did we go from sense to nonsense? Perimeter Institute Lecture Notes on Statistical Physics part 4: Diffusion and Hops Version 1.4. Use λ=2.6 9/11/09 Leo Kadanoff ∫ ∫ 24 . Information gets lost as time goes forward.0) =∂X F(x. Find solution from F(x. Use c=2 A global solution to the diffusion equation is ρ(x. It enables you to look forward or back inﬁnitely far in the future or the past without losing accuracy.0) =1 for 0<x<1 and F(x. But diffusion equation has a solution which does not make sense.t)=G(x-ct)+H(x+ct) This is a global solution.t)= dk exp[ikx-λk2t] g(k) with g(k)= dx exp[-ikx] ρ(x.0) =0 otherwise.0)/(2π) as initial data.Solution to Diffusion Equation Cannot be Carried Backward in Time The wave equation is (∂t2 -∂X2 )F = 0 Its general solution is F(x.2. You cannot extrapolate backward in time. Solve for ρ(x.0) =∂X F(x.0) =0 otherwise.0)=1 for 0<x<1 and ρ(x.