Anisotropic active Brownian particle in two dimensions under stochastic resetting

Anirban Ghosh
Department of Physics, Amity Institute of Applied Sciences, Amity University, Kolkata, 700135, India

Sudipta Mandal
Indian Institute of Science Education and Research, Mohali and
Institute of Electronic Structure and Laser, FORTH-Hellas, Greece
(Dated: January 10, 2025)

We analytically investigate the dynamic behavior of an anisotropic active Brownian particle under
arXiv:2501.05149v1 [cond-mat.stat-mech] 9 Jan 2025

various stochastic resetting protocols in two dimensions. The motion of shape-asymmetric active
Brownian particles in two dimensions leads to anisotropic diffusion at short times, whereas rota-
tional diffusion causes the transport to become isotropic at longer times. We have considered three
different resetting protocols: (a) complete resetting, when both position and orientation are reset
to their initial states, (b) only position is reset to its initial state, (c) only orientation is reset to its
initial state. We reveal that orientational resetting sustains anisotropy even at late times. When
both the spatial position and orientation are subject to resetting, a complex position probability
distribution forms in the steady state. This distribution is shaped by factors such as the initial ori-
entation angle, the anisotropy of the particle, and the resetting rate. We have calculated the exact
expressions for mean-square displacements using a renewal approach for different resetting protocols
and numerically verified the analytical results. When only the translational degrees of freedom are
reset, while the particle’s orientation evolves naturally, the steady-state no longer depends on par-
ticle asymmetry. In contrast, if only the orientation is reset, the long-term probability distribution
becomes a Gaussian, using an effective diffusion tensor—containing non-diagonal elements—defined
by the resetting rate. More broadly, the interaction between translational and rotational dynamics,
in combination with stochastic resetting, produces distinct behaviors at late times that are absent in
symmetric particles. Given recent progress in experimental resetting techniques, these results could
be highly useful for controlling asymmetric active colloids, such as in self-assembly applications.

I. INTRODUCTION matic macromolecules37,38 . Such shape anisotropy oc-

curs across a wide range of scales, from a few nanometers
Active processes represent a class of dynamical systems to several micrometers. Analyzing the diffusive behav-
that are inherently out of equilibrium due to their self- ior of anisotropic particles is naturally more complicated
propelling nature1–4 . Since Vicsek’s pioneering work on than that of spherical particles, primarily due to the cou-
self-propulsion5 , research into active matter systems has pling between rotational and translational motions. In
expanded significantly, covering collective behaviors such the short-time regime, the anisotropic behavior emerges
as flocking6,7 , clustering8–10 , and motility-induced phase from the correlation between the particle’s instantaneous
separation (MIPS)11–14 . Theoretical approaches to un- orientation and its translational motion39,40 . The mo-
derstanding active matter often center around the anal- tion of anisotropic Active Brownian Particles (ABPs) is
ysis of simplified yet analytically solvable models, such marked by a transition from the initial anisotropic diffu-
as the Run-and-Tumble Particle (RTP) and the Active sion, driven by the particle’s starting orientation, to an
Brownian Particle (ABP), as well as various extensions effective isotropic diffusion at longer times.
and modifications of these foundational models3,15–23 . In Stochastic resetting involves intermittently interrupt-
these models, activity emerges from the interaction be- ing and restarting a dynamical process. The inclusion of
tween spatial motion and a stochastic internal orienta- a resetting mechanism in a stochastic system can dramat-
tional degree of freedom. The intrinsic timescale linked ically change both its static and dynamic properties41 .
to the particle’s internal orientation gives rise to a va- This concept is relevant across diverse fields, such as
riety of fascinating phenomena, even at the level of an search problems42–45 , population dynamics46,47 , com-
individual particle, such as spatial anisotropy and ballis- puter science48 , and biological systems49–51 . A well-
tic motion over short durations3,24,25 . known example of stochastic resetting is a Brownian par-
Numerous processes in physics, chemistry, and biol- ticle undergoing diffusion that is periodically reset to its
ogy involve diffusing particles with highly anisotropic initial location at a specific rate52 . This resetting pro-
shapes. Notable instances contain rod-shaped bacte- cess drives the system away from equilibrium, result-
ria or viruses26–29 , anisotropic colloids30–36 , and ne- ing in intriguing outcomes like non-equilibrium steady
 2

states, unique temporal relaxation behaviors, and non- vector n̂i = (cos θi , sin θi ), which defines the direction of
monotonic mean first passage times52–54 . The impact of the propulsion velocity. In the lab frame (x-y plane),
resetting on various diffusive processes has been exten- the particle’s state at any given time t can be described
sively studied in the past decade. While it might seem by the position vector R(t), which indicates the coordi-
intuitive that restarting a stochastic process repeatedly nates of its center of mass. This vector also represents
would delay its overall completion, this is not necessarily the coordinates relative to the particle’s intrinsic body
the case. In fact, a surprising feature of stochastic reset- frame (δ x̃, δ ỹ). The angle θ(t) represents the orientation
ting is its potential to speed up the completion of specific of the particle, specifically the angle between the x axis
random search processes. As an instance, resetting can of the laboratory frame and the x̃ axis of the body frame.
lower the mean first-passage time for a diffusive parti- Understanding the frame of reference is essential for an-
cle to reach a target44,52,55–58 , decrease the runtime of alyzing the behavior of anisotropic particles, as the rota-
stochastic algorithms59–61 , and reduce the turnover time tional and translational motions in the body frame are
in enzymatic reactions62,63 . Therefore, stochastic reset- decoupled. The motion of the anisotropic ABP as shown
ting is an effective method for optimizing the completion in the fig.(1a) is governed by the Langevin equation in
times of stochastic processes and dynamical systems. body frame, which can be expressed as:
While most previous studies on resetting have focused
on symmetric particles, this paper investigates the im- 1 ∂ x̃ v0
= + η̃x (t)
pact of resetting on an anisotropic active Brownian par- Γ∥ ∂t Γ∥
ticle (ABP) in two dimensions. The ABP in our study 1 ∂ ỹ
is an overdamped particle with an orientational degree = η̃y (t) (1)
Γ⊥ ∂t
of freedom, undergoing rotational diffusion. Its behavior 1 ∂ θ̃
in two-dimensional space is defined by its position (x, y) = η̃θ (t)
Γθ ∂t
and orientation θ. We examine three distinct resetting
protocols: (a) complete resetting, where both the par- Here v0 is the propulsion velocity in the body frame, it
ticle’s position and orientation are reset to their initial is directed along the longer axis of the anisotropic parti-
values at a rate r, (b) resetting of only the position, and cle. The mobilities along its longer and shorter axes are
D
(c) resetting of only the orientation angle. In the first respectively Γ∥ = kB∥T and Γ⊥ = kDB⊥T . Γθ = kDBθT is the
two protocols involving position resetting, the particle’s rotational mobility. Here D∥ , D⊥ and Dθ represent dif-
mean-square displacement (MSD) reaches a steady state. fusion coefficients along the parallel, perpendicular and
Conversely, for the protocol that resets only the orienta- rotational axes respectively. The correlations of the ther-
tion, the particle does not attain a stationary state. In- mal fluctuations in the body frame are described as,
stead, it exhibits anisotropic diffusion, with a transition ⟨η̃⟩ = 0
from ballistic to diffusive behavior as time progresses. (2)
⟨η̃i (t)η̃j (t′ )⟩ = 2kB T Γi δij δ(t − t′ )
The structure of the paper is as follows: Sec.(II) intro-
duces the fundamental model and theoretical framework; To derive these equations in the laboratory frame, we
Sec.(III A) elaborates on the complete resetting protocol, must perform a straightforward coordinate rotation. The
detailing its dynamics; Sec.(III B) focuses on the dynam- displacement in the two frames is related with the follow-
ics under the position-reset protocol; Sec.(III C) discusses ing equations;
the effects of orientation resetting; Sec.(IV) presents the
conclusions. Additionally, Appendix(A) provides a com- δx = cos θδ x̃ − sin θδ ỹ
(3)
prehensive analytical derivation of steady state probabil- δy = sin θδ x̃ + cos θδ ỹ
ity distribution function using a perturbative scheme.
Taking the limit δt → 0 after dividing the equations by
δt and substituting the expressions for linear and angular
velocities in the body frame from Eq.(28), we derive the
II. MODEL AND THEORY
final set of equations that describe the motion in the lab
frame,
We have investigated the dynamics of an anisotropic
active Brownian particle (ABP) subjected to reset- ∂x
ting. This particle is confined to move within a two- = v0 cos θ(t) + ηx (t)
∂t
dimensional framework and possesses a propulsion veloc- ∂y
ity along its longer axis, denoted as (x̃). The particle’s = v0 sin θ(t) + ηy (t) (4)
∂t
orientation in the x-y plane plays a crucial role in its ∂θ
motion. This orientation can be characterized by a unit = ηθ (t)
∂t
 3

FIG. 1. (a) The basic model of an anisotropic ABP incorporates parallel, perpendicular, and rotational diffusion coefficients,
with the propulsion velocity directed along the particle’s major axis. (b) In the position-orientation complete resetting protocol,
both the particle’s position and orientation are reset simultaneously. (c) The position-only resetting protocol involves resetting
the particle’s position to the origin while leaving its orientation unchanged. (d) The orientation-only resetting protocol resets
the particle’s orientation to its initial orientation as θ = 0 without altering its position.

The random noise ηi have zero mean and the correla- In the long time regime when t ≫ Dθ−1 , the mean-
tions at fixed θ(t) are given by square displacements along both the x and y axes ex-
hibit diffusive behavior similar to an isotropic particle
⟨ηi (t)ηj (t′ )⟩ = 2kB T Γij [θ(t)]δ(t − t′ ) expressed as24,64 ,
(5)
⟨ηθ (t)ηθ (t′ )⟩ = 2kB T Γθ δ(t − t′ ) !
2
v ∆D 7v 2
Here ⟨∆x2 (t)⟩0 = 2 D̄ + 0 t + − 02
2Dθ 4Dθ 4Dθ
  ! (10)
∆Γ cos 2θ sin 2θ v2 ∆D 5v 2
Γij [θ(t)] = Γ̄δij + (6) ⟨∆y 2 (t)⟩0 = 2 D̄ + 0 t − − 02
2 sin 2θ − cos 2θ 2Dθ 4Dθ 4Dθ
(Γ +Γ )
where Γ̄ = ∥ 2 ⊥ and ∆Γ = (Γ∥ − Γ⊥ ). Furthermore, at the short time regime t < Dθ−1 , we
We integrate Eq.(4) with respect to t for x, and it is find initially instantaneous diffusion regime and then
found as, super-diffusion regime as depicted24
Z t Z t
x(t) = dt′ ηx (t′ ) + dt′ v0 cos θ(t′ ) (7) ⟨∆x2 (t)⟩0 = 2D∥ t − 2Dθ ∆Dt2 + O(t3 )
(11)
0 0
⟨∆y 2 (t)⟩0 = 2D⊥ t + 2Dθ ∆Dt2 + O(t3 )
The second moment of the particle along x and y axes
with an initial angle θ0 without any resetting, is calcu- In this paper, we study the effect of stochastic reset-
lated as24,64 ting on the dynamics of and anisotropic ABP. For ABP,
the presence of rotational and translational degrees of
"
 1 − e−4Dθ t 
# freedom give rise to the chance of resetting in the phase
2
⟨x (t)⟩0 = 2D̄t + ∆D cos 2θ0 space, instead of in position space only. The first ex-
4Dθ plored study shows different resetting protocols involving
v02 cos 2θ0 position and velocity in one-dimensional RTP65 . In the
3 − 4e−Dθ t + e−4Dθ t (8)


+ 2 case of anisotropic ABP, where the translational and ro-
12Dθ
2 tational degrees are the important variables, the similar
v
+ 02 Dθ t + e−Dθ t − 1


situation arises like RTP. In the following, we study three
Dθ
different resetting protocols.
and (I) Position-orientation resetting: All the phase
" #
 1 − e−4Dθ t  variables are reset to its initial values with the reset rate
2
⟨y (t)⟩0 = 2D̄t − ∆D cos 2θ0 r as shown in the fig.(1b). We assumed that the particle
4Dθ
starts from the origin and initial angle θ0 = 0. The par-
v02 cos 2θ0 (9) ticle resets to x = y = 0 and θ0 = 0 at any time t with
3 − 4e−Dθ t + e−4Dθ t


− 2
12Dθ reset rate r.
2
v (II) Position resetting: According to this protocol
+ 02 Dθ t + e−Dθ t − 1


Dθ shown in the fig.(1c), we reset the position of the particle
 4

to the origin with rate r, but the orientation is not reset.

The orientation evolves according to the normal diffusion Z t
dynamics. ⟨xn (t)⟩ = e−rt ⟨xn (t)⟩0 + r dτ e−rτ ⟨xn (τ )⟩0
(III) Orientation resetting: In this protocol shown 0
Z t
(14)
in the fig.(1d), only the orientation θ resets to θ = 0 with n
⟨y (t)⟩ = e −rt n
⟨y (t)⟩0 + r dτ e −rτ n
⟨y (τ )⟩0
rate r, and the position is not reset. 0

Here, ⟨xn (t)⟩0 is the nth moment of the x-coordinate of

III. ANISOTROPIC ABP WITH RESETTING the position in the absence of any resetting events, and
it can be computed explicitly for any value of n.
A. Position-orientation Resetting In the following, we will derive the first two moments
for the x and y components, utilizing the established ex-
pressions for a free ellipsoidal Active Brownian Particle
The most common resetting protocol occurs when both
(ABP). The average values of the x and y coordinates
the position and the orientation of the particle are re-
are given by:
set to their initial values at a rate r. R Our focus lies
on the position distribution P (x, y, t) = dθ P(x, y, θ, t),
t t
which is obtained by integrating P(x, y, θ, t), the proba-
Z Z
⟨x(t)⟩θ00 = v0 dτ ⟨cos θ(τ )⟩θ00 + ⟨ηx (τ )⟩dτ
bility density of the particle being at position (x, y) with 0 0
orientation θ at time t, over all possible orientations θ. Z t Z t (15)
A renewal equation for P(x, y, θ, t), the joint probability ⟨y(t)⟩θ00 = v0 dτ ⟨sin θ(τ )⟩θ00 + ⟨ηy (τ )⟩dτ
0 0
density of the particle’s position and orientation, can be
written directly as: where we have used the superscript θ0 to denote the ini-
tial orientation.
t
⟨cos θ(τ )⟩θ00 = cos θ0 e−Dθ τ
Z
P(x, y, θ, t) = e−rt P0 (x, y, θ, t) + r dτ e−rτ P( x, y, θ, τ ) (16)
0 ⟨sin θ(τ )⟩θ00 = sin θ0 e−Dθ τ
(12)
Here, P0 (x, y, θ, t) represents the probability that, in the The average positions can be computed using the above
absence of resetting, the active Brownian particle (ABP) equations in Eq(15),
is located at position (x, y) with orientation θ at time
t, starting from the initial position x = 0, y = 0. The v0 cos θ0
⟨x(t)⟩θ00 = (1 − e−Dθ t )
renewal equation consists of two terms: Dθ
(17)
The first term accounts for the scenario where no reset- v0 sin θ0
⟨y(t)⟩θ00 = (1 − e−Dθ t )
ting events occur up to time t. The second term captures Dθ
the contribution from trajectories where the most recent Using Eq.(14) we get the first moment after resetting at
resetting event occurred at a time t − τ prior. θ0 = 0
A corresponding renewal equation governing the posi-
v0
tion distribution can be formulated by integrating over ⟨x(t)⟩ = (1 − e−(r+Dθ )t ) (18)
the orientation θ, as follows: r + Dθ
while ⟨y(t)⟩ = 0 for every value of time. This observa-
Z t tion indicates the emergence of a timescale due to the
P (x, y, t) = e−rt P0 (x, y, t) + r dse−rτ P0 (x, y, τ ) (13) resetting mechanism. At short times, specifically for
0
t ≪ (r + Dθ )−1 , the particle exhibits motion along the x-
axis with an effective velocity v0 , similar to the behavior
1. Moments of a free anisotropic Active Brownian Particle (ABP). On
the other hand, at longer times, the particle approaches a
To understand the influence of position-orientation re- stationary position, with its mean position shifting closer
setting on the dynamical behavior of the anisotropic to the origin as the resetting rate r increases.
ABP, we examine the moments of the position coordi- The second moments of x and y without resetting of
nates. Clearly, under the influence of resetting, the mo- ellipsoidal ABP are given in Eqs.(8) and (9). Next, we
ments adhere to a renewal equation analogous to Eq.(13). compute the second moments using Eq.(14), Eq.(8), and
As an illustration, we can multiply both sides by xn and Eq.(9) with n = 2. Here, we investigate the behav-
integrate with respect to both x and y, which leads to ior of the mean squared displacement (MSD), defined
the resulting expression: as σx2 (t) = ⟨x2 (t)⟩ − ⟨x(t)⟩2 for the x-component and
 5

σy2 (t) = ⟨y 2 (t)⟩ for the y-component. We obtain precise

analytical formulas for the higher-order moments of the
x and y components of the position under the position-
orientation resetting protocol. The second moment of
x(t) can be explicitly calculated from Eq.(14) as

2D̄ ∆D cos 2θ0 v02

⟨x2 (t)⟩ = + + FIG. 2. Position-orientation resetting: Mean squared dis-
r r + 4Dθ r(r + Dθ )
" placements of x and y components as a function of time t
v02 cos 2θ0 v 2 cos 2θ0 e−4Dθ t for Dθ = 0.1, D∥ = 1, D⊥ = 0.1, v0 = 1, θ0 = 0 and vary-
+ + e−rt 0
(r + Dθ )(r + 4Dθ ) 3Dθ (r + 4Dθ ) ing resetting rates r. The simulation data are represented by
(19) symbols, while the solid curves correspond to the analytical
v02 cos 2θ0 e−Dθ t v2 2D̄ v02 e−Dθ t
− − 0 − + predictions from Eqs.(21) and (22). The stationary values de-
3Dθ (r + Dθ ) rDθ r Dθ (r + Dθ ) crease as the resetting rate r increases.
#
−4Dθ t
∆D cos 2θ0 e
−
r + 4Dθ

and the second moment of y(t) becomes

2D̄ ∆D cos 2θ0 v02

⟨y 2 (t)⟩ = − +
r r + 4Dθ r(r + Dθ )
"
v02 cos 2θ0 ∆D cos 2θ0 e−4Dθ t
− + e−rt
(r + Dθ )(r + 4Dθ ) r + 4Dθ FIG. 3. Position-orientation resetting: Mean squared dis-
−Dθ t −4Dθ t placements of x and y components as a function of time t for
v02
cos 2θ0 e v02
2D̄ v02
cos 2θ0 e
+ − − − Dθ = 1, D∥ = 1, D⊥ = 0.1, v0 = 1, θ0 = 0 and varying reset-
3Dθ (r + Dθ ) rDθ r 3Dθ (r + 4Dθ ) ting rates r. The simulation data are represented by symbols,
#
2 −Dθ t while the solid curves correspond to the analytical predictions
v0 e
+ from Eqs.(21) and (22). The stationary values decrease as the
Dθ (r + Dθ )
resetting rate r increases.
(20)

For θ0 = 0, where the particle is initially oriented along For very small t, if we expand the exponential terms
the x axis, the mean-squared displacements along the x up to second order, we find
and y directions are given as:
 v02  v02 (2r − 11Dθ )
σx2 (t) = 2D∥ − t− rD∥ +
r + Dθ 6Dθ
2D̄ ∆D 2v02 (r + 2Dθ )
σx2 = + + !
2
!
r r + 4Dθ r(r + Dθ )(r + 4Dθ ) rv (3r + 4D θ )
" + 2Dθ ∆D t2 + 2rDθ ∆D − 0 t3
2 −Dθ t
2v0 e e−4Dθ t  v02  3(r + Dθ )
+ e−rt + − ∆D (21) !
3Dθ (r + Dθ ) r + 4Dθ 3Dθ r2 v02 Dθ (5r + 8Dθ ) 4r2 Dθ2 ∆D 4
# + − t + O(t5 )
v02 2D̄ v02  2 (r + Dθ )(r + 4Dθ ) r + 4Dθ
− − − 1 − e−(r+Dθ )t
rDθ r (r + Dθ ) 2 (23)

and and
 
2D̄ ∆D 4Dθ v02 σy2 (t) = 2D⊥ t + 2Dθ ∆D − rD⊥ t2 − 2rDθ ∆Dt3
σy2 = − +
r r + 4Dθ r(r + Dθ )(r + 4Dθ ) !
4r2 Dθ2 ∆D r3 v02
− t4 + O(t5 )
"
2 −Dθ t
4v0 e e−4Dθ t  v2  r + 4Dθ (r + Dθ )(r + 4Dθ )
+ e−rt + ∆D − 0
3Dθ (r + Dθ ) r + 4Dθ 3Dθ (22) (24)
#
v02 2D̄ It is intriguing to contrast this short-time behav-
− −
rDθ r ior with that of an ellipsoidal active Brownian particle
(ABP) without resetting. When starting from the origin
 6

3.0

1.0
2.5

0.8
2.0

0.6
1.5

0.4
1.0

0.2 0.5

0.0 0.0
10.0 7.5 5.0 2.5 0.0 2.5 5.0 7.5 10.0 10.0 7.5 5.0 2.5 0.0 2.5 5.0 7.5 10.0

FIG. 4. Steady state position distribution along x of FIG. 5. Steady state position distribution along y of
anisotropic ABP under position-orientation resetting when anisotropic ABP under position-orientation resetting when
Dθ = 0.1, D∥ = 1, D⊥ = 0.1, v0 = 1 and θ0 = 0 for dif- Dθ = 0.1, D∥ = 1, D⊥ = 0.1, v0 = 1 and θ0 = 0 for dif-
ferent values of reset rate r as shown in the legends taken at ferent values of reset rate r as shown in the legends taken at
t = 100. t = 100.

and aligned along the x axis, the mean squared displace- Figs.(4) and (5) show the simulation results for the
ment (MSD) of a regular ellipsoidal ABP along the x long time position distribution for both x and y. For a
axis exhibits hyper-diffusive behavior in the short-time small resetting rate, the particle behaves like a free par-
regime24 . In the presence of position-orientation reset- ticle for an extended duration before being reset to the
ting, however, both σx2 and σy2 exhibit growth with a origin. Conversely, for a large resetting rate, the time in-
super-diffusive nature. Additionally, the resetting sig- terval between successive resetting events is significantly
nificantly alters the leading-order behavior of the mean shorter than the rotational diffusion timescale of the free
squared displacement (MSD) as shown in the figs.(2) and anisotropic ABP dynamics. The anisotropic character-
(3). istics of the position distribution, as depicted in figs.(4)
At long times, the particle is anticipated to reach a and (5), reveal that the distribution along the y-direction
stationary state in which the mean squared displacement significantly differs from that along the x-direction.
(MSD) becomes time-independent and attains a constant A novel perturbative scheme is employed to an-
value. alytically determine the distribution function of the
anisotropic active Brownian particle (ABP). The com-
2D̄ ∆D v 2 (r2 + 2rDθ + 4Dθ2 ) plete calculation and details of the method are provided
lim σx2 = + + 0 in Appendix A.
t→∞ r r + 4Dθ r(r + Dθ )2 (r + 4Dθ )
(25)
2 2D̄ ∆D 4Dθ v02
lim σy = − +
t→∞ r r + 4Dθ r(r + Dθ )(r + 4Dθ ) B. Position Resetting
Interestingly, the stationary mean squared displacement
(MSD) exhibits distinct values for the x and y compo- In this section, we analyze the dynamics of the ac-
nents, reflecting the continued presence of anisotropy. tive Brownian particle (ABP) under resetting protocol
This outcome is anticipated, as resetting to θ0 = 0 con- II, wherein only the particle’s position is reset. Specifi-
sistently introduces a pronounced directional bias with cally, the particle is returned to the origin, x = y = 0,
each resetting event. Figs.(2) and (3) illustrate the at a resetting rate r, while its orientation θ remains un-
plots of σx2 and σy2 as functions of time t for various affected by the reset. As in the earlier case, the particle
values of r. As expected, the MSD saturates faster begins its motion at the origin with an initial orientation
to its stationary value with increasing r. The leading- θ = 0 at time t = 0. Consequently, the distribution of
order time-scale of relaxation to the stationary value is θ at any time t remains Gaussian in nature, with a zero
pos+ori
τrelax = min[(r + Dθ )−1 , (r + 4Dθ )−1 ]. mean and a variance of 2Dθ t. The corresponding renewal
 7

equation is formulated as follows: Let us consider the trajectory of the particle over the
interval [θ, t]. If no resetting occurs during this interval,
the position evolves according to the motion of standard
Pr (x, y, θ, t|0, 0, θ0 ) = e−rt P0 (x, y, θ, t|0, 0, θ0 )
Z t Z Z Z active Brownian particle. For trajectories involving at
+r dτ e −rτ
dx ′
dy ′
dθ′ Pr (x′ , y ′ , θ′ , t − τ |0, 0, θ0 ) least one resetting event, let the time elapsed since the
0 last resetting be denoted by τ . The position at time t is
P0 (x, y, θ, τ |0, 0, θ′ ) then determined by the free ABP evolution during the
(26) interval τ , starting from an arbitrary orientation θt−τ ,
which itself is governed by the Brownian motion of θ.
The propagator for x in this scenario can be derived To compute the position distribution, we integrate over
by integrating the previously stated equation over the all possible values of τ ∈ [0, t] and θt−τ ∈ [−∞, ∞]. Com-
variables y, θ, x′ , and y ′ , as follows: bining all these contributions, the renewal equation is
expressed as:
Pr (x, t|0, θ0 ) = e−rt P0 (x, t|0, θ0 )
Z t Z t
Z (27)
+r dτ e−rτ dθ′ P0 (θ′ , t − τ |θ0 )P0 (x, τ |0, θ′ ) P (x, y, t) = e−rt P0 (x, y, t) + r dτ e−rτ ×
0 0
− 4D θ2 (31)
Z ∞
The corresponding nth moment for x in this scenario e θ (t−τ )
Pθ0 (x, y, τ ) p
can be expressed as: −∞ 4πDθ (t − τ )

Here, we use the notation Pθ0 (x, y, τ ) to represent the

n −rt n
⟨x (t)|θ0 ⟩r = e ⟨x (t)|θ0 ⟩0 probability that the free active Brownian particle (ABP)
Z t Z is located at (x, y) at time τ , given that it started with
−rτ
+r dτ e dθ′ P0 (θ′ , t − τ |θ0 )⟨xn (τ )|θ′ ⟩0 an initial orientation θ at τ = 0. The structure of the re-
0
(28) newal equation described above differs significantly from
that obtained for the position-orientation resetting pro-
Now, we have calculated the second moment of x for the tocol. Consequently, the resulting behavior is also antic-
anisotropic ABP putting n = 2 in Eq.(28) and extracting ipated to exhibit distinct characteristics.
⟨x2 (t)|θ0 ⟩0 from Eq.(21) The second moment for initial angle θ0 is calculated as
(detailed calculations are shown in Appendix(B)),
h ∆D cos 2θ0
⟨x2 (t)|θ0 ⟩r = e−rt 2D̄t + (1 − e−4Dθ t )
4Dθ v02 2D̄
2 2 ⟨x2 (t)|θ0 ⟩r = +
v0 cos 2θ0 v i
r(r + D ) r
+ (3 − 4e−Dθ t + e−4Dθ t ) + 02 (Dθ t + e−Dθ t − 1) θ
12Dθ2 Dθ
" #
−rt v02 cos 2θ0 (r − 6Dθ ) v02 2D̄ ∆D cos 2θ0
(θ ′ −θ0 )2 +e − − −
Z t Z −
e 4Dθ (t−τ ) h 6Dθ2 (r − 4Dθ ) rDθ r r − 4Dθ
−rτ ′
+r dτ e dθ p 2D̄τ " #
0 4πDθ (t − τ ) v 2
cos 2θ 0 (r 2
− 6rD θ + 12D 2
) ∆D cos 2θ0
+ e−4Dθ t 0 θ
+
∆D cos 2θ′ −4Dθ τ v02 cos 2θ0 −Dθ τ −4Dθ τ 12Dθ2 (r − 3Dθ )(r − 4Dθ ) r − 4Dθ
+ (1 − e )+ 2 (3 − 4e + e )
4Dθ 12Dθ " #
2 −(r+Dθ )t v02 v02 cos 2θ0 (r − 4Dθ )
v i +e −
+ 02 (Dθ τ + e−Dθ τ − 1) Dθ (r + Dθ ) 4Dθ2 (r − 3Dθ )
Dθ
(29) (32)

We will use the following properties to calculate the Putting n = 1 in Eq.(28) we can find the value of
above integral ⟨x2 (t)|θ0 ⟩r for initial orientation θ0 = 0 as,
v0
⟨x(t)|θ0 ⟩r = (e−Dθ t − e−rt ) (33)
"Z (θ−θ0 )2 # (r − Dθ )
− 4D
e θ (t−τ )
⟨cos 2θ(t)|θ0 ⟩ = Re dθei2θ p and ⟨y(t)⟩ = 0. Note that, for r = Dθ the above equation
4πDθ (t − τ )
remains well defined when t ≪ Dθ−1 . At short times, i.e.
= cos 2θ0 e−4Dθ (t−τ ) (30) for t ≪ min(r−1 , Dθ−1 )
(θ−θ0 )2
Z − 4D (t−τ
e θ )
and dθ p =1
4πDθ (t − τ ) ⟨x(t)⟩ = v0 t − v0 (r + Dθ )t2 + O(t3 ) (34)
 8

 rv02  v02
σx2 (t)pr = 2D∥ + t+ − − rD̄
12Dθ2 2
∆D(r + 4Dθ ) v02 
− + 18Dθ2
2 3(3Dθ − r)(4Dθ − r)
!
49r2 r4 9r3 9r3  2
− 21rDθ + − + − t
FIG. 6. Position resetting: Mean square displacement as a 8 8Dθ2 4Dθ 4
function of time t for Dθ = 1, D∥ = 1, D⊥ = 0.1, v0 = 1, !
θ0 = 0 and different values of resetting rate r. Symbols rep- v02 (r + Dθ )(24Dθ2 − 4rDθ − r2 ) rv02 3
+ − t + O(t4 )
resent the data from simulations. Solid lines are representing 8Dθ (3Dθ − r) 2
expressions from Eq.(35) and (36). (37)

and

The MSD along x direction can be found as σx2 (t)pr = v02 (r − 4Dθ )
σy2 (t)pr = + 2D⊥
⟨x2 (t)|θ0 = 0⟩r − ⟨x(t)|θ0 = 0⟩2r , here σx2 (t)pr denotes as 4Dθ (r − 3Dθ )(r − 4Dθ )
MSD along x due to position reset at θ0 = 0. v02 (27r2 Dθ − 12rDθ2 − r3 − 48Dθ3 )
+
12Dθ2 (r − 3Dθ )(r − 4Dθ )
!
v02 (r2 − 6rDθ + 12Dθ2 ) ∆D(r + 4Dθ )
+ t+ − rD̄
3Dθ (r − 3Dθ )(r − 4Dθ ) 2
v02 2D̄
σx2 (t)pr = + v02 r r2 v02 (r − 6Dθ ) 2v 2 (r2 − 6rDθ + 12Dθ2 )
r(r + Dθ ) r − − 2 − 0
" # 2Dθ 12Dθ (r − 4Dθ ) (r − 4Dθ )(r − 3Dθ )
2 2
v (r − 6Dθ ) v 2 D̄ ∆D !
+ e−rt 0
− 0 − − v02 (r − Dθ )2 (r2 + rDθ − 16Dθ2 ) 2 rv02
6Dθ2 (r − 4Dθ ) rDθ r r − 4Dθ + t − 1
" # 8Dθ2 (r − 3Dθ )(r + Dθ ) 2
−4Dθ t v02 (r2 − 6rDθ + 12Dθ2 ) ∆D !
+e + (35) (r + Dθ )(r − 4Dθ ) 3
12Dθ2 (r − 3Dθ )(r − 4Dθ ) r − 4Dθ + t + O(t4 )
" # 4Dθ (r − 3Dθ )
2 2
v v (r − 4Dθ ) (38)
+ e−(r+Dθ )t 0
− 0
Dθ (r + Dθ ) 4Dθ2 (r − 3Dθ )
At short times, Eqs. (37) and (38) indicate that the sys-
v02 tem exhibits super-diffusive behavior, highlighting the
− (e−Dθ t − e−rt )2
(r − Dθ )2 pronounced anisotropy in the dynamics. Conversely, at
long times t ≫ max(r−1 , Dθ−1 ), both σx2 and σy2 settle
into a stationary state. Interestingly, at this stage, their
Similarly MSD along y axis becomes for θ0 = 0
values become identical.

v02 2D̄ v02 2D̄

σy2 (t)pr = + (σx2 )pr = (σy2 )pr = + (39)
r(r + Dθ ) r r(r + Dθ ) r
" #
2
−rt v0 (r − 6Dθ ) v02 2D̄ ∆D
−e + + − Eq.(39) demonstrates that the anisotropy disappears in
6Dθ2 (r − 4Dθ ) rDθ r r − 4Dθ the stationary state, resulting in effective values for the
" #
∆D v02 (r2 − 6rDθ + 12Dθ2 ) mean squared displacements (MSDs) that are influenced
−4Dθ t
+e − − by the propulsion velocity v0 of the particle and the av-
r − 4Dθ 12Dθ2 (r − 3Dθ )(r − 4Dθ )
" # erage diffusion coefficient D̄. Fig.(6) presents the plots of
−(r+Dθ )t v02 v02 (r − 4Dθ ) σx2 (t)pr and σy2 (t)pr for various resetting rates, along with
+e +
Dθ (r + Dθ ) 4Dθ2 (r − 3Dθ ) the corresponding results from numerical simulations.
(36) The leading order time-scale of relaxation to the sta-
pos
tionary value is τrelax = min[r−1 , (4Dθ )−1 , (r + Dθ )−1 ].
There is a very interesting difference in relaxation time-
For very small t, we can expand the exponential terms of scale of the first and second type resetting protocols.
Eqs.(35) and (36) getting expressions In the position-only reset, angular diffusion (4Dθ term)
 9

1.6 1.6

1.4 1.4

1.2 1.2

1.0 1.0

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
10.0 7.5 5.0 2.5 0.0 2.5 5.0 7.5 10.0 10.0 7.5 5.0 2.5 0.0 2.5 5.0 7.5 10.0

FIG. 7. Position distribution along x of anisotropic ABP un- FIG. 8. Position distribution along y of anisotropic ABP un-
der position resetting rate at t = 100 (stationary state) when der position resetting rate at t = 100 (stationary state) when
Dθ = 0.1, D∥ = 1, D⊥ = 0.1, v0 = 1 and θ0 = 0 for different Dθ = 0.1, D∥ = 1, D⊥ = 0.1, v0 = 1 and θ0 = 0 for different
values of reset rate r as shown in the legends. values of reset rate r as shown in the legends.

C. Orientational Resetting
plays a significant independent role since orientation dy-
namics are not reset. When rotational diffusion constant Let us now study the case of orientational resetting,
is very very small then the relaxation time scale depends when θ undergoes reset to θ = 0 with rate r, while x and y
on the term 4Dθ . This adds a separate time scale to diffuse usually. In this scenario, the position distribution
the competition. In the position-and-orientation reset, does not directly adhere to any renewal equation, whereas
the dynamics of orientation are directly tied to the reset the θ distribution does. We follow the general discussion
rate r, reducing the competition to only two combined of the renewal equation approach
rates. For position-only reset, when Dθ is very small, the
relaxation time scale depends on 4Dθ . For position-and- Z t
orientation reset, when Dθ is very small, the relaxation
Pr (x, y, θ, t|0, 0, θ0 ) = e−rt P0 (x, y, θ, t|0, 0, θ0 ) + r dτ e−rτ
time scale depends on the reset rate r, as orientation is 0
Z Z Z
reset along with position. ′ ′
dx dy dθ Pr (x , y , θ , t − τ |0, 0, θ0 )P0 (x, y, θ, τ |x′ , y ′ , θ0 )
′ ′ ′ ′

Figs.(7) and (8) present the simulation results for the (40)
position distribution in the long-time stationary state for
both the x and y components. As observed in the previ- We examine the third resetting protocol, in which the
ous section, the x and y position probability distributions orientation θ resets to θ = 0 at a rate r, while the po-
are distinct under position-orientation resetting. How- sition remains unchanged. Let P (θ, t|θ′ , t′ ) represent the
ever, in the case where only position resetting is imple- probability that the orientation is θ at time t, given it
mented, the probability distributions become identical was θ′ at an earlier time t′ . This conditional probability
in both the x and y directions. When position resetting P (θ, t|θ′ , t′ ) adheres to a renewal equation:
occurs, the particle’s position is reset independent of its ′
P (θ, t|θ′ , t′ ) = e−r(t−t ) P0 (θ, t|θ′ , t′ )
orientation or velocity direction. The subsequent motion,
Z t−t′ (41)
influenced by the anisotropy, may lead to an anisotropic
+r dτ e−rτ P0 (θ, τ |0, 0)
trajectory, but the distribution of positions after the re- 0
set is governed by the reset distribution itself, which is
usually symmetric in both the x and y directions. Even where P0 (θ, t|θ′ , t′ ) denotes the propagator for the stan-
if the particle is biased in a particular direction by its dard Brownian motion.
self-propulsion, the reset process essentially "erases" the The final form of the second moment of x can be found
effect of previous motion, resetting the position in a sym- using Eqs.(C7),(C9), and (C13) (detailed calculations are
metric manner. done in Appendix(C))
 10

"
 r∆D cos 2θ0 
⟨x2 (t)|θ0 ⟩or = 2D̄ + t
r + 4Dθ
#"
4Dθ ∆D cos 2θ0 −(r+4Dθ )t v02
+ (1 − e ) + r2 t2
(r + 4Dθ )2 (r + Dθ )2
#
2Dθ t(2Dθ2 + 9rDθ + r2 ) 4v 2 e−(r+4Dθ )t
+ + 0
(r + Dθ )(r + 4Dθ ) 3(r + 4Dθ )2

6v02 Dθ2 (2Dθ2 + 16rDθ + 5r2 ) 2v02 e−(r+Dθ )t

− + rDθ t
(r + Dθ )4 (r + 4Dθ )2 (r + Dθ )3
!
4Dθ3 + 33rDθ2 − 2r3
+ FIG. 9. Orientation resetting: Mean square displacement
3(r + Dθ )(r + 4Dθ )
along x as a function of time t for Dθ = 0.1, D∥ = 1,
(42) D⊥ = 0.1, v0 = 1, θ0 = 0 and varying reset rate r. Symbols
represent the data from simulations. Solid lines are represent-
At θ0 = 0 the above expression becomes, ing expression from Eq.(44).
"
2
 r∆D 
⟨x (t)|θ0 = 0⟩or = 2D̄ + t
r + 4Dθ
"#
4Dθ ∆D −(r+4Dθ )t v02
+ (1 − e ) + r 2 t2
(r + 4Dθ )2 (r + Dθ )2
#
2Dθ t(2Dθ2 + 9rDθ + r2 ) 4v 2 e−(r+4Dθ )t (43)
+ + 0
(r + Dθ )(r + 4Dθ ) 3(r + 4Dθ )2
6v02 Dθ2 (2Dθ2 + 16rDθ + 5r2 )
−
(r + Dθ )4 (r + 4Dθ )2
!
2v 2 e−(r+Dθ )t 4Dθ3 + 33rDθ2 − 2r3
+ 0 3
rDθ t +
(r + Dθ ) 3(r + Dθ )(r + 4Dθ )

The MSD σx2 (t)or = ⟨x2 (t)|θ0 ⟩or − ⟨x(t)|θ0 ⟩2or is then FIG. 10. Orientation resetting: Mean square displacement
along y as a function of time t for Dθ = 0.1, D∥ = 1,
given by using Eqs.(43) and (C6)
D⊥ = 0.1, v0 = 1, θ0 = 0 and varying reset rate r. Symbols
represent the data from simulations. Solid lines are represent-
" ing expression from Eq.(45).
 r∆D 
σx2 (t)or = 2D̄ + t
r + 4Dθ
# " Similarly, we can calculated ⟨y 2 (t)|θ0 = 0⟩or as written,
4Dθ ∆D −(r+4Dθ )t v02
+ (1 − e ) + r 2 t2 " ! #
(r + 4Dθ )2 (r + Dθ )2 2 r∆D 4Dθ ∆D  −(r+4Dθ )t

σy (t)or = 2D̄ − t− 1−e
# r + 4Dθ (r + 4Dθ )2
2Dθ t(2Dθ2 + 9rDθ + r2 ) 4v 2 e−(r+4Dθ )t
+ + 0 4v02 Dθ t 4v02 Dθ (2r + 5Dθ )
(r + Dθ )(r + 4Dθ ) 3(r + 4Dθ )2 + −
(44) (r + 4Dθ )(r + Dθ ) (r + 4Dθ )2 (r + Dθ )2
6v 2 D2 (2Dθ2 + 16rDθ + 5r2 )
− 0 θ
" #
(r + Dθ )4 (r + 4Dθ )2 4v02 e−(r+Dθ )t e−(r+4Dθ )t
+ −
! 3 (r + Dθ )2 (r + 4Dθ )2
2v 2 e−(r+Dθ )t 4Dθ3 + 33rDθ2 − 2r3
+ 0 rD θ t + (45)
(r + Dθ )3 3(r + Dθ )(r + 4Dθ )
" #2 Fig. (9) and (10) present the simulation results of the
v02 Dθ
− rt + (1 − e−(r+Dθ )t ) mean-square displacement (MSD) along the x and y-axes
(r + Dθ )2 r + Dθ
respectively as a function of time t. These results align
well with the analytical expressions derived in Eq.(44)
 11

At very short-times, we obtain the MSD x- component

by expanding the exponential terms
"
2 4v02 v02 Dθ2
σx (t)or = − + 2D∥
3(r + 4Dθ )2 (r + Dθ )4
#
20v02 Dθ3 − 4v02 r3 + 12r2 Dθ (1 − v02 ) + 6rDθ2 (8 − v02 )
+ t
3(r + Dθ )3 (r + 4Dθ )
"
2v02 2v 2 (r3 + 4Dθ3 + 9rDθ2 + 6r2 Dθ ) i 2
+ − 2Dθ ∆D − 0 t
3 3(r + Dθ )2 (r + 4Dθ )
"
2rv02 Dθ (10Dθ3 − 3r3 − 24rDθ2 − 12r2 Dθ )
+
3(r + Dθ )3 (r + 4Dθ )
FIG. 11. Orientation resetting: Comparison of mean-square
# "
8r∆DDθ2 2rv02 Dθ 3 16r2 Dθ2 v02
displacements along x and y- axes for Dθ = 0.1,D∥ = 1, + + t +
D⊥ = 0.1, v0 = 1, θ0 = 0 and r = 1.
r + 4Dθ r + Dθ 3(r + 4Dθ )2
16r2 Dθ3 ∆D 4r2 v02 Dθ4 2r2 Dθ2 v02
− 2
− 4
−
(r + 4Dθ ) (r + Dθ ) (r + Dθ )2
#
and (45), validating the theoretical framework. These fig- r2 v02 Dθ2 (8Dθ3 + 18rDθ2 − r3 ) 4
ures illustrate that the particle’s dynamics do not reach + t + O(t5 )
3(r + Dθ )4 (r + 4Dθ )
a steady state despite the implementation of orienta-
(46)
tional resetting. This behavior contrasts with scenarios
involving complete or solely positional resetting proto- Similarly, we also calculate the mean-square displace-
cols, which typically drive the system toward steady-state ment of the y-component at very short time,
dynamics. The dynamics initially exhibit pure diffusive !
behavior, transitioning into a super-diffusive regime over 2 4Dθ v02
σy (t)or = 2D⊥ − t + 2Dθ ∆Dt2
intermediate times. At long times, the system eventually 3(r + Dθ )(r + 4Dθ )
reverts to a diffusive nature. An intriguing phenomenon " #
8v02 rDθ 2rDθ v02 8r∆DDθ2 (r + 4) 3
observed in this case is the persistence of anisotropy even + − − t
3(r + 4Dθ ) 3(r + Dθ ) (r + 4Dθ )2
in the long-time regime, as evidenced by both the sim- " #
ulation results and the analytical expression in Eq.(50). r2 Dθ2 v02 16r2 Dθ2 (∆D + v02 /3) 4
This contrasts sharply with the scenario under no-reset + + t + O(t5 )
3(r + Dθ )2 (r + 4Dθ )2
conditions, as discussed in Mandal et al.24 , where the sys-
(47)
tem transitions to isotropic behavior at long time. This
highlights the role of orientational resetting in maintain- The transient behavior can once again be interpreted
ing anisotropic dynamics over extended periods. In the through an analysis of the dynamical exponents govern-
intermediate super-diffusive regime, characterized by a ing the motion along the x and y directions. In the long-
time dependence of tn (where n > 1), the degree of term limit, since all terms increase linearly over time, we
super-diffusivity is strongly influenced by the resetting can ignore the constant offset in the equations above, and
rate r. Lower resetting rates amplify the super-diffusive the MSD along the x and y axes are given by,
behavior, making it more pronounced, whereas higher re-
setting rates tend to suppress it, steering the dynamics
 r∆D 2v02 Dθ2 (2Dθ + 5r) 
σx2 (t)or ≈ 2D̄ + + t
more quickly toward diffusive behavior. The results cor- r + 4Dθ (r + Dθ )3 (r + 4Dθ )
!
responding to only orientational resetting present a stark r∆D 4v02 Dθ
contrast to those reported by Kumar et al.66 . In their σy2 (t)or ≈ 2D̄ − + t
r + 4Dθ (r + 4Dθ )(r + Dθ )
study, the isotropic nature of active Brownian particles (48)
(ABPs) precluded any initial diffusive behavior. Here,
however, the introduction of anisotropy leads to an ini- The dynamics in x and y directions conversely goes
tial diffusive regime, highlighting the significant impact through super-diffusive regime, before also re-entering a
of anisotropic properties on the dynamics of the system. diffusive regime.
To better understand the nature of these crossovers, we Fig.(11) presents a comparative analysis of the mean-
investigate the short-term and long-term behaviors of the square displacements (MSD) along the x and y directions
mean square displacements. for a specific reset rate of r = 1. The figure reveals that
 12

the anisotropy observed at early times persists even at

long times due to the orientational resetting. This be-
havior is particularly intriguing when contrasted with the 0.030
no-reset condition, as discussed in the work by Mandal
et al.24 . For the no-reset condition, the initial anisotropy 0.025

observed in the anisotropic ABP dynamics diminishes

0.020
over time, eventually vanishing at long times, leading the
particle to behave like an isotropic one. The effect of re- 0.015
setting becomes apparent at higher orders, introducing
an additional anisotropy. This behavior is expected, as 0.010

the resetting condition θ = 0 is fundamentally anisotropic

0.005
in nature. Interestingly, the impact of this anisotropy
persists even at late times – despite the fact that both x- 0.000
and y-motions ultimately become diffusive. 20 0 20 40 60 80 100 120

In the long-term limit, we can anticipate that accord-

ing to the central limit theorem, the two-dimensional
FIG. 12. Plot of long time position distribution along x of
probability distribution in the Cartesian components
anisotropic ABP under orientation resetting when Dθ = 0.1,
takes on a Gaussian form, and is given by D∥ = 1, D⊥ = 0.1, v0 = 1 and θ0 = 0 at t = 100 for different
values of reset rate r as shown in the legends.
" #
1 1
P (x, y, t) = p exp − (x, y)T D−1 ef f (x, y)
4π Det(Def f ) 4t
(49) 0.05

where the effective diffusion coefficients D can be written

for x and y axes as, 0.04

ef f r∆D 2v02 Dθ2 (2Dθ + 5r)

Dxx ≈ 2D̄ + + 0.03
r + 4Dθ (r + Dθ )3 (r + 4Dθ )
(50)
ef f r∆D 4v02 Dθ
Dyy ≈ 2D̄ − + 0.02
r + 4Dθ (r + 4Dθ )(r + Dθ )
The marginal probability distribution of x can be found 0.01
by integrating out the y variable, as given by
0.00
30 20 10 0 10 20 30
Z
1 h (x − µ (t))2 i
x
P (x, t) = dyP (x, y, t) = q exp − x t
x
4πDef t 4D ef f
f
FIG. 13. Plot of long time position distribution along y of
(51) anisotropic ABP under orientation resetting when Dθ = 0.1,
Where µx (t) = ⟨x(t)⟩ which has finite value as shown in D∥ = 1, D⊥ = 0.1, v0 = 1 and θ0 = 0 at t = 100 for different
the Eq.(C6). A similar argument applies to the marginal values of reset rate r as shown in the legends.
probability distribution of y, where the mean µy (t) is
always zero. One of the most intriguing outcomes of
applying orientation resetting in anisotropic ABP is the The propulsion velocity, aligned with the major axis of
absence of a stationary state over extended time periods, the particle, causes the orientation resetting to shift the
with the anisotropy persisting. However, in the case of Gaussian peak in the positive x-direction as the reset-
ABP without resetting, the anisotropy gradually dimin- ting rate increases. This directional bias emerges be-
ishes in the long-time limit24 . cause higher resetting rates reinforce alignment along the
Figs.(12) and (13) depict the simulation results of po- propulsion axis, effectively steering the particle’s position
sition distributions along the x- and y-axes, respectively. distribution towards the positive x-axis over time.
These distributions exhibit Gaussian profiles, reflecting
the expected behavior of the particle under the influ-
ence of orientational resetting. This Gaussian nature
underscores the underlying stochastic processes govern-
ing the particle’s dynamics in both spatial dimensions.
 13

IV. CONCLUSIONS detailed understanding of the system’s behavior under

resetting conditions.
In this article, we investigated how shape asymme- We have demonstrated that the orientation-only reset-
try influences the two-dimensional ABP of an anisotropic ting scheme is insufficient to drive the system to a steady
particle under stochastic resetting. Specifically, we exam- state. However, higher resetting rates effectively suppress
ined the behavior of moments, and steady-state distribu- the intermediate hyper-diffusivity observed in the dy-
tions at both short and long times, considering different namics of the anisotropic ABP. Conversely, when orienta-
resetting schemes. It is well-established that anisotropic tional resetting is introduced, the short-time anisotropy
diffusion only persists over very short time scales, even- in the particle’s motion persists into late times, reflecting
tually transitioning to isotropic diffusion at longer times. the influence of resetting on the dynamics.
In practical scenarios, the short-time dynamics are often This study paves the way for extending investigations
insignificant, enabling the motion of an asymmetric par- into dynamical first passage properties, counting statis-
ticle to be effectively modeled by Langevin equations for tics, exit time statistics, and cover time for anisotropic
a point particle. This simplification uses an isotropic dif- biological particles, which are predominantly observed in
fusion coefficient, calculated as the mean of the diffusion nature.
coefficients along the particle’s principal and secondary
axes. This approximation has been successfully applied
V. ACKNOWLEDGMENT
in contexts such as free particle motion or when the par-
ticle is confined harmonically in two dimensions67 . For
the case of complete resetting, we have derived both the We acknowledge Prof. Sanjib Sabhapandit of RRI,
time-dependent moments and a perturbative steady-state Bengaluru for his valuable comments. AG expresses grat-
solution. Our analysis includes exact calculations up to itude to Dr. Subhasish Chaki, Heinrich Heine University
the first-order approximation of anisotropy, providing a Düsseldorf for important discussions. AG expresses grat-
itude to RRI, Bengaluru, for providing access to their
computing facility.

[1] Sriram Ramaswamy. Annu. Rev. Condens. Matter Phys., Kaupp, Luis Alvarez, Thomas Kiørboe, Eric Lauga, Wil-
1(1):323–345, 2010. son C K Poon, Antonio DeSimone, Santiago Muiños-
[2] M Cristina Marchetti, Jean-François Joanny, Sriram Landin, Alexander Fischer, Nicola A Söker, Frank Ci-
Ramaswamy, Tanniemola B Liverpool, Jacques Prost, chos, Raymond Kapral, Pierre Gaspard, Marisol Ripoll,
Madan Rao, and R Aditi Simha. Reviews of modern Francesc Sagues, Amin Doostmohammadi, Julia M
physics, 85(3):1143, 2013. Yeomans, Igor S Aranson, Clemens Bechinger, Hol-
[3] Borge ten Hagen, Sven van Teeffelen, and Hart- ger Stark, Charlotte K Hemelrijk, François J Nedelec,
mut Löwen. Journal of Physics: Condensed Matter, Trinish Sarkar, Thibault Aryaksama, Mathilde Lacroix,
23(19):194119, 2011. Guillaume Duclos, Victor Yashunsky, Pascal Silberzan,
[4] M Sandoval. Physical Review E, 87(3):032708, 2013. Marino Arroyo, and Sohan Kale. Journal of Physics:
[5] Tamás Vicsek, András Czirók, Eshel Ben-Jacob, Inon Condensed Matter, 32(19):193001, feb 2020.
Cohen, and Ofer Shochet. Physical review letters, [12] J Schwarz-Linek, C Valeriani, A Cacciuto, ME Cates,
75(6):1226, 1995. D Marenduzzo, AN Morozov, and WCK Poon. Proceed-
[6] John Toner, Yuhai Tu, and Sriram Ramaswamy. Annals ings of the National Academy of Sciences, 109(11):4052–
of Physics, 318(1):170–244, 2005. 4057, 2012.
[7] Nitin Kumar, Harsh Soni, Sriram Ramaswamy, and [13] Gabriel S Redner, Michael F Hagan, and Aparna
AK Sood. Nature communications, 5(1):4688, 2014. Baskaran. Biophysical Journal, 104(2):640a, 2013.
[8] Yaouen Fily and M Cristina Marchetti. Physical review [14] Joakim Stenhammar, Raphael Wittkowski, Davide
letters, 108(23):235702, 2012. Marenduzzo, and Michael E Cates. Physical review let-
[9] Jeremie Palacci, Stefano Sacanna, Asher Preska Stein- ters, 114(1):018301, 2015.
berg, David J Pine, and Paul M Chaikin. Science, [15] Peter Siegle, Igor Goychuk, and Peter Hänggi. Physical
339(6122):936–940, 2013. review letters, 105(10):100602, 2010.
[10] AB Slowman, MR Evans, and RA Blythe. Physical re- [16] Ion Santra, Urna Basu, and Sanjib Sabhapandit.
view letters, 116(21):218101, 2016. Journal of Physics A: Mathematical and Theoretical,
[11] Gerhard Gompper, Roland G Winkler, Thomas Speck, 55(38):385002, 2022.
Alexandre Solon, Cesare Nardini, Fernando Peruani, [17] Yunyun Li, Fabio Marchesoni, Tanwi Debnath, and Pu-
Hartmut Löwen, Ramin Golestanian, U Benjamin lak K Ghosh. Physical Review E, 96(6):062138, 2017.
 14

[18] Xue Ao, Pulak K Ghosh, Yunyun Li, Gerhard Schmid, [42] Olivier Bénichou, Claude Loverdo, Michel Moreau, and
Peter Hänggi, and Fabio Marchesoni. The European Raphael Voituriez. Reviews of Modern Physics, 83(1):81,
Physical Journal Special Topics, 223:3227–3242, 2014. 2011.
[19] Alexandre P Solon, Michael E Cates, and Julien [43] A Chechkin and Igor M Sokolov. Physical review letters,
Tailleur. The European Physical Journal Special Topics, 121(5):050601, 2018.
224(7):1231–1262, 2015. [44] Arnab Pal and Shlomi Reuveni. Physical review letters,
[20] Amir Shee, Abhishek Dhar, and Debasish Chaudhuri. 118(3):030603, 2017.
Soft Matter, 16(20):4776–4787, 2020. [45] Arnab Pal, Łukasz Kuśmierz, and Shlomi Reuveni. Phys-
[21] Urna Basu, Satya N Majumdar, Alberto Rosso, and Gré- ical Review Research, 2(4):043174, 2020.
gory Schehr. Physical Review E, 100(6):062116, 2019. [46] Susanna C Manrubia and Damián H Zanette. Physical
[22] J. Tailleur and M. E. Cates. Phys. Rev. Lett., 100:218103, Review E, 59(5):4945, 1999.
May 2008. [47] Paolo Visco, Rosalind J Allen, Satya N Majumdar, and
[23] Ion Santra, Urna Basu, and Sanjib Sabhapandit. Physical Martin R Evans. Biophysical journal, 98(7):1099–1108,
Review E, 104(1):L012601, 2021. 2010.
[24] Sudipta Mandal and Anirban Ghosh. Diffusion dynamics [48] Andrea Montanari and Riccardo Zecchina. Physical re-
of an overdamped active ellipsoidal brownian particle in view letters, 88(17):178701, 2002.
two dimensions. Journal of Statistical Mechanics: Theory [49] Edo Kussell and Stanislas Leibler. Science,
and Experiment, 2024(7):073209, 2024. 309(5743):2075–2078, 2005.
[25] Urna Basu, Satya N. Majumdar, Alberto Rosso, and Gré- [50] Edo Kussell, Roy Kishony, Nathalie Q Balaban, and
gory Schehr. Phys. Rev. E, 98:062121, Dec 2018. Stanislas Leibler. Genetics, 169(4):1807–1814, 2005.
[26] FC Bawden, NW Pirie, JD Bernal, and I Fankuchen. [51] Édgar Roldán, Ana Lisica, Daniel Sánchez-Taltavull, and
Nature, 138(3503):1051–1052, 1936. Stephan W Grill. Physical Review E, 93(6):062411, 2016.
[27] Xin Wen, Robert B Meyer, and DLD Caspar. Physical [52] Martin R Evans and Satya N Majumdar. Physical review
review letters, 63(25):2760, 1989. letters, 106(16):160601, 2011.
[28] Hartmut Graf and Hartmut Löwen. Physical Review E, [53] Martin R Evans and Satya N Majumdar. Journal of
59(2):1932, 1999. Physics A: Mathematical and Theoretical, 44(43):435001,
[29] Amin Doostmohammadi, Sumesh P Thampi, and Ju- 2011.
lia M Yeomans. Physical review letters, 117(4):048102, [54] Satya N Majumdar, Sanjib Sabhapandit, and Grégory
2016. Schehr. Physical Review E, 91(5):052131, 2015.
[30] Alexei Ivlev, Gregor Morfill, Hartmut Lowen, and [55] Shlomi Reuveni. Physical review letters, 116(17):170601,
C Patrick Royall. volume 5. World Scientific Publish- 2016.
ing Company, 2012. [56] Ofir Tal-Friedman, Arnab Pal, Amandeep Sekhon,
[31] Hartmut Löwen. Physical Review E, 50(2):1232, 1994. Shlomi Reuveni, and Yael Roichman. The journal of
[32] Peter Bolhuis and Daan Frenkel. The Journal of chemical physical chemistry letters, 11(17):7350–7355, 2020.
physics, 106(2):666–687, 1997. [57] Benjamin Besga, Alfred Bovon, Artyom Petrosyan,
[33] Marjolein Dijkstra. Advances in chemical physics, Satya N Majumdar, and Sergio Ciliberto. Physical Re-
156:35–71, 2014. view Research, 2(3):032029, 2020.
[34] Hartmut Löwen. Physical Review E, 59(2):1989, 1999. [58] Felix Faisant, Benjamin Besga, Artyom Petrosyan, Sergio
[35] Jörg Roller, Aleena Laganapan, Janne-Mieke Meijer, Ciliberto, and Satya N Majumdar. Journal of Statistical
Matthias Fuchs, and Andreas Zumbusch. Proceedings of Mechanics: Theory and Experiment, 2021(11):113203,
the National Academy of Sciences, 118(3):e2018072118, 2021.
2021. [59] Semih Cayci, Atilla Eryilmaz, and R Srikant. arXiv
[36] Subhasish Chaki, Kristian Stølevik Olsen, and Hartmut preprint arXiv:2007.00081, 2020.
Löwen. Dynamics of a single anisotropic particle under [60] Jan-Hendrik Lorenz. Theory of Computing Systems,
various resetting protocols. Journal of Physics: Con- 65(8):1143–1164, 2021.
densed Matter, 2024. [61] Jan-Hendrik Lorenz. In SOFSEM 2018: Theory
[37] Christopher G Gray, Keith E Gubbins, and Christo- and Practice of Computer Science: 44th International
pher G Joslin. volume 2. Oxford University Press, 1984. Conference on Current Trends in Theory and Prac-
[38] Vijay Narayan, Sriram Ramaswamy, and Narayanan tice of Computer Science, Krems, Austria, January
Menon. Science, 317(5834):105–108, 2007. 29-February 2, 2018, Proceedings 44, pages 493–507.
[39] R Grima and SN Yaliraki. The Journal of chemical Springer, 2018.
physics, 127(8), 2007. [62] Shlomi Reuveni, Michael Urbakh, and Joseph Klafter.
[40] Anirban Ghosh and Dipanjan Chakraborty. The Journal Biophysical Journal, 106(2):677a, 2014.
of Chemical Physics, 152(17), 2020. [63] Tal Rotbart, Shlomi Reuveni, and Michael Urbakh. Phys-
[41] Martin R Evans, Satya N Majumdar, and Grégory ical Review E, 92(6):060101, 2015.
Schehr. Journal of Physics A: Mathematical and The- [64] Anirban Ghosh, Sudipta Mandal, and Dipanjan
oretical, 53(19):193001, 2020. Chakraborty. The Journal of Chemical Physics,
157(19):194905, 2022.
 15

[65] Martin R Evans and Satya N Majumdar. Journal of

Physics A: Mathematical and Theoretical, 51(47):475003,
2018.
[66] Vijay Kumar, Onkar Sadekar, and Urna Basu. Active
brownian motion in two dimensions under stochastic re-
setting. Physical Review E, 102(5):052129, 2020.
[67] R. Grima and S. N. Yaliraki. Brownian motion of an
asymmetrical particle in a potential field. The Journal
of Chemical Physics, 127(8):084511, 08 2007.
 16

Appendix

Appendix A: Perturbative steady state for weak-asymmetric particles

Although we obtain exact formulas for the lower-order moments, we are also interested in gaining a deeper un-
derstanding of the overall form of the non-equilibrium steady state (NESS) under resetting. For this, we adopt a
simplified perturbative method. First, we compute the propagator without considering resetting, and then we apply
the renewal framework to incorporate resetting and determine the steady state.
Without resetting, the dynamics of the system are governed by the Smoluchowski-Perrin equation, which provides
a framework for describing the motion of anisotropic particles in a diffusive medium.

∂t P0 (x, t|x0 ) = Dθ ∂θ2 P0 (x, t|x0 ) + ∇.[D.∇]P0 (x, t|x0 ) − vn̂.∇P0 (x, t|x0 ) (A1)

where D is the diffusion tensor, Dij = kB T Γij = D∥ ni nj + D⊥ (δij − ni nj ), with n̂ = (cos θ, sin θ), the unit vector
pointing along the particle’s major axis, v is the propulsion velocity of the anisotropic particle along its major axis.
Taking a Fourier transform,

Z
P̂0 (k, θ, t|θ0 ) ≡ dxe−ik.x P0 (x, θ, t|θ0 ) (A2)

The transformed equation becomes,

∂t P̂0 (k, θ, t|θ0 ) = Dθ ∂θ2 P̂0 (k, θ, t|θ0 ) + k[D.k]P̂0 (k, θ, t|θ0 ) − ivk cos θP̂0 (k, θ, t|θ0 ) (A3)

To simplify the analysis, we focus on the steady-state behavior in the x-direction. Accordingly, we consider a wave
vector k = kex with ex the unit vector in the x-direction. Under these conditions, the marginal distribution in the
xx-direction satisfies:

" #
 ∆D 
∂t P̂0 (k, θ, t|θ0 ) = Dθ ∂θ2 − k 2 D̄ + cos 2θ P̂0 (k, θ, t|θ0 ) − ivk cos θP̂0 (k, θ, t|θ0 ) (A4)
2

A straightforward interpretation of this equation is that the diffusivity experienced by the particle in the x-direction,
given by D̄ + ∆D 2 cos 2θ , varies depending on the particle’s orientation. To derive an analytical solution that sheds
light on the steady-state shape under resetting, we adopt a perturbative approach. Specifically, we expand the solution
in powers of the dimensionless asymmetry parameter ϵ ≡ ∆D D̄
. The marginal Smoluchowski-Perrin equation can then
be expressed as:

cos 2θ
∂t P̂0 (k, θ, t|θ0 ) = Dθ ∂θ2 P̂0 (k, θ, t|θ0 ) − k 2 D̄P̂0 (k, θ, t|θ0 ) − ϵk 2 D̄ P̂0 (k, θ, t|θ0 ) − ivk cos θP̂0 (k, θ, t|θ0 ) (A5)
2
We expand the solution in powers of the asymmetry as

∞
X
P̂0 (k, θ, t|θ0 ) = pn (k, θ, t|θ0 )ϵn (A6)
n=0

The function pn (k, θ, t|θ0 ) represents the coefficient of the n-th term in the power series expansion. Substituting
this ansatz into the Smoluchowski-Perrin equation results in a set of coupled equations that govern these coefficients.
These equations are derived by equating terms with the same powers of the asymmetry parameter ϵ, leading to a
hierarchy of equations that must be solved sequentially to determine the behavior of the system under the perturbative
expansion.
 17

∂t p0 (k, θ, t|θ0 ) = Dθ ∂θ2 p0 (k, θ, t|θ0 ) − k 2 D̄p0 (k, θ, t|θ0 ) − ivk cos θp0 (k, θ, t|θ0 ) (A7)

k 2 D̄ cos 2θ
∂t pn (k, θ, t|θ0 ) = Dθ ∂θ2 pn (k, θ, t|θ0 ) − k 2 D̄pn (k, θ, t|θ0 ) − pn−1 (k, θ, t|θ0 ) − ivk cos θpn (k, θ, t|θ0 ) (A8)
2
First, it is important to observe that p0 (k, θ, t) corresponds to the solution characteristic of a symmetric colloid.
Since the focus lies on determining the spatial steady-state distribution to first order in the perturbative expansion, we
integrate over the angular degree of freedom for n = 0 and n = 1. This integration results in expressions that describe
the marginal steady-state distributions, incorporating the leading-order effects of anisotropy in the system. The
integration effectively reduces the problem to a spatial one while capturing the perturbative corrections introduced
by the asymmetry parameter.

∂t p0 (k, t|θ0 ) = −k 2 D̄p0 (k, t|θ0 ) − ivk cos θ0 e−Dθ t p0 (k, t|θ0 ) (A9)

k 2 D̄
∂t p1 (k, t|θ0 ) = −k 2 D̄p1 (k, t|θ0 ) − cos 2θ0 e−4Dθ t p0 (k, t|θ0 ) − ivk cos θ0 e−Dθ t p1 (k, t|θ0 ) (A10)
2
Here the zeroth order solution can be written as

ivk cos θ0 ivk cos θ0

2
D̄t − e−Dθ t
p0 (k, t) = e−k e Dθ
e Dθ
(A11)

If we take an approximation that the propulsion velocity v is very small and in long time e−Dθ t term also goes to
very smaller, so we can expand the exponential term upto first order and we get,

!
−k2 D̄t −
ivk cos θ0 ivk cos θ0 −Dθ t
p0 (k, t) ≈ e e Dθ
1+ e (A12)
Dθ

Performing a Laplace transform of the above equation followed by a Fourier inverse transform, we obtain

β(s) −β(s)|(x− v cos θ0

)| v cos θ0 −β(s+Dθ )|(x− v cos θ0
)|
p̃0 (x, s|θ0 ) = e Dθ
− e Dθ
(A13)
2s 2Dθ D̄
p
here β(s) = s/D̄ is the inverse lengthscale.
The integration in the equation for the first-order correction can be performed exactly, yielding an explicit expression
for the first-order term in the perturbative expansion.

!
k 2 D̄ 2 −
ivk cos θ0 ivk cos θ0 −Dθ t
2
∂t p1 (k, t|θ0 ) = −k D̄p1 (k, t|θ0 )− cos 2θ0 e−(4Dθ +k D̄)t e Dθ
1+ e −ivk cos θ0 e−Dθ t p1 (k, t|θ0 )
2 Dθ
(A14)
Performing a Laplace transform results in

ivk cos θ0
−

k 2 D̄ cos 2θ0 e Dθ Dθ s + k 2 D̄ + 5Dθ + ivk cos θ0 s + k 2 D̄ + 4Dθ )
p̂˜1 (k, s) = −   (A15)
2Dθ 2


2


2


s + k D̄ + 4Dθ s + k D̄ + 5Dθ s + k D̄ + s + Dθ ivk cos θ0

By inverting the Fourier transform, we ultimately derive the first-order correction to the propagator as
 18

q q
s+4Dθ v cos θ s+4Dθ v cos θ
(x− D 0 ) (x− D 0 )
" #
−
cos 2θ0 e D̄ θ e D̄ θ v cos θ0 cos 2θ0 (s + 5Dθ )
p̃1 (x, s|θ0 ) = q − q +
4 4Dθ s+4D D̄
− v cos θ0 (s + Dθ ) 4Dθ s+4D D̄
+ v cos θ0 (s + Dθ ) 4Dθ
θ θ
q q
s+5Dθ v cos θ s+5Dθ v cos θ
(x− D 0 ) (x− D 0 )
" #
−
e D̄ θ e D̄ θ
p + p
5Dθ D̄ − v cos θ0 (s + Dθ ) D̄(s + Dθ ) 5Dθ D̄ + v cos θ0 (s + Dθ ) D̄(s + Dθ )
 − 1 −α(s,v,Dθ )+√α2 (s,v,Dθ )+4sD̄ (x− v cos θ0 )
 
" 2 p
2 2 D̄ Dθ
cos 2θ0 α (s, v, Dθ ) + 2sD̄ − α(s, v, Dθ ) α (s, v, Dθ ) + 4sD̄ e
+ h i
2
p p
α2 (s, v, Dθ ) + 4sD̄ 8Dθ D̄ − α(s, v, Dθ )(α(s, v, Dθ ) − α2 (s, v, Dθ ) + 4sD̄)
 1 α(s,v,Dθ )+√α2 (s,v,Dθ )+4sD̄ (x− v cos θ0 ) #
 
 p Dθ
α2 (s, v, Dθ ) + 2sD̄ + α(s, v, Dθ ) α2 (s, v, Dθ ) + 4sD̄ e 2D̄
− p h p i
α2 (s, v, Dθ ) + 4sD̄ 8Dθ D̄ − α(s, v, Dθ )(α(s, v, Dθ ) + α2 (s, v, Dθ ) + 4sD̄)
 1 α(s,v,Dθ )+√α2 (s,v,Dθ )+4sD̄ (x− v cos θ0 )
 
" 3 √ Dθ
2 2 2 3/2 2 D̄
v cos θ0 cos 2θ0 4α + 3α α + 4sD̄ + 12αsD̄ + (α + 4sD̄) e
+ √  √ 
8Dθ D̄ α2 + 4sD̄ 10Dθ D̄ − α(α + α2 + 4sD̄)
 − 1 −α(s,v,Dθ )+√α2 (s,v,Dθ )+4sD̄ (x− v cos θ0 ) #
 
 √ Dθ
4α3 − 3α2 α2 + 4sD̄ + 12αsD̄ − (α2 + 4sD̄)3/2 e 2D̄
− √  √ 
α2 + 4sD̄ 10Dθ D̄ − α(α − α2 + 4sD̄)
(A16)

here α(s, v, Dθ ) = v cos θ0 (s + Dθ )

The spatial probability distribution up to the first order correction without resetting is the combination of both
p̃0 (x, s) and p̃1 (x, s)

P̃0 (x, s|θ0 ) = p̃0 (x, s|θ0 ) + ϵp̃1 (x, s|θ0 ) (A17)

The term P̃0 (x, s|θ0 ) represents a normalized probability distribution. We can now readily obtain the steady state
distribution under the effect of resetting by using the renewal equation Eq.(13)

Z ∞
Pr (x|0, θ0 ) = r dτ e−rτ P0 (x, τ |0, θ0 ) = rP̃0 (x, r|0, θ0 ) (A18)
0

Substituting the value of P̃0 (x, s) from Eq.(A17) in the Eq.(A18) and replacing s as r, we get
 19

v cos θ
β(r+4Dθ )(x− D 0 )
"
β(r) −β(r)|(x− v cos θ0
)| v cos θ0 −β(r+Dθ )|(x− v cos θ0
)| ϵ cos 2θ0 re θ
P̃r (x|θ0 ) = e Dθ
− e Dθ
+ q
2 2Dθ D̄ 4 4Dθ r+4D D̄
− v cos θ0 (r + Dθ )
θ
v cos θ0 # " v cos θ0
−β(r+4Dθ )(x− D ) β(r+5Dθ )(x− D )
re θ ϵv cos θ0 cos 2θ0 (r + 5Dθ ) re θ
− q + p
4D D̄
+ v cos θ (r + D ) 4Dθ 5Dθ D̄ − v cos θ0 (r + Dθ ) D̄(r + Dθ )
θ r+4Dθ 0 0 θ
v cos θ
−β(r+5Dθ )(x− D 0
#
)
re θ
+ p
5Dθ D̄ + v cos θ0 (r + Dθ ) D̄(r + Dθ )
 − 1 −α(r,v,Dθ )+√α2 (r,v,Dθ )+4rD̄ (x− v cos θ0 )
 
" 2
p
2 2D̄ Dθ
ϵr cos 2θ0 α (r, v, Dθ ) + 2rD̄ − α(r, v, Dθ ) α (r, v, Dθ ) + 4rD̄ e
+ h i
2
p p
α2 (r, v, Dθ ) + 4rD̄ 8Dθ D̄ − α(r, v, Dθ )(α(r, v, Dθ ) − α2 (r, v, Dθ ) + 4rD̄)
 1 α(r,v,Dθ )+√α2 (r,v,Dθ )+4rD̄ (x− v cos θ0 ) #
 
 p Dθ
α2 (r, v, Dθ ) + 2rD̄ + α(r, v, Dθ ) α2 (r, v, Dθ ) + 4rD̄ e 2D̄

− p h p i
α2 (r, v, Dθ ) + 4rD̄ 8Dθ D̄ − α(r, v, Dθ )(α(r, v, Dθ ) + α2 (r, v, Dθ ) + 4rD̄)
(A19)

There are several noteworthy aspects to consider in the case of an anisotropic active Brownian particle (ABP). The
steady-state probability distribution at zeroth order is defined by a characteristic length scale, represented by β(r),
and the particle’s activity introduces an additional length scale, β(r + Dθ ), which is influenced by the resetting rate.
The distribution exhibits a peak at x = 0 as a result of the restrictive influence exerted by resetting at the origin.
This peak value rises with an increase in the resetting rate r, as anticipated.
When the anisotropy of the particle is considered, the solution becomes dependent on the changes in length scales
and is influenced by the propulsion velocity, reset rate, rotational diffusivity and the initial orientation angle θ0 ,
introduced by the anisotropy of the particle. Interestingly, the rotational dynamics influence the spatial steady-state
distribution through the coupling between translational and rotational degrees of freedom, an effect that endures even
in the long-time regime.

Appendix B: EXACT COMPUTATION OF THE MOMENTS FOR THE POSITION RESETTING

The second moment of x for the anisotropic ABP putting n = 2 in Eq.(28)

h ∆D cos 2θ0 v 2 cos 2θ0 v02 i
⟨x2 (t)|θ0 ⟩r = e−rt 2D̄t + (1 − e−4Dθ t ) + 0 (3 − 4e −Dθ t
+ e −4Dθ t
) + (D θ t + e −Dθ t
− 1)
4Dθ 12Dθ2 Dθ2
′
(θ −θ0 )2
Z t −
∆D cos 2θ′ v 2 cos 2θ0
Z
−rτ ′ e 4Dθ (t−τ ) h
+r dτ e dθ p 2D̄τ + (1 − e−4Dθ τ ) + 0 2 (3 − 4e
−Dθ τ
+ e−4Dθ τ )
0 4πDθ (t − τ ) 4D θ 12D θ
v02 i
+ 2 (Dθ τ + e−Dθ τ − 1)
Dθ
(B1)

We will use the following properties to calculate the above integral

"Z (θ−θ0 )2 #
− 4D (t−τ
e θ )
⟨cos 2θ(t)|θ0 ⟩ = Re⟨ei2θ |θ0 ⟩ = Re dθei2θ p = cos 2θ0 e−4Dθ (t−τ )
4πDθ (t − τ )
(B2)
(θ−θ0 )2
Z −
e 4Dθ (t−τ )
dθ p =1
4πDθ (t − τ )
 20

Appendix C: EXACT COMPUTATION OF THE MOMENTS FOR THE ORIENTATION RESETTING

The Langevin Eqs.(4) can be formally integrated to write

Z t Z t
x(t) = dt′ ηx (t′ ) + dt′ v0 cos θ(t′ )
0 0
Z t Z t (C1)
y(t) = dt′ ηx (t′ ) + dt′ v0 sin θ(t′ )
0 0

Taking the statistical average over all possible trajectories of θ, we obtain the following expression
Z t D E
⟨x(t)⟩ = dt′ v0 cos θ(t′ )
0
Z t D E (C2)
⟨y(t)⟩ = dt′ v0 sin θ(t′ )
0

Let P(θ, t|θ , t ) represent the probability that the orientation takes the value θ at time t given that it was θ′ at an
′ ′

earlier time t′ . The evolution of this probability distribution satisfies a renewal equation, which captures the dynamics
of the system in the presence of resetting events.

Z (t−t′ )
′ ′ −r(t−t′ ) ′ ′
P(θ, t|θ , t ) = e P0 (θ, t|θ , t ) + r dτ e−rτ P0 (θ, τ |0, 0) (C3)
0

where P0 (θ, t|θ′ , t′ ) denotes the propagator for the standard Brownian motion, given as
" #
′ ′ 1 (θ − θ′ )2
P0 (θ, t|θ , t ) = p exp − (C4)
4πDθ (t − t′ ) 4Dθ (t − t′ )

The averages appearing on the right-hand side of the Eq.(C2) can be computed using the renewal equation Eq.(C3)
for P(θ, t). We get,
Z +∞
Dθ r
⟨cos θ(τ )⟩ = dθ cos θP(θ, τ ) = e−(r+Dθ )τ + (C5)
−∞ r + Dθ r + Dθ
and ⟨sin θ(s)⟩ = 0. Using the above expression in Eq.(C2) we can calculate the exact expression for the mean of x
position as,
" #
v0 Dθ  
⟨x(t)|θ0 ⟩ = rt + 1 − e−(r+Dθ )t (C6)
r + Dθ r + Dθ

And ⟨y(t)⟩ = 0

1. Variance of x(t) and y(t)

We get from Eq.(C1)

Z t D Z t E Z t Z t D E
⟨x2 (t)|θ0 ⟩or = v02 dτ ′ cos θ(τ ) cos θ(τ ′ ) +
dτ dτ dτ ′ η(τ )η(τ ′ )
0 0 0 0
Z t Z t D E Z t Z t D E (C7)
2
⟨y (t)|θ0 ⟩or = v02 dτ dτ ′ sin θ(τ ) sin θ(τ ′ ) + dτ dτ ′ η(τ )η(τ ′ )
0 0 0 0

To compute the position moments, it is essential to first evaluate the two-time auto-correlations in the governing
equations. Beginning with the two-time correlation of cos θ; for τ > τ ′ we get,
D E
C(τ, τ ′ ) ≡ cos θ(τ ) cos θ(τ ′ )
Z (C8)
= dθdθ′ cos θ cos θ′ P(θ, τ |θ′ , τ ′ )P(θ, τ ′ |0, 0)
 21

where the propagator P(θ, τ |θ′ , τ ′ ) satisfies the renewal Eq.(C3). Using Eq.(C3) in the above equation and performing
the integrals, we get, for τ > τ ′ ,

r2 2Dθ −(r+Dθ )τ −3Dθ τ ′ rDθ  −(r+Dθ )τ ′ 

C(τ, τ ′ ) = + e + e − e −(r+Dθ )τ
(r + Dθ )2 r + 4Dθ (r + Dθ )2
(C9)
Dθ2 (5r + 2Dθ ) ′
+ 2
e−(r+Dθ )(τ −τ )
(r + 4Dθ )(r + Dθ )

To calculate the second term containing ⟨η(τ )η(τ ′ )⟩ in RHS of Eq.(C7), we approach as
" #
Z t Z t Z t Z t
′ ′ ∆Γ
′
dτ dτ ⟨η(τ )η(τ )⟩ = 2kB T dτ dτ Γ̄ + ⟨cos 2θ(τ ′ )|θ0 ⟩or (C10)
0 0 0 0 2

To determine ⟨cos θ(τ ′ )|θ0 ⟩or under resetting conditions, we apply the renewal equation specifically for θ, which
simplifies to the following form:

Z t
−rt
Pr (θ, t|θ0 ) = e P0 (θ, t|θ0 ) + r dτ e−rτ P0 (θ, τ |θ0 ) (C11)
0

By multiplying the governing equation with cos 2θ and integrating, we obtain

Z t
′
−rt
⟨cos 2θ(τ )|θ0 ⟩or = e ⟨cos 2θ(τ )|θ0 ⟩0 + r dt′ e−rt ⟨cos 2θ(t′ )|θ0 ⟩0
0 (C12)
−(r+4Dθ )t r cos 2θ0  
= cos 2θ0 e + 1 − e−(r+4Dθ )t
r + 4Dθ

Using the above expression in Eq.(C10) we find,

Z t Z t Z t
h ∆Γ i
dτ dτ ′ ⟨η(τ )η(τ ′ )⟩ = 2kB T dτ ′ Γ̄ + ⟨cos 2θ(τ ′ )|θ0 ⟩or
0 0 2
"0 # (C13)
 r∆Γ cos 2θ0  2Dθ ∆Γ cos 2θ0 −(r+4Dθ )t
= 2kB T Γ̄ + t− (1 − e )
2(r + 4Dθ ) (r + 4Dθ )2

Repeating the same exercise for sin θ, we get

D E 2Dθ ′
h ′
i
sin θ(τ ) sin θ(τ ′ ) = e−(r+Dθ )(τ −τ ) 1 − e−(r+4Dθ )τ (C14)
r + 4Dθ

Similarly, the second moment of the y-component is calculated using the renewal framework and incorporating the
effects of resetting and anisotropy

" ! #
2 r∆D 4Dθ ∆D  −(r+4Dθ )t
 4v02 Dθ t 4v02 Dθ (2r + 5Dθ )
⟨y (t)|θ0 ⟩or = 2 D̄ − t− 2
1−e + −
r + 4Dθ (r + 4Dθ ) (r + 4Dθ )(r + Dθ ) (r + 4Dθ )2 (r + Dθ )2
!
4v02 e−(r+Dθ )t e−(r+4Dθ )t
+ −
3 (r + Dθ )2 (r + 4Dθ )2
(C15)