Steady state in ultrastrong coupling regime: expansion and first orders

Camille L Latune
ENSL, CNRS, Laboratoire de physique, F-69342 Lyon, France
Quantum Research Group, School of Chemistry and Physics,
University of KwaZulu-Natal, Durban, KwaZulu-Natal, 4001, South Africa
National Institute for Theoretical Physics (NITheP), KwaZulu-Natal, 4001, South Africa
(Dated: October 28, 2022)
Understanding better the dynamics and steady states of systems strongly coupled to thermal baths
is a great theoretical challenge with promising applications in several fields of quantum technologies.
Among several strategies to gain access to the steady state, one consists in obtaining approximate
expressions of the mean force Gibbs state, the reduced state of the global system-bath thermal state,
largely credited to be the steady state. Here, we present analytical expressions of corrective terms
to the ultrastrong coupling limit of the mean force Gibbs state, which has been recently derived.
We find that the first order term precisely coincides with the first order correction obtained from
a dynamical approach – master equation in the strong-decoherence regime. This strengthens the
identification of the reduced steady state with the mean force Gibbs state. Additionally, we also
compare our expressions with an other recent result obtained from a high temperature expansion
of the mean force Gibbs state. We observe numerically a good agreement for ultra strong coupling
as well as for high temperatures. This confirms the validity of all these results. In particular, we
show that, in term of coherences, all three results allow one to sketch the transition from ultrastrong
coupling to weak coupling.

The dynamics of quantum systems strongly coupled
to thermal baths have recently received a lot of inter-
est fuelled by hopes of understanding or even discover-
ing new phenomena in quantum transport [1–8], quan-
tum thermodynamics [9–15], quantum sensing [16–18],
as well as understanding better the underlying physics
of some essential biological functions [19–22]. In most of
these applications, to know the steady state of the sys-
tem strongly coupled to the bath is often essential and
sufficient. Such steady states greatly depart from usual
equilibrium steady states [23–26]. While the properties
of the most general open quantum evolutions have been
known for long time thanks to the seminal paper [27],
their precise time evolutions and steady states are still a
current challenge of the theory of open quantum system
[28].
To obtain some information about strongly coupled
steady states, several strategies have been developed, in-
cluding embedding techniques like reaction coordinate
[10, 23, 29, 30] and pseudo-mode [31–34], or numerical
techniques (Hierarchical Equation of Motion) [22, 35, 36].
One alternative strategy consists in focusing directly on
the steady state without going through the description
of the whole dynamics. In this perspective, the global
steady state of the system and bath is expected to be
the global system-bath thermal state at the bath tem-
perature [37–43]. The steady state of the system is then
given by tracing out the bath, which is sometimes re-
ferred to as the mean force Gibbs state [25, 43, 44]. Such
partial trace is usually very challenging, but can be done
at least approximately assuming for instance a weak cou-
pling [24, 25, 40, 45]. One other interesting regime, and
potentially containing more novelty, is the ultrastrong
coupling regime, when the strength of the coupling is
larger than the system’s energy scale. However, only few
papers considered such situations. In [25], Cresser and
Anders provide the explicit expression of the mean force
Gibbs state in the limit of infinite coupling. In particu-
lar, they show that it is close to the form of the steady
state obtained in [46, 47] using arguments from eisens-
election [48], although there are also some slight differ-
ences. Despite being an interesting result, it would be
welcome to have also information on how the transition
from the weak coupling limit to the ultrastrong coupling
limit happens, as well as on the steady state in interme-
diate regimes which are experimentally more accessible.
In this perspective, introducing a technique inspired
from the displaced oscillator picture [49, 50] for diago-
nalization of the quantum Rabi model as well as remi-
niscent of the polaron transformation [51, 52], we recover
the infinite coupling limit of [25] and go beyond by pro-
viding first order corrections. In the regime of high bath
temperature or low bath frequency, we derive a very sim-
ple approximate expression. We compare our results to
two very recent derivations. The first one [44] was ob-
tained from a master equation in the so-called strong-
decoherence regime, a generalization of the ultra-strong
coupling regime, and actually coincides with the first or-
der expansion derived here. This confirms that a system
interacting with a thermal bath does converge, at least up
to first order, to the mean force Gibbs state even in the
ultra-strong coupling regime. The second recent deriva-
tion [54] consists in a high temperature expansion of the
mean force Gibbs state, a generalization of a derivation
introduced in [53]. The comparison with this result is es-
sentially numeric, and we find an overall good agreement
in the expected regime of validity.
Additionally, we show that the provided corrective
terms allow to sketch the transition from the ultrastrong
coupling limit to the weak coupling limit.
 2

I. MEAN FORCE GIBBS STATE with T representing the “β-ordering operator”, acting on
inverse temperatures in the same way as the usual time
We consider the following total Hamiltonian ordering operator acts on exponential time integrals, and

HSB = HS + HB + λHI , (1) H̃S (u) := eu(HB +λAB+λ

2
QA2 )
HS e−u(HB +λAB+λ
2
QA2 )
.
(6)
where HB := k ωk a†k ak is the bath Hamiltonian com-
P
Now, by noticing that HB + λAB + λ2 QA2 =
posed of the bosonic creation and annihilation operators λg
P † † − ω k (b†k −bk )A
a†k and ak , λ is a dimensionless parameter representing k ωk Dk b k bk Dk , where D k := e k is a dis-
the strength of the system-bath coupling, HI := AB is placement operator by an “amount” λgk A/ωk acting on
the system-bath coupling term, and B := k gk (ak + a†k )
P the mode k, we can re-write HB + λAB + λ2 QA2 as a
is the usual coupling bath operator. As in the quantum “mixture of displaced baths”. To see that, we denote by
Brownian motion [28], there is an extra term λ2 QA2 al and |al ⟩ the eigenvalues (assumed to be non-degenerate
appearing during the derivation of the solution which for simplicity) and corresponding eigenvectors of the ob-
λgk
− (b† −b )a
corresponds to the renormalization of the system’s en- servable A, and by Dk,l := e ωk k k l the displace-
ergies due to the interaction with the bath. This is ment operator by the quantity λgk al /ωk acting only on
also taken into account explicitly in [25]. The quantity the mode k. Then, we have
Q, theR ∞“re-organization energy” [25, 55], is defined as
Q := 0 dωJ(ω)/ω, where J(ω) is the bath spectral den- ωk Dk b†k bk Dk†
X
HB + λAB + λ2 QA2 =
sity, J(ω) = k gk2 δ(ω − ωk ). Such extra term is usually
P
k
added initially, “by hand”, so that it cancels out later on
|al ⟩⟨al |Dk b†k bk Dk†
X X
during the calculation when the renormalization takes = ωk
k l
place. Following this procedure, we add the extra term
|al ⟩⟨al |Dk,l b†k bk Dk,l
†
X X
λ2 QA2 so that the total Hamiltonian is now = ωk
k l
HSB := HS + λ2 QA2 + λAB + HB . (2) X
= |al ⟩⟨al |HB,l , (7)
Note the difference of notation between the “natural” (1) l
and renormalized (2) Hamiltonians. As a matter of com-
pleteness, we also mention in section V of the Supplemen- where we defined the “displaced bath” HB,l :=
P † † 2 2
tary Material the derivation starting from the “natural” k ωk Dk,l bk bk Dk,l = HB + λal B + λ al Q, reminiscent
Hamiltonian (1), which leads to renormalized final ener- of the displaced oscillator picture [49, 50]. From there we
gies (or pseudo-energies). This shows in particular that obtain,
both derivations are equivalent, as expected, but that the
2
renormalization has to take place at some stage, either QA2 )
P
u l |al ⟩⟨al |HB,l
eu(HB +λAB+λ = eX
initially or finally.
As mentioned in the introduction, the global steady = |al ⟩⟨al |euHB,l , (8)
l
state of the system and bath is expected to be the global
thermal state (assuming [HS , A] ̸= 0) [37–43] so that
−1 −βHSB
ρth
SB := ZSB e , (3) X
H̃S (u) = |al ⟩⟨al |euHB,l HS |al′ ⟩⟨al′ |e−uHB,l′
where ZSB := TrSB e−βHSB is the partition function.


l,l′
Then, the reduced steady state of S is obtained by
= HSpop + H̃Scoh (u), (9)
−1 −βHSB
ρss
S := ZSB TrB [e ], (4)
where HSpop := coh
P
l hl |al ⟩⟨al | and H̃S (u) :=
which is sometimes called the mean force Gibbs state P uH −uH
l̸=l′ hl,l |al ⟩⟨al |e
′
′ ′ B,l
e B,l , defining hl :=
[25, 43, 44]. We are going to approximate this state in the
⟨al |HS |al ⟩, and hl,l′ := ⟨al |HS |al′ ⟩.
regime of ultrastrong coupling, when λ becomes larger Rβ
than the energy scale of HS . Taking out HSpop from the exponential e−T 0
duH̃S (u)
,
we obtain,
≈
Rβ pop Rβ
II. GENERAL DERIVATION e−T 0
duH̃S (u)
= e−βHS e−T 0
coh
duHS (u)
, (10)

We first split HSB in two, HS on one hand, and with

λ2 QA2 + λAB + HB on the other hand, and take it out
≈ pop pop
of the exponential using the usual identities [56]. It leads HScoh (u) := euHS H̃Scoh (u)e−uHS
to X
2
Rβ = hl,l′ euωl,l′ |al ⟩⟨al′ |euHB,l e−uHB,l′ , (11)
QA2 ) −T
e−βHSB = e−β(HB +λAB+λ e 0
duH̃S (u)
, (5) l̸=l′
 3

defining ωl,l′ := hl − hl′ . Combining the identities (5), This leads to the following expression for the reduced
(8) and (10), we arrive at steady state of S,
≈
pop
−T 0β duHScoh
X R
−1 −βHB,l −βHS
ρth (u)
 ≈

= Z |a ⟩⟨a |e e e −1
X Rβ coh
SB SB l l
ρss
S = ZSB e−βhl |al ⟩⟨al |TrB e−βHB,l e−T 0
duHS (u)
.
l
Rβ ≈ l
X coh
−1
= ZSB e−βhl |al ⟩⟨al |e−βHB,l e−T 0
duHS (u)
. (13)
l
(12) The toughest part is, as expected,

 ≈
 ∞ Z β Z u1 Z un−1 ≈ ≈ ≈
Rβ coh X h i
TrB e−βHB,l e−T 0
duHS (u)
= (−1)n du1 du2 ... dun TrB e−βHB,l HScoh (u1 )HScoh (u2 )...HScoh (un ) .
n=0 0 0 0
(14)

In the following, when we mention “first order term” or ZB is the partition function of the uncoupled bath. Note
“higher order terms”, we refer to the terms appearing in that since ρss,∞
S is a normalized state, we also have the
identity ZSss = l e−βhl . The above expression (15) is
P
the above expansion (13). So far, we made no approx-
imation, making (13) an exact expression. Since this exactly equal to the one derived in [25], and recently re-
problem is not exactly solvable, we have to make some obtained in [44].
approximations in order to reach an explicit form. Before
that, let us see how one can recover the infinite coupling
limit. B. First approximation

≈
As seen in the previous section, the term HScoh (u)
A. Recovering the infinite coupling limit contains overlaps of displaced baths, soi that terms
≈ ≈ ≈
h
TrB e−βHB,l HScoh (u1 )HScoh (u2 )...HScoh (un ) of increasing
From expression (13), it might not be obvious how one
recovers the infinite coupling limit. It actually comes order contain overlaps of increasing order and are there-
≈ fore significantly smaller than terms of lower orders (for
from the terms HScoh (u) which contain overlaps between
large coupling strength). With this assumption, we are
displaced baths, euHB,l e−uHB,l′ . The strength of the
going to retain only the first and second term,
coupling between S and each mode of the bath is rep- (
resented by gk . Thus, when the strength of the cou- −1
X
−βhl
ss
|al ⟩⟨al | TrB e−βHB,l
 
pling goes to infinity, the displaced baths HB,l and HB,l′ ρS ≃ ZSB e
tend to be displaced infinitely far apart from each other l
" #)
(forl ̸= l′ ). Consequently, the overlap and the expecta- β
Z
≈

tion value TrB [e−βHB,l euHB,l′ e−uHB,l′′ ] tends to zero for −TrB e−βHB,l duHScoh (u) . (16)
0
increasing coupling strength.
Extending this reasoning to higher The first term gives TrB e−βHB,l = ZB , as already seen
 
orderh terms, we see that i all terms in the previous section. The second term gives,
≈ ≈ ≈
−βHB,l coh coh coh
TrB e HS (u1 )HS (u2 )...HS (un ) contain " Z β #
≈ X
−βHB,l coh
multiple overlaps of different displaced baths, and there- TrB e duHS (u) = hl′ ,l′′ |al′ ⟩⟨al′′ |
fore tends to zero as the coupling strength increases. 0 l′ ̸=l′′
In section III of the Supplemental Material, we show Z β
dueuωl′ ,l′′ TrB e−βHB,l euHB,l′ e−uHB,l′′ .
 
numerically (for a two-level systems) that the higher × (17)
0
order terms tend to zero as the coupling strength
increases. Thus, in the infinite coupling limit, only the Since, when injected in (16), this expression will be mul-
first term of the sum in (13) is different from zero. It tiplied by |al ⟩⟨al | on the left-hand side, one only needs
leads to to compute
X Z β
ρss,∞ −1
e−βhl |al ⟩⟨al |TrB e−βHB,l
 
= ZSB
X
dueuωl,l′ TrB e−βHB,l euHB,l e−uHB,l′ ,
 
S hl,l′ |al ⟩⟨al′ |
l 0
X l′ ̸=l
= (ZSss )−1 e−βhl |al ⟩⟨al |, (15) (18)
l
′
where the index l and l were made equal. After some
ZSB
and TrB e−βHB,l = TrB e−βHB :=
   
where ZSss := ZB manipulations, we can obtain the following expression for
 4

TrB e−βHB,l euHB,l e−uHB,l′ (details provided in section

 
and
I of the Supplementary Material), Z β 2 2
dueuωl,l′ e−λ al′ ,l u(1− β )Q
u
fl,l′ (β) ≃
TrB e−βHB,l euHB,l e−uHB,l′
 
0
s ( " s #
−uω
 
−λ2 a2l′ ,l
R∞ J(ω)
(euω −1) 1− 1−e−ωβ 1 β 1 β 2 2
= ZB e 0 ω2 1−e , (19) = DF (λ al′ ,l Q − ωl,l′ )
λ|al′ ,l | Q 2λ|al′ ,l | Q
where al′ ,l := al′ − al . We finally obtain, up to second " s #)
1 β
order, +eβωl,l′ DF (λ2 a2l′ ,l Q + ωl,l′ ) ,
2λ|al′ ,l | Q
X X
ρss
S = pss
l |al ⟩⟨al | − pss
l hl,l′ fl,l′ (β)|al ⟩⟨al′ |, (24)
l l,l′ ;l̸=l′ 2 Rx 2

(20) where DF(x) := e−x 0 du eu is sometimes referred to

as the Dawson function.
−βhl
with pss
l := e /ZSss and
The above expression can be further simplified as fol-
Z β R∞ J(ω)lows. The strong coupling regime can be characterized
 −ωu

−λ2 a2l′ ,l 0 dω ω2 (euω −1) 1− 1−e−ωβ
fl,l′ (β) = dueuωl,l′ e . by a re-organization energy λ2 Q comparable to, or larger
1−e

0 than, the energy scale of the system [25, 26]. Thus,

(21) in the strong coupling regime one can expect to have
λ2 Q ≫ maxl |hl |, implying
Note that we have the following identity (see section II s s
of the Supplementary Material) pss ss
l fl,l′ (β) = pl′ fl′ ,l (β), 1 β 2 2 1 β 2 2
implying ⟨al |ρss ∗ ss
S |al′ ⟩ = ⟨al′ |ρS |al ⟩, as it should be.
(λ al′ ,l Q ± ωl,l′ ) ∼ λ al′ ,l Q
2λ|al′ ,l | Q 2λ|al′ ,l | Q
Additionally, f+,− (β) tends to zero when λ goes to in- p
finity, so that we recover the infinite coupling limit (15). ∼ βλ2 Q. (25)
One can also see that the first order corrections affect
Finally, since DF (x) = 2x 1
+ O(x−3 ) for x ≫ 1 (the ap-
only the coherences (in the eigenbasis of A). These ob-
proximation is actaully very good for x ≥ 3), we can find
servations coincide with the ones in [44]. Finally, the
the following simple approximate expression for fl,l′ (β)
expression obtained starting from the natural Hamilto-
asuming λ2 Q ≫ β −1 ,
nian (1) instead of the renormalized one (2) are the same
as (20) and (21) but substituting the pseudo-energies hl 1 + eωl,l′ β ωl,l′ (1 − eωl,l′ β )
by the renormalized ones hl − λ2 a2l Q (see section V of fl,l′ (β) = 2 2 + + O[(λ2 Qβ)−3 ].
λ al′ ,l Q λ4 a4l′ ,l Q2
the Supplementary Material).
(26)

C. Approximate expression of fl,l′ (β) III. EXAMPLE: SPIN-BOSON MODEL

Depending on the bath spectral density, it might not As illustration of our results, we consider the versatile
be possible to obtain an exact analytical expression of and famous spin-boson model [57] characterised by the
fl,l′ (β), so that some approximations would have to be following total Hamiltonian
made. However, assuming the bath temperature is high,
or equivalently that the dominant frequencies in the bath ϵ ∆
HSB = σz + σx + λσz B + HB , (27)
are low, we can obtain an approximate expression of 2 2
fl,l′ (β) witout even specifying the form of the spectral where σx and σz are the Pauli matrices. The spin Hamil-
density. More precisely, we assume that J(ω) vanishes tonian HS can be re-written as
−1 −1
for ω ≥ ωc , where  ωc ≤ β .Then, for ω ∈ [0; β ], the
1−e−uω ϵ ∆ ωS
factor (euω − 1) 1 −


1−e−ωβ
can be approximated by HS = σz + σx = |e⟩⟨e| − |g⟩⟨g| , (28)
2 2 2
√
2 2
euω − 1 1 − e−uω with ωS := ϵ + ∆ ,
   
u u
1− = 1− + O(ωβ), (22)
ω2 1 − e−ωβ ω β (ωS + ϵ)|+⟩ + ∆|−⟩
|e⟩ := p ,
2ωS (ωS + ϵ)
which is actually a very good approximation as soon as
ωβ ≤ 1. With that we obtain −∆|+⟩ + (ωS + ϵ)|−⟩
|g⟩ := p , (29)
 −ωu
 2ωS (ωS + ϵ)
R∞ J(ω)
−λ2 a2l′ ,l dω (euω −1) 1− 1−e−ωβ 2 2
(1− uβ )Q ,
e 0 ω2 1−e ≃ e−λ al′ ,l u
and |±⟩ denotes the eigenstates of σz . Since coupling
(23) observable A is equal to σz , we have al=± = ±1, al,l′ =
 5

a+,− = −a−,+ = 2, hl=± = ±ϵ/2, hl,l′ = h+,− = h−,+ = a high temperature expansion of the mean force Gibbs
∆/2, and ωl,l′ = hl − hl′ = ω+− = −ω−+ = ϵ. Addition- state was recently derived in [54]. In this section, we
ally, since σz2 = I, the renormalized Hamiltonian is equal compare our result with the aforementioned ones. We
to start by briefly introducing them.
ϵ ∆
HSB = σz + σx + λσz B + HB + λ2 Q, (30) • In [44], the author derives the first order correction
2 2
to the stationary state in the ultra strong coupling
which simply corresponds to redefining the origin of the limit, which actually coincides with the general ex-
spin energy. This can also be verified from the renormal- pressions (20) and (21) (see details in section VI of
ized pseudo-energies defined in section V of the Supple- the Supplemental Material). As a technical note,
mentary Material, which are given by hl=± := hl=± − the results in [44] are actually valid when the bath
a2l λ2 Q = ±ϵ/2 − λ2 Q. Thus, for the spin-boson model, spectral density satisfies limω→+∞ J(ω) ω 2 > 0, and
the energy renormalization induced by the interaction the equivalence of the first order correction in [44]
with the bath has no impact on the reduced steady state. with (20) and (21) is guaranteed only within this
Applying expression (20), we obtain condition on the bath spectral density (satisfied by
1 h usual spectral densities).
ρss
S = −βϵ/2 e−βϵ/2 |+⟩⟨+| + eβϵ/2 |−⟩⟨−|
e + eβϵ/2 As a sanity check for the numerical simulations be-
∆ ∆ i low, we compare, for the spin-boson model, our
− e−βϵ/2 f+,− (β)|+⟩⟨−| − eβϵ/2 f−,+ (β)|−⟩⟨+| ,
2 2 approximate expression (33) of f+,− (β) with the
(31) high temperature approximation in [44] based on
an over-damped spectral density, also sometimes
with called Lorentz-Drude spectral density,
Z β R∞ J(ω)
 −ωu

ϵu −4λ
2
dω ω2
(euω −1) 1− 1−e ωβ
f+,− (β) = due e 0 1−e , 2Q ωc ω
0 J(ω) = , (35)
β
π ωc2 + ω 2
1−e−ωu
Z 2 ∞
R J(ω) uω
 
−ϵu −4λ 0 dω ω2 (e −1) 1− 1−eωβ
f−,+ (β) = due e .
0 where Q is precisely the re-organization energy as-
(32) sociated with the over-damped spectral density,
and ωc represents the cutoff frequency. Note that
the dimensionless factor λ is not explicitly present
A. High temperature approximation in [44], but one can make it appearing by multi-
plying the bath spectral density by λ2 . Then, the
Without specifying explicitly the bath spectral density, reorganisation energy becomes λ2 Q. Additionally,
if we consider that J(ω) vanishes for frequencies smaller the steady state populations obtained in [44] cor-
than β −1 , then the approximation (24) of section II C respond to the one obtained for infinite coupling
applies, and we have, limit [25], as we also derived in (15) and (20).
s ( " s #
f+,− (β) =
1 β
DF
1 β 2
(4λ Q − ϵ) • The derivation in [54], a generalization to arbitrary
2λ Q 4λ Q system-bath coupling of [53], consists in expressing
" s #) the mean force Gibbs state through approximating
ϵβ 1 β 2 the Hamiltonian of mean force. In other words,
+e DF (4λ Q + ϵ) . (33) −1 −βHSB
4λ Q ρss
S = ZSB TrB [e ] is expressed in the form
−1 −βHM F
ZM F e , and an approximated expression of
Assuming furthermore λ2 Q ≫ ϵ and λ2 Qβ ≫ 1, charac- HM F , the Hamiltonian of mean force, is provided
teristics of the ultrastrong coupling regime, we can use to second order in β in the following form[54]
expression (26), leading to
X 2
1 + eϵβ ϵ(1 − eϵβ ) HM F = HSdiag − ΛA2 + Jnm e−βΛ(an −am ) /6
|n⟩⟨m|
f+,− (β) = 2
+ + O[(λ2 Qβ)−3 ]. (34)
4λ Q 16λ4 Q2 n̸=m

+O(β 4 ), (36)
IV. COMPARISON WITH PREVIOUS RESULTS
where Λ P is the re-organization energy,
In a recent paper [44], the author obtains corrections to A = n an |n⟩⟨n| is the system oper-
the ultrastrong coupling limit by deriving and solving a ator coupling with the bath, as previ-
ously, HSdiag
P
strong-decoherence regime master equation (generaliza- = n ⟨n|HS |n⟩|n⟩⟨n|, and
tion of the ultra-strong coupling regime). Additionally, Jnm := ⟨n|HS |m⟩. Re-expressed using the
 6

notations used throughout our paper, we have css λ2 Q

0.1 1 10 100 ϵ
- 0.01
X
HM F = hl |al ⟩⟨al | − λ2 QA2
l
X 2
Q(al −a′l )2 /6 - 0.02
+ hl,l′ e−βλ |al ⟩⟨al′ |. (37)
l̸=l′
- 0.03

The term −λ2 QA2 corresponds to the renormal- (a)- 0.04

ization of the system’s energies, which has to be css eg
performed initially as we did above, or finally, as 0.03
shown in the Supplemental Material. The parti-
tion function is given by ZM F = TrB [e−βHM F ]. 0.02
Applying the above expression to the spin-boson
model, we obtain 0.01

λ2 Q
ϵ ∆ 2
0.1 1 10 100
HM F = σz − λ2 Q + e−2βλ Q/3 σx (b) ϵ
2 2
ϵ ∆ 2
FIG. 1. Plots in the semi-log scale of the steady state co-
= σz + e−2βλ Q/3 σx , (38)
2 2 herences (a) css := ⟨+|ρssS |−⟩ in the eigenbasis of A, and (b)
css ss
e,g := ⟨e|ρS |g⟩ in the eigenbasis of HS , in function of the
coupling strength λ2 Q in unit of ϵ, for ∆ = 1.5ϵ, ωc = 2ϵ and
where we dropped the renormalization term in the ϵβ = 0.1. The orange thick solid curve corresponds to (31)
second line since in this situation it only corre- using the high temperature approximation (33) of f+,− (β),
sponds to redefining the origin of the energies. while the gray thin dashed line corresponds to the further sim-
Thus, the mean force Gibbs state is simply given plified expression (34) of f+,− (β). The purple thick dashed
by line is the steady state coherences given by the high temper-
ature expression derived in [44]. Finally, the black thin line
is the steady state coherences given by (40).
h ′ ′
i
−1 −ωS β/2 ′
ρM F = Z M F e |e ⟩⟨e′ | + eωS β/2 |g ′ ⟩⟨g ′ | , (39)
In Fig. 1, we have plotted in semi-log scale the steady
state coherences (a) css := ⟨+|ρss S |−⟩ in the eigenbasis of
√ 2 A, and (b) css := ⟨e|ρss
|g⟩ in the eigenbasis of HS , in
with ωS′ := ϵ2 + ∆′2 , ∆′ := e−2βλ Q/3 ∆, |e′ ⟩ := e,g S
′ ′
(ωS +ϵ)|+⟩+∆ |−⟩ ′
−∆ |+⟩+(ωS′ function of the coupling strength λ2 Q in unit of ϵ, for
√ ′ ′ , |g ′ ⟩ := √ ′ ′ +ϵ)|−⟩ , and
2ωS (ωS +ϵ) 2ωS (ωS +ϵ) ∆ = 1.5ϵ, ωc = 2ϵ and ϵβ = 0.1. The orange thick solid
′ ′
ZM F = e−ωS β/2 +e ωS β/2
. This expression has the curve corresponds to (31) using the high temperature ap-
merit of tending to the right limit when λ → 0, proximation (33) of f+,− (β), while the gray thin dashed
namely ρM F → Z −1 e−βHS , the usual thermal equi- line corresponds to the further simplified expression (34)
librium state. This is not the case of the expression of f+,− (β). The purple thick dashed line is the steady
(31), which is expected since it is meant to be valid state coherences given by the high temperature expres-
in the opposite regime, when λ2 Qβ ≫ 1. In the sion derived in [44]. Finally, the black thin line is the
eigenbasis of σz , the expression of ρM F becomes steady state coherences given by Eq. (40).
As expected, on can see an excellent agreement be-
  tween (33) (orange curve) and the expression given in
1 ϵ ′ [44] (purple dashed line). The very slight discrepancy
ρM F = 1 − ′ tanh ωS β/2 |+⟩⟨+|
2 ωS appearing at intermediate coupling strength comes from
1

ϵ
 a slight difference in the derivation of high temperature
′
+ 1 + ′ tanh ωS β/2 |−⟩⟨−| approximation between (22)-(23) and eqs. (56)-(57) of
2 ωS [44] (high temperature approximation before time inte-
∆′  
gration of the bath correlation function). However, there
− ′ tanh ωS′ β/2 |+⟩⟨−| + |−⟩⟨+| . (40)
2ωS is a significant discrepancy with (40) (black line) at in-
termediary and strong coupling strength. Beyond that,
it appears from (a) that for increasing coupling strength,
A brief analytical comparison between (40) and the steady state tends to a diagonal state in the eigenba-
(31) is provided in the Supplemental Material. sis of A, as expected from the ultrastrong coupling limit
 7

(15) and [25, 46, 47]. Conversely, for decreasing coupling pss+
strength, one can see in (b) the emergence of a progres-
0.4
sive transition to a diagonal state in the eigenbasis of HS ,
as expected from the weak coupling limit.

0.35

0.3 λ2 Q
css
ϵβ (a) 0 10 20 30 40 ϵ
0.1 0.5 1 5 10 ss
p e

- 0.01 0.4

- 0.02 0.3
(a)
css eg
0.2 λ2 Q
0.4
(b) 0 5 10 15 20 ϵ
0.3
FIG. 3. Plots of the steady state excited population (a) pss+ :=

0.2
0.1
ϵβ
(b) 0.1 0.5 1 5 10
FIG. 2. Plots in the semi-log scale of the steady state co-
herences (a) css := ⟨+|ρss S |−⟩ in the eigenbasis of A, and (b)
css ss
e,g := ⟨e|ρS |g⟩ in the eigenbasis of HS , in function of the in-
verse temperature ϵβ, for ∆ = 1.5ϵ, ωc = 1ϵ and λ2 Q = 10ϵ.
The colour convention is the same as in Fig. 1, namely, the
orange thick solid curve corresponds to (31) using the high
temperature approximation (33) of f+,− (β), while the gray
thin dashed line (almost indistinguishable from the orange
curve) corresponds to the further simplified expression (34)
of f+,− (β). The purple thick dashed line is the steady state
coherences given by the high temperature expression derived
in [44]. Finally, the black thin line is the steady state coher-
ences given by (40).
In Fig. 2, we plotted in the semi-log scale of the steady
state coherences (a) css := ⟨+|ρssS |−⟩ in the eigenbasis of
A, and (b) css ss
e,g := ⟨e|ρS |g⟩ in the eigenbasis of HS , in
function of the inverse temperature ϵβ, for ∆ = 1.5ϵ,
ωc = 1ϵ and λ2 Q = 10ϵ. The colour convention is the
same as in the previous figure Fig. 1. Again, as expected,
we observe a very good agreement between (33) (orange
curve) and the expression given in [44] (purple dashed
line). We also observe a significant discrepancy with (40)
(black line) out side the high temperature regime.
⟨+|ρss ss ss
S |+⟩ in the eigenbasis of A, and (b) pe := ⟨e|ρS |e⟩ in the
eigenbasis of HS , in function of the coupling strength λ2 Q in
unit of ϵ, for ∆ = 1.5ϵ and ϵβ = 0.5. The orange thick
solid curve corresponds to the steady state excited population
obtained in (33), [44], and [25, 46, 47] (all coinciding). The
black thin curve corresponds to the excited population given
by (40). The blue dotted curve corresponds to the thermal
−ωS β
population in the vanishing coupling limit, pthe = e /(1 +
−ωS β
e ).
In the two following figures, Fig. 3 and Fig. 4, we
plotted the the steady state excited population (a) pss+ :=
⟨+|ρss ss ss
S |+⟩ in the eigenbasis of A, and (b) pe := ⟨e|ρS |e⟩
in the eigenbasis of HS . As for the coherence, Fig. 3 is
in function of the coupling strength λ2 Q in unit of ϵ, for
∆ = 1.5ϵ and ϵβ = 0.5, while in Fig. 4, the plots are
in function of the inverse temperature ϵβ in a semi-log
scale, for ∆ = 1.5ϵ and λ2 Q = 5ϵ. For both figures, the
orange thick solid curve corresponds to the steady state
excited population obtained in (33), [44], and [25, 46, 47]
(all coinciding), while the black thin curve corresponds to
the excited population given by (40). Finally, the blue
dotted curve corresponds to the thermal population in
the vanishing coupling limit, pth −ωS β
/(1 + e−ωS β ).
e =e
We can draw conclusions similar to Figs. 1 and 2,
namely that both predictions coincide very well at high
temperature as well as for ultra strong coupling. How-
ever, some discrepancies emerge for intermediate and
large coupling strength as well as for large β.
 8

pss + uations in which the steady state of a system interacting

0.5 with a thermal bath does converge to the mean force
0.4 Gibbs state. This convergence was recently shown for
weak coupling up to the first order [24, 40, 45], for the
0.3 ultra strong coupling regime up to the zeroth order [25],
and is now confirmed in the ultra-strong coupling regime
0.2 up to the first order, thanks to the complementary results
0.1 in [44].
In the high temperature regime, a simple and explicit
ϵβ general expression is derived (24) for the first order cor-
(a) 0.1 0.5 1 5 10
rections. This expression is numerically compared for
psse the spin-boson model with the result of [54] obtained
by high temperature expansion of the mean force Gibbs
0.4 state. For sanity check, a comparison of (24) with the
high temperature approximation of [44] is also made, and
0.3 the very good agreement in all regimes of parameters is
confirmed. Very good agreement is also observed at ul-
0.2 tra strong coupling as well as high temperatures between
0.1 our result and the one from [54]. This strengthens the
validity of all three approaches. Thanks to them, we are
ϵβ able to draft the transition from the ultra-strong coupling
(b) 0.1 0.5 1 5 10
regime to the weak coupling regime, Fig. 1 b.
However, going further, significant discrepancies
FIG. 4. Plots of the steady state excited population (a) pss+ :=
⟨+|ρss ss ss emerge between our result and [54] at moderate coupling
S |+⟩ in the eigenbasis of A, and (b) pe := ⟨e|ρS |e⟩ in the
eigenbasis of HS , in function of the inverse temperature ϵβ in strength and moderate temperatures. For those values
a semi-log scale, for ∆ = 1.5ϵ and λ2 Q = 10ϵ. The color of parameters, further analysis with higher orders are
convention is the same as in Fig. 3, namely, the orange thick needed in order to confirm which result is the more ac-
solid curve corresponds to the steady state excited population curate, as well as to reconstruct accurately the full tran-
obtained in (33), [44], and [25, 46, 47] (all coinciding). The sition weak-to-ultra-strong coupling and provide a good
black thin curve corresponds to the excited population given approximation of the steady state in all regimes. This
by (40). The blue dotted curve corresponds to the thermal does not seem out-of-reach according to the preliminary
−ωS β
population in the vanishing coupling limit, pth e = e /(1 + results on higher order terms obtained in the Supple-
−ωS β
e ).
mental Material. Additionally, benchmarking the results
with other techniques like reaction coordinate or pseudo-
modes would be interesting and instructive.
V. CONCLUSION

We obtain a perturbative expansion (13) and (14) of ACKNOWLEDGEMENTS

the mean force Gibbs state in the ultra strong coupling
regime, using a technique inspired from the displaced os-
cillator picture [49, 50] and polaron transformation. The
zero order term confirms the result of [25, 44]. Analyti-
cal expression of the first order term is derived, (20) and
(21), and is found to coincide with the first order correc-
tion derived in [44] (within the mild validity condition
of [44], see section VI of the Supplemental Material for
more detail). This increases the range of the known sit-
I would like to thank Anton Trushechkin for very
interesting and helpful discussions on this topic. I also
would like to thank the anonymous referees for their
constructive comments, participating in improving the
quality of the paper. This work was supported by
funding from the National Institute for Theoretical
Physics (NITheP) of the Republic of South Africa.

[1] A. W. Chin, S. F. Huelga ans M. B. Plenio, Coher-
ence and decoherence in biological systems: principles of
noise-assisted transport and the origin of long-lived co-
herences, Phil. Trans. R. Soc. A (2012) 370, 3638–3657.
doi:10.1098/rsta.2011.0224
[2] P. Ribeiro and V. R. Vieira, Non-Markovian effects
in electronic and spin transport, Phys. Rev. B 92,
100302(R) (2015). doi:10.1103/PhysRevB.92.100302
[3] P. Strasberg, G. Schaller, T. L. Schmidt, and M. Es-
posito, Fermionic reaction coordinates and their appli-
cation to an autonomous Maxwell demon in the strong-
coupling regime, Phys. Rev. B 97, 205405 (2018).
doi:10.1103/PhysRevB.97.205405
[4] L. A. Correa, B. Xu, B. Morris, and G. Adesso, Pushing
the limits of the reaction-coordinate mapping, J. Chem.
Phys. 151, 094107 (2019). doi:10.1063/1.5114690

[5] S. V. Moreira, B. Marques, R. R. Paiva, L. S.
Cruz, D. O. Soares-Pinto, and F. L. Semião, En-
hancing quantum transport efficiency by tuning non-
Markovian dephasing, Phys. Rev. A 101, 012123 (2020).
doi:10.1103/PhysRevA.101.012123
[6] E. Zerah-Harush and Y. Dubi, Effects of disor-
der and interactions in environment assisted quan-
tum transport, Phys. Rev. Research 2, 023294 (2020).
doi:10.1103/PhysRevResearch.2.023294
[7] D. Dwiputra and F. P. Zen, Environment-assisted quan-
tum transport and mobility edges, Phys. Rev. A 104,
022205 (2021). doi:10.1103/PhysRevA.104.022205
[8] Nicholas Anto-Sztrikacs and Dvira Segal, Strong coupling
effects in quantum thermal transport with the reaction
coordinate method, arXiv:2103.05670.
[9] D. Gelbwaser-Klimovsky and A. Aspuru-Guzik, Strongly
Coupled Quantum Heat Machines, J. Phys. Chem. Lett.
6, 3477-3482 (2015). doi:10.1021/acs.jpclett.5b01404
[10] P. Strasberg, G. Schaller, N. Lambert, and T. Bran-
des, Nonequilibrium thermodynamics in the strong cou-
pling and non-Markovian regime based on a reaction
coordinate mapping, New J. Phys. 18, 073007 (2016).
doi:10.1088/1367-2630/18/7/073007
[11] D. Newman, F. Mintert, and A. Nazir, Perfor-
mance of a quantum heat engine at strong reser-
voir coupling, Phys. Rev. E 95, 032139 (2017).
doi:10.1103/PhysRevE.95.032139
[12] M. Perarnau-Llobet, H. Wilming, A. Riera, R. Gallego,
and J. Eisert, Strong Coupling Corrections in Quantum
Thermodynamics, Phys. Rev. Lett. 120, 120602 (2018).
doi:10.1103/PhysRevLett.120.120602
[13] M.Wertnik, A. Chin, F. Nori, and N. Lambert, Op-
timizing co-operative multi-environment dynamics in a
dark-state-enhanced photosynthetic heat engine, J. Chem.
Phys. 149, 084112 (2018). doi:10.1063/1.5040898
[14] D. Newman, F. Mintert, and A. Nazir, Quantum limit
to nonequilibrium heat-engine performance imposed by
strong system-reservoir coupling, Phys. Rev. E 101,
052129 (2020). doi:10.1103/PhysRevE.101.052129
[15] M Wiedmann, J T Stockburger, and J Ankerhold,
Non-Markovian dynamics of a quantum heat engine:
out-of-equilibrium operation and thermal coupling con-
trol, New J. Phys. 22, 033007 (2020). doi:10.1088/1367-
2630/ab725a
[16] L. A. Correa, M. Perarnau-Llobet, K. V. Hovhan-
nisyan, S. Hernández-Santana, M. Mehboudi, and A.
Sanpera, Enhancement of low-temperature thermometry
by strong coupling, Phys. Rev. A 96, 062103 (2017).
doi:10.1103/PhysRevA.96.062103
[17] M. Mehboudi, A. Lampo, C. Charalambous, L. A. Cor-
rea, M. A. Garcı́a-March and M. Lewenstein, Using Po-
larons for sub-nK Quantum Nondemolition Thermome-
try in a Bose-Einstein Condensate, Phys. Rev. Lett. 122,
030403 (2019). doi:10.1103/PhysRevLett.122.030403
[18] M Salado-Mejı́a, R. Román-Ancheyta, F. Soto-Eguibar,
and H. M. Moya-Cessa, Spectroscopy and criti-
cal quantum thermometry in the ultrastrong coupling
regime, Quantum Sci. Technol. 6, 025010 (2021).
doi:10.1088/2058-9565/abdca5
[19] A. Kolli, E. J. O’Reilly, G. D. Scholes, and A. Olaya-
Castro, The fundamental role of quantized vibrations in
coherent light harvesting by cryptophyte algae, J. Chem.
Phys. 137, 174109 (2012). doi:10.1063/1.4764100
9
[20] N. Lambert, Y.-N. Chen, Y.-C. Cheng, C.-M. Li, G.-Y.
Chen and Franco Nori, Quantum biology, Nat. Physics 9,
10-18 (2013). doi:10.1038/nphys2474
[21] G. Scholes, G. Fleming, L. Chen, et al., Using coherence
to enhance function in chemical and biophysical systems,
Nature 543, 647–656 (2017). doi:10.1038/nature21425
[22] N. Lambert, T. Raheja, S. Ahmed, A. Pitchford, F.
Nori, BoFiN-HEOM: A bosonic and fermionic numerical
hierarchical-equations-of-motion library with applications
in light-harvesting, quantum control, and single-molecule
electronics, arXiv:2010.10806.
[23] J. Iles-Smith, N. Lambert, and A. Nazir, En-
vironmental dynamics, correlations, and the emer-
gence of noncanonical equilibrium states in open quan-
tum systems, Phys. Rev. A 90, 032114 (2014).
doi:10.1103/PhysRevA.90.032114
[24] A. Purkayastha, G. Guarnieri, M. T. Mitchison, R. Filip,
and J. Goold, Tunable phonon-induced steady-state co-
herence in a double-quantum-dot charge qubit, npj Quant.
Inf. 6, 27 (2020). doi:10.1038/s41534-020-0256-6
[25] J. Cresser and J. Anders, Weak and ultrastrong cou-
pling limits of the quantum mean force Gibbs state,
Phys. Rev. Lett. 127, 250601 (2021). doi: 10.1103/Phys-
RevLett.127.250601
[26] C. L. Latune, Steady state in strong system-bath cou-
pling regime: Reaction coordinate versus perturba-
tive expansion, Phys. Rev. E 105, 024126 (2022).
doi:10.1103/PhysRevE.105.024126
[27] E. C. G. Sudarshan, P. M. Mathews, and
Jayaseetha Rau, Stochastic Dynamics of Quantum-
Mechanical Systems, Phys. Rev. 121, 920 (1961).
doi:10.1103/PhysRev.121.920
[28] H. Breuer and F. Petruccione, Theory of Open Quantum
Systems, (Oxford, Oxford, 2002).
[29] A. Garg, J. N. Onuchic, and V. Ambegaokar, Effect of
friction on electron transfer in biomolecules, J. Chem.
Phys. 83, 4491 (1985). doi:10.1063/1.449017
[30] Jake Iles-Smith, Arend G. Dijkstra, Neill Lambert, and
Ahsan Nazir, Energy transfer in structured and unstruc-
tured environments: Master equations beyond the Born-
Markov approximations, J. Chem. Phys. 144, 044110
(2016). doi:10.1063/1.4940218
[31] B. M. Garraway, Nonperturbative decay of an atomic
system in a cavity, Phys. Rev. A. 55, 2290 (1997).
doi:10.1103/PhysRevA.55.2290
[32] G. Pleasance and B. M. Garraway, Application of quan-
tum Darwinism to a structured environment, Phys. Rev.
A 96, 062105 (2017). doi:10.1103/PhysRevA.96.062105
[33] A.E. Teretenkov, Pseudomode Approach and Vibronic
Non-Markovian Phenomena in Light-Harvesting Com-
plexes, Proc. Steklov Inst. Math. 306, 242–256 (2019).
doi:10.1134/S0081543819050201
[34] Graeme Pleasance, Barry M. Garraway, and Francesco
Petruccione, Generalized theory of pseudomodes
for exact descriptions of non-Markovian quan-
tum processes, Phys. Rev. R. 2, 043058 (2020).
doi:10.1103/PhysRevResearch.2.043058
[35] Yoshitaka Tanimura, Numerically “exact” approach to
open quantum dynamics: The hierarchical equations of
motion (HEOM), J. Chem. Phys. 153, 020901 (2020).
doi:10.1063/5.0011599
[36] N. Lambert, S. Ahmed, M. Cirio, F. Nori, Mod-
elling the ultra-strongly coupled spin-boson model with
unphysical modes, Nat Commun 10, 3721 (2019).

doi:10.1038/s41467-019-11656-1
[37] V. Bach, J. Fröhlich, and I. M. Sigal, Return
to Equilibrium, J. Math. Phys. 41, 3985 (2000).
doi:10.1063/1.533334
[38] J. Frohlich, Marco Merkli, Another Return of “Return
to Equilibrium”, Commun. Math. Phys. 251, 235–262
(2004). doi:10.1007/s00220-004-1176-6
[39] M. Merkli, I. M. Sigal, and G. P. Berman, Decoherence
and Thermalization, Phys. Rev. Lett. 98, 130401 (2007).
doi:10.1103/PhysRevLett.98.130401
[40] T. Mori and S. Miyashita, Dynamics of the Density Ma-
trix in Contact with a Thermal Bath and the Quan-
tum Master Equation, Journal of the Physical Society
of Japan 7, 124005 (2008). doi:10.1143/JPSJ.77.124005
[41] M. Könenberg and M. Merkli, On the irreversible dynam-
ics emerging from quantum resonances, J. Math. Phys.
57, 033302 (2016). doi:10.1063/1.4944614
[42] Marco Merkli, Quantum Markovian master equations:
Resonance theory shows validity for all time scales, Ann.
Phys. 412, 16799 (2020). doi:10.1016/j.aop.2019.167996
[43] A. S. Trushechkin, M. Merkli, J. D. Cresser, and J.
Anders, Open quantum system dynamics and the mean
force Gibbs state, AVS Quantum Sci. 4, 012301 (2022).
doi:10.1116/5.0073853
[44] A. Trushechkin, Quantum master equations and steady
states for the ultrastrong-coupling limit and the strong-
decoherence limit, arxiv:2109.01888
[45] Y. Subaşı, C. H. Fleming, J. M. Taylor, and B. L.
Hu, Equilibrium states of open quantum systems in the
strong coupling regime, Phys. Rev. E 86, 061132 (2012).
doi:10.1103/PhysRevE.86.061132
[46] P. L. Orman and R. Kawai, A qubit strongly interacting
with a bosonic environment: Geometry of thermal states,
arXiv:2010.09201v1 (2020).
[47] K. Goyal and R. Kawai, Steady state thermodynam-
ics of two qubits strongly coupled to bosonic envi-
ronments, Phys. Rev. Research 1, 033018 (2019).
10
doi:10.1103/PhysRevResearch.1.033018
[48] W. H. Zurek, Decoherence, einselection, and the quantum
origins of the classical, Rev. Mod. Phys. 75, 715 (2003).
doi:10.1103/RevModPhys.75.715
[49] E. K. Irish, J. Gea-Banacloche, I. Martin, and K. C.
Schwab, Dynamics of a two-level system strongly coupled
to a high-frequency quantum oscillator, Phys. Rev. B 72,
195410 (2005). doi:10.1103/PhysRevB.72.195410
[50] Zi-Min Li and Murray T. Batchelor, Gener-
alized adiabatic approximation to the quantum
Rabi model, Phys. Rev. A 104, 033712 (2021).
doi:10.1103/PhysRevA.104.033712
[51] A. Kolli, A. Nazir, and A. Olaya-Castro, Electronic exci-
tation dynamics in multichromophoric systems described
via a polaron-representation master equation, J. Chem.
Phys. 135, 154112 (2011). doi:10.1063/1.3652227
[52] C. K. Lee, J. Moix, and J. Cao, Accuracy of second or-
der perturbation theory in the polaron and variational po-
laron frames, J. Chem. Phys. 136, 204120 (2012). doi:
10.1063/1.4722336
[53] A. Gelzinis and L. Valkunas, Analytical derivation of
equilibrium state for open quantum system, J. Chem.
Phys. 152, 051103 (2020). doi: 10.1063/1.5141519
[54] G. Timofeev and A. Trushechkin, Hamiltonian of mean
force in the weak-coupling and high-temperature ap-
proximations and refined quantum master equations,
arXiv:2204.00599.
[55] V. May and O Kühn, Charge and Energy Transfer Dy-
namics in Molecular Systems (Wiley-VCH, Weinheim,
2011).
[56] Richard P. Feynman, An Operator Calculus Having Ap-
plications in Quantum Electrodynamics, Phys. Rev. 84,
108-128 (1951). doi:10.1103/PhysRev.84.108
[57] A. J. Leggett, S. Chakravarty, A. T. Dorsey, M. Fisher,
A. Garg, and W. Zwerger, Dynamics of the dissipa-
tive two-state system, Rev. Mod. Phys. 59, 1 (1987).
doi:10.1103/RevModPhys.59.1