Analytical Methods of calculating end to end Mean First

Encounter Times of Polymer Chains

Soham Sawant
Indian Institute of Technology Bombay
[email protected]

Abstract
We present a detailed study of various analytical approaches for calculating the Mean First En-
counter Time (MFET) between the ends of a polymer chain. Our analysis begins with an introduction
to the Rouse model, a fundamental framework for describing polymer dynamics. Building on this,
we derive an explicit expression for the MFET within the Rouse model using normal mode analysis.
We then explore approximate methods for evaluating the MFET, focusing on the Szabo-Schulten-
Schulten (SSS) approximation and the Wilemski-Fixman (WF) theory, both of which simplify the
complex many-body problem into more tractable forms.Following this, we extend our discussion to
more realistic polymer systems by investigating the effects of excluded volume interactions. In par-
ticular, we use the SSS approximation to study the MFET in self-avoiding polymers and analyze
its non-monotonic dependence on chain compactness. This behavior reveals subtle physical insights
about the balance between entropic stretching and excluded volume effects in determining encounter
dynamics.Through these methods, we highlight how different theoretical frameworks capture the
underlying physics of polymer looping, and we emphasize the limitations and applicability of each
approach.

1 Introduction
The Mean First Encounter Time (MFET) of polymer chain ends is the average time it takes for the
two ends of a polymer to meet for the first time. This might sound like a small event, but it plays a
very big role in many important systems. In biology, for example, the ends of DNA or chromatin chains
can come together to control how genes are turned on or off. In proteins, the meeting of chain ends
can trigger the folding process that gives a protein its working shape. In man-made polymers, MFET
helps us understand how long it takes for a single chain to loop back on itself, or for chains to link up
to form networks. This is useful in designing materials that need to change shape, heal themselves, or
stick together in a certain way. By knowing MFET, scientists can better control reactions in materials or
biological systems by adjusting chain length, temperature, or how flexible the chains are from a science
perspective, MFET is a great way to study how tiny random movements of molecules can lead to bigger
changes. It connects ideas from physics, chemistry, and biology in a useful way.
In short, knowing when the ends of a polymer meet helps us understand how molecules behave, react,
and organize themselves. It’s a simple concept with powerful applications in everything from cell biology
to advanced materials.

2 The Rouse Model

2.1 The Gaussian Chain
The Gaussian chain model assumes that each bond vector is Gaussian distributed. The distribution is
given by (1):
  23
3 3r 2
P (r) = e− 2b2 (1)
2πb2
The Gaussian chain can be visualized as a sequence of beads connected by harmonic springs of
strength k, where k = 3kbB2 T , as illustrated in Fig. 1. This model serves as the foundation for the Rouse
chain. The potential energy of the chain is given by:

1
 Figure 1: Schematic representation of a Gaussian chain as beads connected by harmonic springs.

n n
X 1 X 1
Φ(r) = kri2 = k(Ri+1 − Ri )2 (2)
i=1
2 i=1
2
Here, Ri denotes the position of the i-th bead.

2.2 The Rouse Chain

Consider a Gaussian chain consisting of N beads connected by N − 1 springs, each with spring constant
k = 3kbB2 T . If we examine the dynamics of a single bead while holding the others fixed, we find that the
external field Φ acting on that bead is generated by its connections to neighboring beads.
We assume each bead experiences the same friction coefficient ζ, moves in the overdamped regime
(neglecting inertial effects), and has a position-independent diffusion coefficient D = kBζ T . This setup
defines the Rouse chain model.
The dynamics of the beads are described by the following Langevin equations:
dR1 3kB T
=− (R1 − R2 ) + f1 (3)
dt ζb2
dRn 3kB T
=− (2Rn − Rn−1 − Rn+1 ) + fn (4)
dt ζb2
dRN 3kB T
=− (RN − RN −1 ) + fN (5)
dt ζb2
Note that the equations for the terminal beads differ from those for the internal beads, as they are
influenced by only one neighboring spring.
The random forces fn (t) are stochastic, with the following statistical properties:

⟨fn (t)⟩ = 0 (6)

′ ′
⟨fn (t)fm (t )⟩ = 2Dδnm δ(t − t ) (7)

2.3 Normal Mode Analysis

Equations (3)–(5) are 3N coupled stochastic differential equations. To solve them, we first ignore the
stochastic forces fn and try specific solutions of the form(1):

Rn (t) = X(t) cos(an + c) (8)

Using this ansatz and the boundary conditions at n = 1 and n = N , we obtain the values of a and
c. We can represent Rn as a sum of normal modes:
N −1   
X pπ 1
Rn = X0 + Xp cos n− (9)
p=1
N 2
N −1   
1 X pπ 1
cos n− = δp0 (10)
N n=0 N 2
N −1   
1 X pπ 1
Xp = Rn cos n− (11)
N n=0 N 2

2
 The equation of motion for a normal mode is:
dXp 3kB T 2
 pπ 
=− 4 sin X p + Fp (12)
dt ζb2 2N
Where kp = 4k sin2 (pπ/2N ). The noise properties are:

⟨Fp (t)⟩ = 0 (13)

D
⟨Fp (t)Fq (t′ )⟩ = δpq δ(t − t′ ) (14)
N
2D
⟨F0 (t)F0 (t′ )⟩ = δ(t − t′ ) (15)
N
1
X0 is the center-of-mass coordinate which diffuses with a diffusion constant scaling as N:
Z t
X0 (t) = X0 (0) + dτ F0 (τ ) (16)
0
6D
gcm (t) = ⟨(X0 (t) − X0 (0))2 ⟩ =
t (17)
N
All other modes (p = 1, 2, . . . , N − 1) perform independent vibrations governed by:
Z t
−t/τp
Xp (t) = Xp (0)e + dτ e−(t−τ )/τp Fp (τ ) (18)
0

ζb2
 h  pπ i−1
τp = 4 sin2 (19)
3kB T 2N
ζb2 N 2 1
For small wavelengths (p/N ≪ 1), τp ≈ 3π 2 kB T · p2 .
The correlation ofXp (t) -
 
t
⟨Xp (t) · Xp (0)⟩ = ⟨Xp2 ⟩ exp − (20)
τp
We now calculate the equilibrium expectation values of Xp2 i.e., the amplitudes of the normal modes. To
this end, first consider the statistical weight of a configurationR1 , ..., RN in Cartesian coordinates(1).
N −1
!
1 3 X 2
P(R1 , . . . , RN ) = exp − 2 (Rn+1 − Rn ) (21)
Z 2b n=1

N −1 N −1
X pπ 1 pπ 1 X pπ pπn
Rn+1 − Rn = XP [cos( (n + )) − cos( (n − ))] = − XP [2sin( )sin( )] (22)
p=1
N 2 N 2 p=1
2N N
N −1 −1
X X NX
X
2 pπn qπn pπ qπ
(Rn+1 − Rn ) = 4Xp Xq sin( )sin( )sin( )sin( ) (23)
n=1 p q n=1
N N 2N 2N
PN −1
Since n=1 Xp Xq sin( pπn qπn
N )sin( N ) = δpqN .

N −1
!
1 6N X  pπ 
P (X0 , . . . , XN ) = exp − 2 Xp · Xp sin2 (24)
Z b p=1 2N

where Z is a normalization constant .

b2 N b2 1
⟨Xp2 ⟩ =  pπ  ≈ 2 · 2 (25)
4N sin2 π p
2N
Again, the last approximation is valid whenn p << N .

3
2.4 Correlation of the end to end vector
N
X −1  pπ  X  pπ 
R = RN − R1 = Xp [(−1)p − 1] cos = −2 Xp cos (26)
p=1
2N 2N
p∈odd

We need to calculate < R(t).R(0) >

XX  pπ   qπ 
< R(t).R(0) >= 4 < Xp (t)Xq (0) > cos cos (27)
p q
2N 2N
Z t
> −t/τp
< Xp (t).Xq (0) >=< Xp (0).Xq (0) e + dτ e−(t−τ )/τp < Fp (τ ).Xq (0) >=< Xp2 > e−t/τp δpq (28)
0
Which further implies that -
N −1
X pπ
< R(t).R(0) >= 4 Xp (0)2 e−t/τp cos2 ( ) (29)
p=1
2N

pπ
Since Xp (0)2 ∝ p12 the higher modes do not contribute much generally.So we can take cos2 ( 2N ) ≈ 0 for
high N . Thus we can write a normalized version of this correlation .
P∞ − t
τp
e
< R(t).R(0) > p=1 p2
= P∞ 1 (30)
< R(0)2 > p=1 p2

In most cases we just take this value to be e−t/τp .

2.5 Mean Square Displacement of Segments in the Rouse Model

In this section we calculate the mean square displacement gseg (t) of the individual segments. Using
Eq. (9) and the fact that different modes are uncorrelated, we get for segment n:

N −1   
X pπ 1
⟨(Rn (t) − Rn (0))2 ⟩ = ⟨(X0 (t) − X0 (0))2 ⟩ + ⟨(Xp (t) − Xp (0))2 ⟩ cos2 n− (31)
p=1
N 2

Averaging over all segments , the mean square displacement of a typical segment in the Rouse model
is as follows:
N
1 X
gseg (t) = ⟨(Rn (t) − Rn (0))2 ⟩
N n=1
N
(32)
D X  
=6 t+ ⟨X2p ⟩ 1 − e−t/τp .
N p=1

Two limiting cases can be identified. First, for very large times t ≫ τ1 , the first term in Eq. (32)
dominates, yielding:
D
gseg (t) ≈ 6
t (t ≫ τ1 ). (33)
N
This result is consistent with the fact that the polymer as a whole diffuses with diffusion coefficient
DG . Secondly, when t ≪ τ1 , the sum over p in Eq. (32) dominates. If N ≫ 1, the relaxation times
can be approximated using Eq. (19), the Rouse mode amplitudes using Eq. (25), and the sum can be
approximated by an integral:

4
 Z ∞
b2 1  −tp2 /τ1

gseg (t) = dp 1 − e
π2 N 0 p2
2 Z t Z ∞
b ′ 2
= 2 dt′ dp e−t p /τ1
π N τ1 0 0
Z t (34)
b2 1 p
= 2 dt′ πτ1 /t′
π N τ1 0 2
1/2
12kB T b2

= t1/2 (τN ≪ t ≪ τ1 , N ≫ 1).
πζ
Thus, at short times, the mean square displacement of a typical segment is subdiffusive with an
exponent of 1/2, and is independent of the number of segments N in the chain. For extremely short
times, t << τN (e−t/τp = 1 − t/τp ), -
N
X t
gseg (t) ≈ <Xp2 > = 3Dt (35)
p=1
τp

So at really small time scales segmental motion is diffusive .

3 Computation of the Mean First-Encounter Time Between the

Ends of a Polymer Chain
We will use a novel theoretical approach which was used in the paper by Holcman (2 ) . This is a general
theory that will help us compute the MFET without any major simplifications .The mean first-encounter
time (MFET) is defined as the first arrival time for the end monomer into a ball of radius ϵ, centered at
the other polymer end .MFET does not depend only on the radius ϵ but also on the polymer length N .
As we have seen before when we use the SSS approximation , MFET depends on the initial√end-to-end
3/2
distribution . It scales
√ as N and depends on ϵ . This approximation was only valid when N ϵ/b is of
order 1 . But when N ϵ/b >> 1 this approximation (SSS) is no longer valid and the MFET depends
on the relaxation time of the rouse modes (τp ) which is proportional to N 2 and is independent of ϵ . So
MFET shows mixed scaling laws in different regimes , using this method we will be able to show both
regimes in a single expression .

3.1 End to End encounter in the configuration space

We define a Rouse chain made up of N monomers at the points Rn (n=1,2,..N) and driven by independent
Brownian motions in a force generated by the harmonic potential .
N
κX 2
ΦRouse = (Rn − Rn−1 ) (36)
2 n=2

Each monomer interacts only with its two neighbors (except for the end points). We neglect all the
other possible interactions such as hydro- dynamics and allow the polymer to cross itself. The Langevin
equation is given by -
dRn √ dwn
= −D∇Rn ΦRouse + 2D , for n = 1, . . . , N (37)
dt dt

5
Where wn are independent d - dimensional brownian motions with mean zero and variance one .
In the normal Rouse coordinates eqn(9)-
N −1   
X pπ 1
Rn = u0 + up cos n− (38)
p=1
N 2

The condition for encounter is given as -

|RN − R1 | < ϵ (39)

This can be reduced to - X  pπ 

|2 up cos |<ϵ (40)
2N
p odd

The end-to-end encounter is independent of the coordinate u0 , which is the center of mass. Thus, the
MFET becomes the mean first passage time for the (N-1)d-dimensional stochastic dynamical system -

u(t) = (u1 (t), . . . , uN −1 (t)) ∈ Ω × Ω..... × Ω = Ω̃, where Ω ⊂ R2 or R3 (41)

Here u(t) is basically space of all normal modes .

dup p dw̃p
= −κp Dp up + 2Dp (42)
dt dt
Where κp = 4ksin2 (pπ/2N ) and Dp = D.
 
 X  pπ  
S = (u1 , . . . , uN −1 ) ∈ Ω̃ | up cos =0 (43)
 2N 
p odd

S is a submanifold in Ω̃ which basically gives us the space in which the reaction occurs . But we defined
a radius in which we the reaction occurs or the polymer ends meet . Thus we define a boundary -
 
ϵ
Sϵ = P ∈ Ω̃ | dist(P, S) < √ (44)
2

We define a probability density function p(u(t) = x, t) which satisfies the Fokker Planck Equation -

1 ∂p(x, t)
= ∇2 p(x, t) + ∇ · (ϕp ∇p(x, t)) = Lp (45)
D ∂t
Where p(x, 0) = p0 (x) = N e−ϕp and ϕp = p 21 κp u2p .Here L is the forward Fokker Planck Operator .
P
p(x, t) = 0 for x ∈ ∂Sϵ , which just means that we impose an absorbing boundary condition when we
reach the submanifold which implies that the the two ends meet for the first time .The solution to this
forward Fokker Planck equation can be expanded as (3) -
∞ ∞
X ϵ X ϵ
p(x, t) = ai wλϵi (x)e−λi tD peq (x) = ai wλϵi (x)e−λi tD e−ϕ(x) , (46)
i=0 i=0

Where wλϵi (x) and λi are the eigenfunctions and the eigenvalues of the Backward Fokker Planck
Operator , ϵ denotes that L† is in Ωϵ = Ω̃ − Sϵ . Holcman makes a mistake here by stating that these
are the eigenfunctions of the forward Fokker Planck operator in his orignal paper (2). But it can be seen
in the reference (3) that it should be the backward Fokker Planck Equation which is given by -

1 ∂p(x, t)
= ∇2 p(x, t) + ∇ϕp · (∇p(x, t)) = L† p (47)
D ∂t
Now the probability distribution that the two ends have not met till time t is -
Z
p(t) = p(x, t)dx (48)
Ωϵ

6
The equilibrium distribution is given by -

e−ϕp (x)
p0 (x) = (49)
|Ω̃|

Since the eigenfunctions of the Backward Fokker Planck Operator are orthogonal -
Z
ai = p0 (x)wλϵi (x)dx (50)
Ω̃

So we can write p(t) as - X ϵ

p(t) = Ci e−λi Dt (51)
i
Z Z
Ci = p0 (x)w (x)dx
λϵi wλϵi (x)e−ϕ(x) dx (52)
Ωϵ Ωϵ
Z 2
e −1
Ci = |Ω| wλϵi (x)e −ϕ(x)
dx (53)
Ωϵ

The first-passage time density f (t) is related to the survival probability p(t) = P rob{τ > t} as follows:

d
f (t) = − p(t) (54)
dt
Therefore, the mean first-passage time ⟨τ ⟩ is:
Z ∞
⟨τ ⟩ = tf (t) dt (55)
0

Now using integration by parts, we rewrite the expression for ⟨τ ⟩:

Z ∞  
dp
⟨τ ⟩ = t − dt (56)
0 dt

Applying the integration by parts formula:

Z ∞   Z ∞
dp ∞
t − dt = [−tp(t)]0 + p(t) dt (57)
0 dt 0

Assuming that limt→∞ tp(t) = 0 and p(0) = 1, we obtain:

Z ∞ Z ∞ ∞
X ϵ X Ci
⟨τ ⟩ = p(t) dt = Ci e−λi Dt dt = (58)
0 i 0 i=0
Dλϵi

In practice, only the first few eigenvalues and corresponding coefficients are retained when approximating
the first-encounter time (FET) distribution. This simplification is both physically justified and supported
by numerical observations. We will see when we will solve the backward Fokker Planck λ0 < λ1 < λ2 ... so
the higher eigenvalues die out quicker in p(t) so they will not contribute much . Empirical results confirm
this approximation: for polymers of moderate length (N=16,32), a single exponential term accurately
fits the FET distribution, indicating dominance of the lowest eigenvalue λ0 . For longer chains (N=64),
inclusion of a second exponential term improves the fit, capturing early-time behavior associated with
λ1 . But if we increase the capture radius too much (eg ϵ = 0.4b) then the short time behavior also
becomes important so we need to include higher terms as well .

3.2 Solutions of the Backward Fokker Planck Operator

We will now solve the Backward FPE -
1 ∂p(x, t)
= ∇2 p(x, t) + ∇ϕp · (∇p(x, t)) = L† p (59)
D ∂t

7
 P∞
Where p(x, t) = i=0 ai wλi (x)e−λi tD e−ϕ(x)

∂p(x, t)
= −λi Dp(x, t) (60)
∂t
X
∇2 p(x, t) = ai e−λi Dt e−ϕp (x) [∇2 wλi (x) − 2(∇wλi (x)).(∇ϕp ) − wλi (x)∇2 ϕp + wλi (x)(∇ϕp ).(∇ϕp )]
i
X (61)
−λi Dt −ϕp (x) 2
∇.(∇ϕp p(x, t)) = ai e e [wλi (x)∇ ϕp + ∇wλi (x).∇ϕp − wλi (x)(∇ϕp ).(∇ϕp )] (62)
i
X
LHS = − λi ai e−λi Dt e−ϕp (x) wλi (x) (63)
i
X
RHS = ai e−λi Dt e−ϕp (x) [∇2 wλi (x) − ∇wλi (x).∇ϕp ] (64)
i

Eigen functions thus are governed by -

−λi wλi (x) = ∇2 wλi (x) − ∇wλi (x).∇ϕp (65)

N −1 N −1
X ∂ 2 wλi (x) X ∂wλi (x)
− κi ui = −λi wλi (x) (66)
j=1
∂ 2 uj j=1
∂uj

∂ 2 wλij (x) ∂wλij (x)

− κj uj = −λij wλij (x) (67)
∂ 2 uj ∂uj
PN −1 QN −1
Where λi = j=1 λij and wλi (x) = j=1 wλij (x) . Now we will calculate the eigenvalue corresponding
to i=0 and i=1 .
For i=0 λ0 = 0 which implies that wλ0 (x) is a constant function . The normalization condition is given by -
Z
|wλi (x)|2 e−ϕp (x) dx = 1 (68)
Ω̃
1 1
wλ0 (x) = R −ϕ (x) = (69)
( Ω̃ e p dx)1/2 |Ω̃|1/2
The generalqsolutions of equation (51) are well known they are called Hermite polynomials Now for i=1
κj PN −1
wλij (x) = u
Ω̃ j
and λ 1 = j=1 κj .

N −1
Y (κj )1/2 uj
wλ1 (x) = (70)
j=1
|Ω̃|1/2

3.3 Calculation of eigenvalues using the absorbing boundary condition

The eigenvalues and eigen functions we calculated earlier were for the backward Fokker Planck equation
without the boundary condition at the reaction boundary . The new perturbed eigenvalue due to this
absorbing boundary condition for (d=3,2) respectively is given by (4) -
Z
λϵi = λ0i + c2 ϵ wλ2 0 dVx + O(ϵ2 ), (71)
i
S

Z  2 !
2π 1
λϵi = λ0i + wλ2 0 dVx + O , (72)
log ϵ S
i log ϵ

Here dVx = e−ϕp (x) dx and the integral is over the submanifold S (eqn 43) . Using wλ0 (x) obtained in
the earlier section we will now calculate λϵ0 -

c2 ϵ S e−ϕ(x) dxg
R
ϵ
λ0 = + O(ϵ2 ). (73)
|Ω̃|

8
   d2
Z N −1 Z ∞  (2π)(N −1) 
P 1 2 Y 1 2
−
|Ω̃| = e p 2 κp up du1 du2 ..duN −1 = e− 2 κp up dup =  N −1 (74)
 

p=1 −∞  Q 
κp
l=1
−ϕ(x)
R P
To get S
e dxg we integrate over the surface S = p∈odd up ωp where ωp = cos(pπ/2N ) .
 d/2
N −2 2
Q
(2π) ω
Z
p
e−ϕ(x) dxg =  Q P p odd 2   , (75)
ωp
p κp
S
p odd κp

. Finally, summarizing these results, using a direct computation, we obtain that for fixed N and small ϵ
. 
3/2
 Nκπ 4πϵ + O(ϵ2 )  for d = 3,
ϵ
λ0 = (76)
 2
 2κ 1 + O 1
log ϵ for d = 2.
N log( ϵ )

Using a similar analysis, we obtain for large N that

 κ 3/2  8

λϵ1 ≈ κ1 + ϵc2 1 − 2 + O(ϵ2 ), (77)
Nπ π

Because C0 ≈ 1, using the first eigenvalue in relation (76), we obtain the approximation (in dimension
3).
√ !
N 2b b2
⟨τϵ ⟩2d = log + A2 N 2 + O(1), (78)
2Dκ ϵ D
3/2 √
b2

Nπ 2
⟨τϵ ⟩3d = + A3 N 2 + O(1), (79)
κ D4πϵ D

4 Szabo Schulten Schulten Approximation

The Smoluchowski or Fokker Planck equation for a polymer with N+1 beads with P (R1 , R2 , ...., RN , t)is
given by -
N
∂P X ∂ ∂ P
= DP = D0 · Peq . (80)
∂t j=0
∂R j ∂R j Peq

The SSS approximation(9) is a local equilibrium approximation where we take that the internal
modes are already relaxed in the timescale we are looking at which implies that t >> τp . We make the
assumption that -
p(r, t)
P (RN +1 , t) ≈ Peq (RN +1 ) (81)
peq (r)
Where p(r,t) is the distribution of end to end distance . using this approximation we get a diffusion
equation only as a function of end to end distance(5) -

∂p 1 ∂ ∂ p
= D 2 r2 peq (82)
∂t r ∂r ∂r peq

with the absorbing boundary condition p = 0 when r = a. The equilibrium distribution of the
end-to-end distance r, peq (r), is Gaussian with the mean squared distance L2 = N b2 ,
3/2
3r2
  
3
peq (r) = exp − 2 . (83)
2πL2 2L
The first passage time τ (r) starting from distance r satisfies the backward diffusion equation .
1 1 ∂ 2 ∂
D r peq τ (r) = −1 (84)
peq r2 ∂r ∂r

9
 Figure 2: x−3/2 ex (Γ(3/2, x))2

Putting eqn(83) in eqn(84) we get -

2 2
1 ∂ 2 −3r2 /2L2 ∂ e−3r /2L
2
r e τ (r) = − (85)
r ∂r ∂r D
Then we do a variable change -
r
3r2 2x L
x= ;r = L; dr = √ dx (86)
2L2 3 6x
We get -
d 3/2 −x dτ L2 −x 1/2
(x e )=− e x (87)
dx dx 6D
D = 2D0 From eqn(87) -
dτ L2 −3/2 x
=− x e Γ(3/2, x) (88)
dx 12D0
Z ∞
Γ(3/2, x) = x1/2 e−x dx (89)
x
R∞
peq (r)τ (r) dr (r2 )
τavg = a R∞ (90)
a
peq (r)r2 dr
After variable tarnsformation -
R∞
x0
x1/2 e−x τ (x) dx 3a2
τ̄ = ; x0 = (91)
Γ( 32 , x0 ) 2L2
Now we use the BC that τ (x0 ) = 0 -
Z ∞ Z ∞
dτ
x1/2 e−x τ (x) dx = [Γ(3/2, ∞)τ (∞) − Γ(3/2, x0 )τ (x0 )] − Γ(3/2, x)dx (92)
x0 x0 dx
Using the fact that Γ(3/2, ∞) = 0 and τ (x0 ) = 0 and eqn 92 we get -
R ∞ −3/2 x

2
L2 x0 x e Γ 32 , x dx
τ̄ = 3

(93)
12D0 Γ 2 , x0
The function in the integrand of the numerator (Fig -2) is near zero when x > 1 so the integral
contribution is mostly from small x . So we can use this asymptotic approximation for small x0 -
3
x−3/2 ex (Γ( , x))2 ≈ x−3/2 Γ(3/2)2 (94)
2
Z ∞
L2
τ= Γ(3/2) x−3/2 dx (95)
12D0 x0
We get final expression as - √ √
2π L3 2πb3 N 3/2
τ̄ → √ = √ . (96)
12 3 D0 a 12 3D0 a

10
4.1 Polymer with Self avoidance
Now we take a polymer model with self avoidance . The end to end distance can be modeled as (6) -
r
Peq (r) = Arg exp(−α( √ )δ (97)
< r2 >
Where this√ ensures that the probabilities of small r goes to zero which ensures self avoidance . In this
1
equation < r2 > = bN ν and δ = 1−ν . For a Rouse polymer g = 0, ν = 0.5 , α = 1.5 . The equivalent
potential used here is -
α
U (r) = −gkB T ln(r) + δ νδ kB T rδ (98)
b N
Now we put eqn(40) into eqn (27 ) and use the following substitution -
α x
x = crδ ; c = ; r = ( )1/δ (99)
bδ N νδ c

dx = cδrδ−1 dr = c1/δ δx(δ−1)/δ dr (100)

We now use this variable change and then simplify the diffusion equation -
d g+δ+1 −x dτ 1 3+g
(x δ e ) = − 2 2/δ e−x x δ −1 (101)
dx dx δ Dc
Using the same procedure as before we get -
dτ 1 1+g+δ g+3
= − 2 2/δ x− δ ex Γ( , x) (102)
dx δ Dc δ
αaδ
Now we calculate τavg using the BC that τ (x0 ) = 0; x0 = bδ N νδ
R∞ g+3
x δ −1 e−x τ (x) dx
x0
τ̄ = (103)
Γ( g+3
δ , x0 )
Z ∞ Z ∞
g+3
−1 −x dτ g + 3
x δ e τ (x) dx = − Γ( , x) (104)
x0 x0 dx δ
Now using this we get the final τ̄ as -
R∞ g+δ+1 g+3


2
b2 N 2ν x0
x− δ ex Γ δ ,x dx
τ̄ = 2 g+3 (105)
δ Dα2/δ


Γ δ , x0

The integrand in eq(105) also has the same form as fig 2 so we can take asymptotic limit to be -
∞ 2
g + 3 2 ∞ − g+δ+1
Z   Z
g+δ+1 g+3
x− δ ex Γ ,x dx = (Γ( )) x δ dx (106)
x0 δ δ x0

b3+g g + 3 ν(3+g)
τ̄ = g+3 Γ( )N (107)
δ(g + 1)α δ Dag+1 δ

5 Wilemski Fixman
The Wilemski–Fixman (WF) approximation is a well-known method(8) used to estimate how long it
takes for two parts of a flexible molecule, like a polymer chain, to find each other and react. This is
especially important in problems like DNA looping, protein folding, or polymer chemistry, where two ends
of a molecule must come close together before anything happens. The WF approximation simplifies this
complex process by focusing on how likely it is that the two reactive parts are near each other at a given
time. Instead of tracking every detail of the molecule’s motion, the method uses an effective equation
involving a correlation function between the two ends. While it makes some simplifying assumptions—like
treating the motion as purely diffusive and ignoring certain higher-order effects—it still gives results that
match experiments and simulations quite well, especially when the reaction happens slowly compared to
the motion (i.e., in the diffusion-limited regime).

11
In the style of modern nonequilibrium statistical mechanics, Wilemski and Fixman relate < τ >to the
time integral of a sink–sink time correlation function. They add a sink term kS(r), to the Fokker Planck
equation.
∂p(x, t)
= Lp(x, t) − kS(r) (108)
∂t
Here L represents the Fokker planck operator in 3 dimensions and S(r) is a sink function where r is the
end to end distance of the polymer. Their choice for the sink functionS(r), depending on the end-to-end
distance r, is the Heaviside sink . (
1, if r < a
S(r) = (109)
0, if r ≥ a
The limit of infinite k corresponds to complete absorption in the sink, and provides the theoretical
connection to the first passage time. Other sinks can be used such as a delta function sink or a gaussian
sink. We define C(t) as a sink-sink time correlation function.
Z ∞ Z ∞
C(t) = dx dx0 S(x)G(x, t|x0 , 0)S(x0 )Peq (x0 ) (110)
−∞ −∞

Here x represents the end to end distance , Peq (x0 ) is the equilibrium end to end distance.G(x, t|x0 , 0)
is the probability of the end to end distance going from x0 at time t=0 to x at time t .

(x − ϕ(t)x0 )2
 
3 3/2 1 3
G(x, t | x0 , 0) = ( ) exp − · (111)
2πL2 [1 − ϕ2 (t)]
3/2 2L2 (1 − ϕ2 (t))

Here ϕ(t) represents the end to end correlation of the Rouse chain .

⟨r(t) · r(0)⟩eq
ϕ(t) = (112)
⟨r2 ⟩eq

Using the rouse model we get -

P∞ − t
e τp
p=1 p2
ϕ(t) = P∞ 1 (113)
p=1 p2

b2 N 2
Here τp = 3π 2 Dp2 , ϕ(t) can be approximated as e−t/τ1 as that is the most dominant timescale .
 
3ϕ(t)xx0
3 ∞ ∞  sinh
3(x2 + x20 )
 
L2 (1 − ϕ2 (t))
Z Z
3 1
C(t) = 4πx2 S(x) dx 4πx20 S(x0 ) dx0 exp −
2πL2 (1 − ϕ2 (t))3/2 0 0 2L2 (1 − ϕ2 (t)) 3ϕ(t)xx0
L2 (1 − ϕ2 (t))
(114)
This integrand can be solved quite easily using the Delta Function Sink -

S(x) = δ(x − a) (115)

Where a is defined as the contact radius of the two ends where the reaction occurs .
We now need an expression connecting C(t) to the MFET. C(t) describes how the probability of the
polymer being in a reactive state at time t is correlated with its having been in that state at t = 0.
In other words, if C(t) decays rapidly, the polymer often finds reactive configurations. If C(t) decays
slowly, the polymer takes longer to return to a reactive state after leaving it.If the polymer chain diffuses
quickly, it rapidly samples different conformations, some of which bring the ends back together. Because
of this rapid exploration, the polymer often finds itself in a reactive configuration again.That is, the rate
at which C(t)decreases corresponds to how often the ends are encountering for the first time.C(t) acts
kind of like a survival probability -
Z ∞   Z ∞
dC(t) ∞
⟨τ ⟩ = t − dt = [−tC(t)]0 + C(t) dt (116)
0 dt 0

12
 Figure 3: C(t) for N=100 ,τ = 0.1 ,a=0.1b

When we plot C(t) we notice that the function does not converge as time goes to infinity.This would
imply that < τ > will diverge . So what we do is we subtract C(∞) from C(t) and then calculate τ .We
are interested in the time it takes for the polymer ends to meet for the first time.The long-time behavior
of C(t) represents an equilibrium situation where the polymer reaches a steady-state probability of being
in the reactive region. However, first-passage problems must focus on the transient decay of C(t), not
the long-time equilibrium behavior.The function C(t) − C(∞) isolates the non-equilibrium dynamics of
the polymer.This ensures that the integral properly captures how long it takes for the polymer ends to
meet for the first time before equilibrium dominates.We define C̃(t) = C(t) − C(∞).
Z ∞ !
dC̃(t) h i∞ Z ∞
⟨τ ⟩ = t − dt = −tC̃(t) + C̃(t) dt (117)
0 dt 0 0

We also normalize the transient behaviour by dividing by C(∞)

Z ∞  
C(t)
< τ >= dt −1 (118)
0 C(∞)

C(t) exp(−2x0 ϕ(t)2 /1 − ϕ(t)2 )sinh(2x0 ϕ(t)/1 − ϕ(t)2 )

= p (119)
C(∞) 2x0 ϕ(t) 1 − ϕ(t)2

C(t)
Figure 4: C(∞ −1

Here x0 = 3a2 /2N b2 .For large N , x0 is very small so we can write e−x ≈ 1 − x and sinh(x) ≈ x for
small x in eqn (119) .
2x0 ϕ(t)2 2x0 ϕ(t)
p
C(t) (1 − 1−ϕ(t)2 ) 1−ϕ(t)2− 2x0 ϕ(t) 1 − ϕ(t)2 )
−1= p (120)
C(∞) (2x0 ϕ(t) 1 − ϕ(t)2 )

13
Assuming ϕ(t) ≈ e−t/τ1 and further taking the approximation that the internal modes equilibrate fast
that implies that τ1 is small .So we can take 1 − e−2t/τ1 ≈ 1 . Using all the above approximations in the
limit of large N and fixed a we get the result -
< τ >∝ τ1 (121)
2
The final result we get by this method is that τ is proportional to N .

6 Comments on SSS and Wilemski Fixman Approximations

N a τ ±µ WF ( Delta Sink) SSS

50 0.1 570 ± 30 564 424
0.5 174 ± 10 205 82
1.0 110 ± 5 141 39
75 0.1 1070 ± 70 1107 779
0.5 410 ± 20 446 152
1.0 250 ± 10 326 73
100 0.1 1800 ± 80 1796 1201
0.5 680 ± 30 778 235
1.0 450 ± 20 590 114

Table 1: Comparison of MFET (τ ) for different chain lengths (N ) and sink radii (a) using various
theoretical methods: Wilemski-Fixman (DS), and SSS theory.

Table I (5) shows that while the theoretical treatments of Wilemski-Fixman (WF) and Szabo-
Schulten-Schulten (SSS) do not perfectly match simulation data, they both exhibit improved accuracy as
the contact distance a decreases. Notably, the delta-sink (DS) version of WF demonstrates remarkable
agreement with simulation for a = 0.1.
This improved accuracy for small a can be understood by examining the assumptions underlying
these theoretical models. The SSS approximation, for instance, can be derived from the full diffusion
equation by applying a local equilibrium assumption to the time-dependent distribution function:

p(r, t)
P (RN +1 , t) ≈ P eq (RN +1 ) , (122)
peq (r)
Where p(r, t) is the probability distribution of the end-to-end distance r at time t, and P eq denotes
the equilibrium distribution over all internal coordinates given r.
Substituting this approximation into the full diffusion equation and integrating over the internal
coordinates leads to the SSS diffusion equation for the end-to-end distance. where D0 is the diffusion
coefficient of a single monomer.
However, the validity of this local equilibrium approximation hinges on the relative time scales of
the system. Specifically, it is only justified if the relaxation time to local equilibrium is much shorter
than the trapping (reaction) time. The longest relaxation time in the Rouse model is associated with
the slowest mode and is given by:

N2
τR = , (123)
3π 2
(in units where b = D0 = 1). Importantly, τR is independent of the contact distance a. In the limit
of small a, the mean first passage time predicted by the SSS model asymptotically approaches:
√ √
2π L3 2πb3 N 3/2
τ̄ → √ = √ . (124)
12 3 D0 a 12 3D0 a
As an example, for N = 50 and a = 1.0, we have τR = 84 and the mean first passage time is τ̄ ≈ 1.3τR .
In contrast, for a = 0.1, the trapping time increases to τ̄ ≈ 7τR . This suggests that the local equilibrium
assumption becomes valid only when a is sufficiently small. We can also write as -
π 3/2 b
τ̄ ≈ τR √ √ (125)
2 6 Na

14
 √
For τ̄ >> τR a needs to be much smaller than b/ N .
The WF approach, which relates t to C(t), also uses a local equilibrium approximation. Inside the sink,
the correct distribution is replaced by an equilibrium distribution having the same sink occupancy as
the correct distribution.
R
dRN +1 S(r)P (RN +1 , t)
S(r)P (RN +1 , t) ≈ S(r)P eq (RN +1 ) R . (126)
dRN +1 S(r)P eq (RN +1 )

This approximation is considered to be more benign than the one used in the SSS theory because
it retains some non-Markovian characteristics of the end-to-end vector dynamics, unlike the SSS model
which simplifies the process to a Markovian one.
Note that a potential source of confusion in the WF approach is the apparent lack of continuity in taking
limits. In the limit of small a at fixed N , τ is of the order of N 3/2 /a. We observed that in the limit of
large N at fixed a, τ appears to be of the order of N 2 and independent of a. But in fact, τ must combine
both kinds of behavior, N 3/2 /a and N 2 .This can be seen in eqn(79).

7 Non-Monotonicity of MFET
As a polymer collapses, a naive intuition suggests that the Mean First Encounter Time (MFET) between
its ends should decrease monotonically. This is because, as the polymer becomes more compact, the two
ends are more likely to come into contact and react. This behavior is indeed observed in models like the
3
β-polymer model (7), where the MFET scales as τ ∼ N 2 (β−1) for β > 1. The parameter β describes the
degree of polymer collapse: the Rouse model is recovered when β = 2.
In this model, the Langevin equation for the Rouse normal modes is modified as:
dup p dwp
= −Dp κ̃p up + 2Dp , (127)
dt dt
with mode-dependent relaxation coefficients defined as:
 pπ 
κ̃p = 4κ sinβ . (128)
2N
As β decreases, the effective springs become less stiff, leading to a more collapsed polymer configu-
ration. Consequently, the MFET decreases monotonically with collapse in this simplified picture.
However, this approach neglects an important physical effect — self-avoidance of polymer beads. In
reality, polymer beads cannot overlap due to excluded volume effects. This is typically modeled using a
repulsive potential, such as the Lennard-Jones (LJ) potential. As the polymer collapses, these repulsive
interactions become increasingly significant, resisting further compaction and thereby increasing the
MFET.
This interplay leads to a non-monotonic behavior in MFET as a function of polymer collapse. Such
behavior can be captured analytically using the Szabo-Schulten-Schulten (SSS) approximation applied
to self-avoiding polymers. In this framework, the MFET scales as:

τ ∝ N 3ν+νg , (129)

where the radius of gyration scales as R2 ∝ N ν . The exponent ν decreases as the polymer collapses,
reflecting the reduction in spatial extent. However, the factor g accounts for the suppressed probability
of very short end-to-end distances due to repulsive interactions. Importantly, g increases with collapse
to ensure that the end-to-end distribution vanishes as distance approaches zero.
Therefore, while ν decreases with collapse, g increases, and the product νg introduces a competition
that results in non-monotonicity in MFET. Initially, MFET decreases as collapse brings the ends closer
together, but beyond a point, repulsive effects dominate and the MFET increases.One example of this
was acheived by using crowders (10) , where looping time showed non monotonic behavior as seen in
Fig(5) .

15
 Figure 5: Non Monotonic Behavior due to crowders

References
(1)Theory of Polymer Dynamics Lecture notes (Johan T Padding)
(2)A. Amitai and D. Holcman, Phys. Rev. Lett. 110, 248105 (2013)
(3)M. Doi and S. F. Edwards, The Theory of Polymer Dynamics (Clarendon Press, Oxford, UK, 1986
(4)I. Chavel and E. A. Feldman, Duke Math. J. 56, 399 (1988).
(5) Diffusion limited first contact of the ends of a polymer: Comparison of theory with simulation
Richard W. Pastor; Robert Zwanzig; Attila Szabo
(6) Polymer Physics Michael Rubenstein and Ralph H Colby
(7) Polymer model with long-range interactions: Analysis and applications to the chromatin structure
A. Amitai and D. Holcman PHYSICAL REVIEW E 88, 052604 (2013)
(8)G. Wilemski and M. Fixman, J. Chem. Phys. 60, 878 (1974).
(9)A. Szabo, K. Schulten, and Z. Schulten, J. Chem. Phys. 72, 4350(1980).
(10)Kinetics of polymer looping with macromolecular crowding: effects of volume fraction and crowder
size Jaeoh Shin, Andrey G. Cherstvya and Ralf Metzler Soft Matter, 2015, 11, 472

16