Arxiv:2011.04053V2 [Quant-Ph] 20 Apr 2021 Where Needn Eso Fe.( Eq

Home , Open quantum system

arXiv:2011.04053v2 [quant-ph] 20 Apr 2021 where needn eso fE.( Eq. time- of The version operators. trace-less independent orthonormal of set the are rlcniin,i nw ob ecie ytegeneral the by described be equation to master known is conditions, eral ern ok [ works neering 23 thermody- quantum [ in walks those as [ such namics areas in ifications or absence [ the tures [ showing effects structure memory of rich presence a with evolution where cinwt nabetevrnet[ environment ambient an with action C)dnmclmp hc sgvnb operator-sum positive by completely given a is representation: which Kraus) i.e., (or map, channel, dynamical a (CP) by described be ∗ ‡ † [email protected] [email protected] [email protected] noe unu ytmeouin ne ut gen- quite under evolution, system quantum open An pnqatmsses hc r ytm ninter- in systems are which systems, quantum Open qiaety noe unu ytmeouincan evolution system quantum open an Equivalently, [ (cf. others among ], ρ ˙ ( t 1 γ 10 = ) 19 hoeia cecsDvso,Ponpan nttt of Institute Poornaprajna Division, Sciences Theoretical K j 14 L ≡ − ( – j , t 13 ( ρ ) 1 2 – − 20 t ( 16 { r h iedpnetdcyrts and rates, decay time-dependent the are ) iglrte,mxn n o-akvaiyo al dynam Pauli of non-Markovianity and mixing Singularities, ( t .Oe ytmeet aepoon ram- profound have effects system Open ]. a etitoso h xeietlraiaino non-in poi of is realization indivisibility) experimental mixing. channel the positive on completely restrictions of cha ial resultant sense the whic the in channels, singularity (in o singular the elimination of of the properties types to different two lead between can distinguish singularity We a with channels ing nwrteqeto ntengtv ntecneto ui Pa qubit of context the channels in non-singular negative the mixing in by analogou question produced an the answer be consider We can channel authors. the various by recently served ~ L i ,qatmcreain n oeec [ coherence and correlations quantum ], t = ) ,qatmcytgah [ cryptography quantum ], )[ r h ru prtr.Tedynami- The operators. Kraus the are [ j † unu o-akvaiyo hnescnb rdcdb mix by produced be can channels of non-Markovianity Quantum H 25 ( ρ t ( S hiatUtagi, Shrikant E ) t , L ( )] 3 ( 2 26 t t nedsilnr rga nQatmIfrainadComp and Information Quantum on Program Interdisciplinary .Introduction I. j rdaeSuis aia cdm fHge dcto,Man Education, Higher of Academy Manipal Studies, Graduate ) , )[ ( ρ , ]. t ρ ) ( 0]= (0)] ρ , 24 t ]+ )] ( ]). t ) 2 } – X 1 X 9 j ,wssuidi h pio- the in studied was ), j ,uia rnnuia fea- non-unital or unital ], ninIsiueo ehooy ohu-407 India. Jodhpur-342037, Technology, of Institute Indian ,2, 1, γ K j ( j t ∗ ( ) t ) io .Rao, N. Vinod ρK 1 L 17 ,eprec an experience ], j ( j † , t ( 18 ) t ρ ) ( , ,quantum ], t ) L j † ( t { ) L 21 1 (1) (2) j .Srikanth, R. } – cetfi eerh iau eglr–526,India 562164, Bengaluru– Bidalur Research, Scientific hntemap the then ie by given ffrall for If sC fadol fteCo matrix Choi the if only and if CP is h solution the E h w-aaee opsto faC map CP a of composition Consider two-parameter state. entangled the maximally unnormalized an is a map cal hne ntesneo Pidvsblt,a endabove. a defined as of CP-indivisibility, non-Markovianity of sense of the concept in channel evo- [ the non-Markovian others consider we of for work, plethora criteria been a the has among of lution, CP-indivisibility one as CP-indivisible. proposed is it erwise, fsnua onso h a n fsnua channels. singular of and map the of points singular of I o all for pnst iglrt ntegenerator the in singularity a to sponds E o nyis only not opsto q ( Eq. composition ibr pc fa of space Hilbert hwn htnnivriiiyo h map the of non-invertibility that showing n furthermore, And o-netbeadthus and non-invertible iglrpit ftechannel the of point) singular nend hntepoint the then undefined, ˙ )[ ( ( t ,t t, ento 1. Definition notions the are work, this in purpose, our for Important opeepstvt of positivity Complete | = ) ψ etbeqatmcanl yapoesof process a by channels quantum vertible nl h oncint non-Markovianity to connection The nnel. + i tdot hs eut moenontriv- impose results These out. nted ) ih t 1, ⊗ i L iglrte nterslatchannel. resultant the in singularities f ψ † t edudrmxn obodyquite broadly to mixing under lead h ≤ l hnes nteohrhn,mix- hand, other the On channels. uli usino hte iglrte of singularities whether of question s ( E f + I .. nsta akte.Hr we Here them. lack that ones i.e., , t n uhsihBanerjee. Subhashish and d ( )[ s | ≥ t ] where , E ) ≤ E ≥ E t E ( ( n akva hnes sob- as channels, Markovian ing tefoestemse qain[ equation master the obeys itself t ( ≥ ( t ,t t, E )] 0 ,t t, where , t fteei time a is there If ( f , t o all for d t tto (IDRP-QIC), utation i t , i i f ota h iedpnetmphas map time-dependent the that so ) dmninlacla h map The ancilla. -dimensional L pl-714 India. -576104, ipal 5 h nemdaemap intermediate the , = ) t , i al,bcuetemap the because fails, ) ( ) I oiie u oi n extension any is so but positive, t i d = ) scle Pdvsbe[ CP-divisible called is = ) T T steiett prtri the in operator identity the is t E exp E ( ≥ t E E t stetm-reigoperator. time-ordering the is ˙ ( ∗ f ( ( ,t t, t t t t , scle h iglrt (or, singularity the called is i ) f E where , E Z i ∗ t , ( ) t ) − clmaps ical t i t ) f sterqieetthat requirement the is 1 E ≡ E L t , ( ( t t ( ,t t, i ) s 6 ) , 3, = ) , ( [ | ds t i ‡ 34 ψ 8 f ) t , . t , + χ L ∗ .Furthermore, ]. 29 i ( i , uhta the that such E t ) ( = = ) – E ( . 5 33 t ( E , f | t E 00 ( ∗ t , 28 E .I this In ]. ( t t , t ) E ( i ) ∗ .Oth- ]. ,t t, E i ( + t , 7 sCP, is ) corre- t ( − , f ,t t, i i | t , ) ) 1 11 (5) 27 (3) (4) ⊗ i i is is ) ) i ] 2 the channel is called “singular”. If no such singular points II. Mixing non-singular Pauli channels t∗ exist, then the channel is said to be non-singular (or, regular). A general Pauli dynamical map is given by Note that the singularity of the channel can be accompanied by perfectly regular dynamics, i.e., the map 3 † (t ,t ) itself is well-defined (albeit non-invertible) [27, (t)[ρ]= ki(t)σiρσi , (6) ∗ i E E35]. Xi=0 Ref. [36] discusses a method for making singularities where σ = I, and σi, i 1, 2, 3 are Pauli X, Y, Z op- 0 ∈{ } tractable in the context of the definition of CP-divisibility erators respectively, and 3 k (t)=1. The canonical of maps [5]. Building thereon, a measure of singularities i=0 i form of master equationP corresponding to the map (6) of the maps is presented in [34]. An account of handling has the form the singularities was reviewed in [29, Sec. 4.3]. These measures based on CP-indivisibility are equivalent to the 3 † one based on decay rate [7] up to a constant factor. ρ˙(t)= (t)[ρ(t)] = γj (t)(σj ρ(t)σ ρ(t)) (7) L j − The effect of mixing different quantum evolutions has Xj=1 attracted attention of late. References. [37, 38] show where γ (t) are the rates. that a convex combination of semigroup dynamical maps j can lead to a deviation from the semigroup structure. The decay rates may be readily obtained using Eq. (4) Quite interestingly, the convex combination ′ = (1 [43]. Noting that E − p) 1 + p 2 of two semigroup (hence, CP-divisible) maps E E (t)[σj ]= λj (t)σj . (8) 1 = exp t 1 and 2 = exp(t 2) may give rise to CP- E E { L } E L indivisible (even eternally CP-indivisible) evolution [39]. we have More recently various authors have shown that it is possible to obtain a CP-indivisible Pauli channel by mixing −1 1 ˙(t)[σj ]= λ˙ j (t)σj ; (t)[σj ]= σj , (9) CP-divisible Pauli channels [39–41], implying that the set E E λj of CP-divisible channels is not convex. For almost all relevant works in the literature, includ- showing that the vanishing of a λj at some time t∗ cor- ing those cited above [29, 34–36, 42], instances of sin- responds to non-invertibility of the map at that instant, gularity of a channel are always accompanied by non- and thus to a singularity of the Pauli channel, per the Markovianity in the sense of CP-indivisibility (though argument following Eq. (4). This is made more explicit the converse is not true). In this light, our above ob- below. servation concerning the mixing of Markovian channels Now, from (9) and (4), we readily obtain the rates in prompts the question of whether an analogous behavior the master equation Eq. (7): holds with respect to mixing singular channels. This will ˙ ˙ ˙ be important for understanding the geometry of quantum 1 λ1(t) λ2(t) λ3(t) γ1(t)= , channels. 4λ1(t) − λ2(t) − λ3(t)

In particular, restricting to the context of mixing Pauli 1 λ˙ 2(t) λ˙ 1(t) λ˙ 3(t) channels, we ask whether singular channels can be pro- γ2(t)= , 4λ2(t) − λ1(t) − λ3(t) duced by mixing non-singular ones, and answer the ques- ˙ ˙ ˙ tion in the negative. This negative result implies that 1 λ3(t) λ1(t) λ2(t) γ3(t)= . (10) non-singular channels form a convex set. On the other 4λ3(t) − λ1(t) − λ2(t) hand, mixing singular channels does not necessarily result in a singularity of the resultant channel, showing Note in particular that the γj ’s have a singular point that singular channels do not form a convex set. Finally, when any of the λi vanishes. Our ﬁrst result, below, es- we explain why in the context of mixing Pauli channels, sentially asserts the convexity of non-singular Pauli chan- singularities of the channel imply CP-indivisibility, but nels. the converse is not true. This connection between sin- Lemma 1. It is impossible to produce a singular Pauli gularity and CP-indivisibility does not hold in general. channel by mixing only non-singular Pauli channels. Proof. Let the Pauli channels that are being mixed be This work is organized as follows. In Sec. II, we show given by that it is not possible to produce singularities of the chan- (ρ) (1 p(t))ρ(0) + p(t)σ ρσ , nel by mixing non-singular channels. In Sec. III and E1 ≡ − 1 1 IV, we discuss the results pertaining to mixing singular 2(ρ) (1 q(t))ρ(0) + q(t)σ2ρσ2, E ≡ − channels of two broad types. The interplay of singulari- 3(ρ) (1 r(t))ρ(0) + r(t)σ3ρσ3, (11) ties and non-Markovianity is discussed in Sec. V. In all E ≡ − cases, the results are illustrated with examples. Finally, where the functions p, q, and r quantify the degree of de- we conclude in Sec. VI. coherence of the channels and must satisfy 0 p,q,r 1 ≤ ≤ 3

1 to ensure complete positivity of the maps. The corre- p(t), q(t) or r(t) in Eq. (11) is 2 . sponding individual Lindblad rates are In this case, the occurrence of non-Markovianity (CP- indivisibility) can be attributed to the non-monotonicity η˙ p t γη = − , (12) of the decoherence functions ( ) etc., leading to reco- 1 2η herence in the negative slope region of the functions. − Typical instances of interest here would be channels for where η p(t), q(t), r(t) . Let the three channels in ∈ { } which the decoherence function is non-monotonic, oscil- Eq. (11) be mixed with probabilities a,b and c, where lating between 0 and 1 . In Sec. III, we shall show that: 0 a,b,c 1 and a + b + c = 1. This gives rise to the 2 p ≤ ≤ (a) mixing two such channels with singular points t∗ and channel: q t∗ produces a singularity only if their singularities are simultaneous (tp = tq ), and moreover the resultant sin- ˜(ρ)= a (ρ)+ b (ρ)+ c (ρ) ∗ ∗ E E1 E2 E3 gularity occurs at the same time; and furthermore, (b) = (1 ap bq cr)ρ + apσ ρσ + bqσ ρσ − − − 1 1 2 2 three-way mixing of such channels can never produce a + crσ3ρσ3. (13) singularity. Definition 3. Channels of type II: Those in which By assumption, the mixing maps 1, 2 and 3 are non- the maximum value attained by the decoherence func- singular. In view of Eq. (12), thisE impliesE thatE 1 tion p(t), q(t) or r(t) in Eq. (11) can exceed 2 . Typical instances of interest here would be channels for 1 0 p(t), q(t), r(t) < (14) which the decoherence function is monotonic, reaching an ≤ 2 1 asymptotic value in the interval ( 2 , 1]. The positive slope for finite time t. The time-dependent eigenvalues of the region of the decoherence function p(t) etc., when they map ˜ from Eq. (13) read exceed half, corresponds to recoherence of the system, E leading to non-Markovianity. Unlike in the case of Type λ (t)=1 2(bq + cr), (15a) I channels, we will find, in Sec. IV, that the features 1 − λ (t)=1 2(ap + cr), (15b) (a) and (b) do not hold in this case, i.e., singularities 2 − need not be simultaneous, and the restriction to two-way λ (t)=1 2(ap + bq). (15c) 3 − mixing is not needed. We shall find below that the conditions under which The condition for a singularity in the resultant channel mixing of channels leads to a singularity in the resultant is that one or more of λ in Eq. (15) should vanish at j channel depends on the type of the channels being mixed. a certain finite time(s) ts. For example, consider λ1 in Eq. (15a). Given the range restriction Eq. (14) on the We may note that Type I is a more usual occurrence, decoherence functions p(t) and q(t), we have and can for example result when a qubit system and its qubit environment evolve according to a Hamiltonian given by ω( 01 10 + 10 01 ) acting on the initial state λ1(t) > 1 (b + c) 0 (16) − ≥ 01 , where |ω isi h a real| | number.i h | The joint system remains | i for finite t. Repeating the argument for λ2 and λ3, we in the subspace spanned by 01 , 10 , and the reduced 2 {| i | i} 2 conclude that there can be no singularity in the mixed state of the system is cos (ωt) 0 0 + sin (ωt) 1 1 . | i h | | i h | channel.

III. Mixing of Type I channels It follows from Eq. (16) and analogous results for λ2 and λ3 that the non-singular mixing channels necessarily have positive decay rates γj , and therefore are CP- In Eq. (15), suppose λ1(t) vanishes at some finite time divisible. Thus, as a corollary of Lemma 1, we find that t∗. Observe that this can happen only if q(t∗)= r(t∗)= 1 it is impossible to produce a singularity by mixing CP- . In other words, the mixing channels 2 and 3 must 2 E E divisible Pauli channels. each possess a singularity such that these singularities Lemma 1 does not address the question of whether occur simultaneously at t∗, which coincides with the sin- mixing singular channels produces a singularity in the gularity in the resultant channel. Moreover, we require resultant channel. To address this question, it is conve- that the mixing parameter a =0, meaning that only two nient to distinguish two types of singular channels. It is of three channels should be mixed. A similar argument clear from Eq. (12) that for a Pauli channel to be sin- holds for λ2(t) and λ3(t). gular, the decoherence function p(t), q(t) or r(t), as the To summarize, in the context of mixing Type I chan- 1 case may be, should attain the value of 2 at some finite nels to produce a singularity, precisely two channels time t. Accordingly, the two types of Pauli singular chan- should be mixed, and they should be synchronized in 1 nels are those where the value 2 is the maximum or is the occurrence of their singularities. If they are not syn- exceeded. It turns out that they evince quite different chronized, then the singularity will be eliminated in the behaviors under mixing. resultant channel. Definition 2. Channels of Type I: Those in which the The fact that the mixing of two singular channels can maximum value attained by the decoherence function produce a non-singular channel can be compared to the 4 situation that mixing CP-indivisible channels can result mixing of the X,Y, and Z Pauli RTN channels, as a result in a CP-divisible channel - even a semigroup [39, 40]. of which we obtain an isotropic RTN Pauli channel. A consequence is that, like CP-indivisible channels, Now consider individual RTN Pauli channels of Type singular ones also form a nonconvex set. I with their respective decoherence function being

Example 1. Let the mixing channels be 1 and 2, 1 Λ(t) 1 2 1 E 2 E p(t)= q(t)= r(t)= − , (20) with p(t) = 2 [1 cos (µt)] and q(t) = 2 [1 cos (νt)], 2 a,b > 0 and a + −b =1 in Eq. (13). In view of− Eq. (10), where Λ(t) is given by Eq. (19) with µ = µ = µ . the eigenvalues λi(t) of the resultant channel read 1 2 3 We now consider the question of whether the above RTN 2 λ1(t)=1 b sin (νt), non-Markovian depolarizing channel can be reproduced − by mixing the individual RTN Pauli channels of Type I. λ (t)=1 a sin2(µt), 2 − For d w, the zeros of Λ occur periodically, making 2 2 ≫ λ3(t)=1 a sin (µt) b sin (νt), (17) the channel possess an inﬁnite number of singularities. − − For d < w, Λ(t) attains zero only at t = , making the showing that there is a singularity only from the zeros of channel non-singular. ∞ λ3(t), and furthermore this happens if and only if the two Eq. (15) yields the following eigenvalues of the resul- trigonometric terms attain 1 at the same time t∗, which tant channel: will also be the singular point of the resultant channel.

A simple way to ensure this is by having µ = ν, in which λ1(t)=1 (b + c)(1 Λ), case singularities occur in the resultant channel for t = − − λ (t)=1 (a + c)(1 Λ), n π . 2 − − 2 λ (t)=1 (a + b)(1 Λ). (21) 3 − − Whilst in general Λ takes values in the range ( 1, +1], It is not hard to show that this behavior, of the sin- − gularities of the mixing channels to be simultaneous at but to conform to Type I, the parameters d and w must some time t∗, and leading to a singularity at the same be so chosen that Λ(t) is confined in the range [0, 1], with R the singularity occurring when Λ(t)=0. If a,b,c > 0, time t∗ = t∗ in the resultant channel, is general for Type I channels. then the sum of any two of them is strictly less than The following example illustrates the idea that the 1. It follows from Eq. (21) and the Type I restriction number of mixing channels should not exceed 2. Other- (requiring that Λ is bounded below by 0) that each λj in wise the singularity is eliminated. Eq. (21) never vanishes. We now consider an analogous result when type II Example 2. A depolarizing colored noise is the Random channels are mixed, and show how it contrasts with the telegraph noise (RTN) non-Markovian depolarizing above two examples. † channel [ρ]= AiρA , with the Kraus operators [44] E i i Ai = √Piσi, whereP σ0 = I, σx = σ1, σy = σ2, σz = σ3 are Pauli operators. Here, IV. Mixing of Type II channels 1 P0 = [1 + Λ1 +Λ2 +Λ3], In Equation (15), if we relax the requirement that 4 p, q and r are bounded by 1 , then we obtain Type II 1 2 P1 = [1 + Λ1 Λ2 Λ3], channels. In this case, we will find that neither the 4 − − synchronization nor restriction of the mixing channels 1 P = [1 Λ +Λ Λ ], to two, is required, for producing a singular channel. 2 4 − 1 2 − 3 1 Example 3. Consider the same system as in Ex- P3 = [1 Λ1 Λ2 +Λ3], (18) 1 4 − − ample 2, but letting p, q, and r to exceed 2 . Accordingly, Λ takes values in the range ( 1, +1]. For simplicity, let where 1 − a = b = c = 3 , in which case the eigenvalues become sin(wtµi) Λi = exp( wt) + cos(wtµi) , (19) 1 − µi λ = [1 + 2Λ(t)]. (22) 1,2,3 3 1 The quantity w = 2τ is the spectral bandwidth while τ The singularity of the resulting RTN depolarizing chan- is the rate of fluctuation of the environment affecting the 1 nel occurs when Λ(t)= 2 . From Eq. (20), we find that 2di 2 − 3 qubit, and µi = 1, with di representing the this singularity happens when p(t) = q(t) = r(t) = , w − 4 coupling strengthsq corresponding to the ith Pauli chan- which is permissible for mixing channels of type II. nel. For the present purpose, let all di’s in Eq. (19) be As a final example, consider the three mixing channels taken to be equal, given by d. This corresponds to equal to be of type II, having possibly different functional 5 forms, but all reaching an asymptotic value greater than VI. Discussions and Conclusions 4 (say) 5 . This work discusses the problem of producing a sin- R R 3 Example 4. At some time t∗ , let q(t∗ ) = 5 and gular general Pauli dynamical map by the mixing non- R 4 r(t∗ ) = 5 . The singularities of Y and Z occur, singular (or, regular) Pauli channels. We point out that 2 3 2 E 3 E 1 respectively, at t∗ and t∗, where q(t∗)= r(t∗)= 2 , where it is impossible to do so. Different conditions on the 2 3 in general we don’t require t∗ = t∗, i.e., the singularities classes of mixing singular channels are considered in or- of the mixing channels are not necessarily synchronized. der to guarantee that the resultant channel is singular. 1 1 Furthermore let the mixing fraction b = 6 and c = 2 , so In particular, we show that: (i) for Type I channels it that a = 1 b c = 1 , implying that there is a finite is possible to produce a singular channel by mixing two − − 3 fraction of the channel X in the mixing. It follows from singular Pauli channels, provided the occurrence of their E R singularities is synchronized; (ii) mixing three Type I sin- Eq. (15a) that λ1 = 0 at t∗ , meaning that this is a R gular channels results in the elimination of singularities singularity of the resultant channel. Note that t∗ need j i.e., such a convex combination results in a regular chan- not coincide with either t∗ (j =2, 3). nel; and (iii) by contrast, in the case of Type II channels, we have shown that mixing two or three singular chan- V. Interplay of singularities and non-Markovianity nels can result in a singular channel, and the singularities need not be synchronized in their occurrence. A further question that may be considered here is It turns out that for the resultant channels considered the power of mixing weaker forms of non-Markovianity here, singular channels are necessarily non-Markovian (in than CP-indivisibility (cf. Reference [8] and references the CP-indivisible sense). In Eq. (15), consider the point therein), in terms of generating singularities and/or t∗ where the first singularity is encountered, i.e., one of stronger forms of non-Markovianity. An aspect of this for the λj (t∗) vanish, say λ1(t∗)=0. From Eq. (15a), we the measure of CP-indivisible maps produced by mixing ˙ have λ1(t) 2(bq˙ + cr˙). In the case of Type I channels, a class of Pauli maps was considered in Reference [41]. ≡− 1 both q(t) and r(t) reach 2 and fall off at the same time. A future direction would be to explore the results of this Therefore, λ˙ 1(t) is negative just before t∗, and flips sign paper from a geometric point of view [45–47], in particu- to positive just after t∗. On the other hand, λ1 remains lar to quantify the measure of non-singular channels pro- positive for all time. Thus: duced by mixing singular ones, analogous to results for CP-divisible channels in References [40, 41]. Reference ˙ λ1 [48] shows how an operation of coarse-graining in time lim = . (23) ± t→t∗ λ1 ±∞ while transforming a master equation from a nonlocal integro-differential form to a time-local one can modify This implies that γ2 and γ3 flip the sign from positive to the CP-indivisibility property of the dynamics [48]. An negative at the singularity, and γ1 the other way. interesting question here would be whether this approx- For Type II channels, by virtue of monotonic increase imation procedure can also modify the (non-)singular ˙ of q(t) and r(t), λ1(t) 2(bq˙ + cr˙) < 0 for all time. On property of the dynamics. ≡− the other hand, λ1 flips its sign from positive to negative Ref. [49] reports on an experimental implementation of at the singularity. Thus, Eqs. (23) and the attendant producing non-Markovianity by two-way mixing of Pauli consequences for the decay rates hold here too. There- semigroups in a linear quantum optical platform, an ex- fore, curiously, despite the contrasting behavior in the tension of which to three-way mixing of more general CP- eigenvalues and the rate of change, yet in both Type I and divisible maps has recently been proposed [41]. This idea Type II channels, singularities signal CP-indivisibility in may be adapted for mixing singular channels, which can a similar way. be produced by suitable bath engineering, possibly with It may be worth pointing out that singularities do appropriate modifications to the experimental implemen- not necessarily imply CP-indivisibility. For illustration, tation in Ref. [50] in the case of a photonic realization, or consider a CP-indivisible dephasing channel described in Reference [51], in the case of a semiconductor quantum by [ρ] γ(t)(σ3ρσ3 ρ) with the decay rate γ(t) = optical realization. tan(Lωt),≡ where ω is some− real number. The channel has Our result here, that singularities cannot be produced (2n+1)π by mixing non-singular channels, is shown only for qubit an infinitely many singularities at t∗ = 2ω for inte- ger n, which will flip the sign of the rate, and thus signal Pauli channels. We expect that this is true quite gener- CP-indivisibility in a similar manner as discussed above. ally, and in particular, applicable to non-Pauli maps. By contrast, consider the same channel, but with decay rate γ(t) = tan2(ωt). This channel contains singularities at the same instants as the above channel, but the sign Acknowledgments of the rate never flips from positive to negative at any of these singularities and thus corresponds to a CP-divisible SU and VNR acknowledge Admar Mutt Educa- process. tion Foundation for support from a scholarship. RS 6 acknowledges the support of Department of Sci- Physical Systems (ICPS) program of the Department ence and Technology (DST), India, under Grant No. of Science and Technology (DST), India, under Grants MTR/2019/001516. RS and SB acknowledge, respec- No.: DST/ICPS/QuEST/Theme-1/2019/14 and No. tively, the support from the Interdisciplinary Cyber DST/ICPS/QuEST/Theme-1/2019/6.

[1] Subhashish Banerjee. Open quantum systems : Dynamics [17] Kishore Thapliyal, Anirban Pathak, and Subhashish of Nonclassical Evolution, volume 20. Springer Nature Banerjee. Quantum cryptography over non-markovian Singapore, 2018. channels. Quantum Information Processing, 16(5):115, [2] Heinz-Peter Breuer and Francesco Petruccione. The the- 2017. ory of open quantum systems. Oxford University Press, [18] Shrikant Utagi, R Srikanth, and Subashish Banerjee. 2002. Ping-pong quantum key distribution with trusted noise: [3] R Srikanth and Subhashish Banerjee. Squeezed gen- Non-Markovian advantage. Quantum Inf.Process.19,366 eralized amplitude damping channel. Phys. Rev. A, (2020). 77(1):012318, 2008. [19] N Pradeep Kumar, Subhashish Banerjee, R Srikanth, [4] S Omkar, R Srikanth, and Subhashish Banerjee. Dissipa- Vinayak Jagadish, and Francesco Petruccione. Non- tive and non-dissipative single-qubit channels: dynamics markovian evolution: a quantum walk perspective. Open and geometry. Qu. Inf. Proc, 12(12):3725–3744, 2013. Systems & Information Dynamics, 25(03):1850014, 2018. [5] Ángel Rivas, Susana F Huelga, and Martin B Plenio. [20] Javid Naikoo, Subhashish Banerjee, and CM Chan- Entanglement and non-markovianity of quantum evolu- drashekar. Non-markovian channel from the reduced dy- tions. Phys. Rev. Lett, 105(5):050403, 2010. namics of coin in quantum walk. Physical Review A, [6] Heinz-Peter Breuer, Elsi-Mari Laine, and Jyrki Piilo. 102(6), 062209 (2020). Measure for the degree of non-markovian behavior of [21] Samyadeb Bhattacharya, Subhashish Banerjee, and quantum processes in open systems. Phys. Rev. Lett, Arun Kumar Pati. Evolution of coherence and 103(21):210401, 2009. non-classicality under global environmental interaction. [7] Michael J. W. Hall, James D. Cresser, Li Li, and Erika Quantum Information Processing, 17(9):236, 2018. Andersson. Canonical form of master equations and [22] Subhashish Banerjee, V Ravishankar, and R Srikanth. characterization of non-markovianity. Phys. Rev. A, Dynamics of entanglement in two-qubit open system in- 89:042120, Apr 2014. teracting with a squeezed thermal bath via dissipative [8] Shrikant Utagi, R Srikanth, and Subhashish Banerjee. interaction. Annals of Physics, 325(4):816–834, 2010. Temporal self-similarity of quantum dynamical maps as [23] Javid Naikoo, Supriyo Dutta, and Subhashish Baner- a concept of memorylessness. Scientific Reports, 10(1):1– jee. Facets of quantum information under non-markovian 10, 2020. evolution. Phys. Rev. A, 99:042128, Apr 2019. [9] C-F Li, G-C Guo, and J Piilo. Non-Markovian quan- [24] C-F Li, G-C Guo, and J Piilo. Non-Markovian quantum dynamics: What does it mean? EPL (Europhysics tum dynamics: What is it good for? EPL (Europhysics Letters), 127(5):50001, 2019. Letters), 128(3):30001, 2020. [10] Frank Verstraete and Henri Verschelde. On quantum [25] Vittorio Gorini, Andrzej Kossakowski, and Ennackal channels. arXiv:0202124, 2002. Chandy George Sudarshan. Completely positive dynam- [11] Mary Beth Ruskai, Stanislaw Szarek, and Elisabeth ical semigroups of n-level systems. Journal of Mathemat- Werner. An analysis of completely-positive trace- ical Physics, 17(5):821–825, 1976. preserving maps on m2. Linear algebra and its appli- [26] Göran Lindblad. Completely positive maps and entropy cations, 347(1-3):159–187, 2002. inequalities. Communications in Mathematical Physics, [12] Subhashish Banerjee and R Ghosh. Dynamics of deco- 40(2):147–151, 1975. herence without dissipation in a squeezed thermal bath. [27] Dariusz Chruściński and Andrzej Kossakowski. Non- Journal of Physics A: Mathematical and Theoretical, markovian quantum dynamics: local versus nonlocal. 40(45):13735, 2007. Physical review letters, 104(7):070406, 2010. [13] Javid Naikoo and Subhashish Banerjee. Coherence-based [28] Dariusz Chruściński, Andrzej Kossakowski, and Án- measure of quantumness in (non-) markovian channels. gel Rivas. Measures of non-markovianity: Divisibil- Quantum Information Processing, 19(1):29, 2020. ity versus backflow of information. Physical Review A, [14] Massimiliano Esposito, Upendra Harbola, and Shaul 83(5):052128, 2011. Mukamel. Nonequilibrium fluctuations, fluctuation the- [29] Angel Rivas, Susana F Huelga, and Martin B Plenio. orems, and counting statistics in quantum systems. Re- Quantum non-markovianity: characterization, quantifi- views of modern physics, 81(4):1665, 2009. cation and detection. Rep. Prog. Phys, 77(9):094001, [15] Michele Campisi, Peter Hänggi, and Peter Talkner. Col- 2014. loquium: Quantum fluctuation relations: Foundations [30] Dariusz Chruściński and Sabrina Maniscalco. Degree of and applications. Reviews of Modern Physics, 83(3):771, non-markovianity of quantum evolution. Physical review 2011. letters, 112(12):120404, 2014. [16] George Thomas, Nana Siddharth, Subhashish Banerjee, [31] B Bylicka, D Chruściński, and Sci Maniscalco. Non- and Sibasish Ghosh. Thermodynamics of non-markovian markovianity and reservoir memory of quantum channels: reservoirs and heat engines. Phys. Rev. E, 97:062108, Jun a quantum information theory perspective. Sci. Rep, 4,1 2018. (2014). 7

[32] Li Li, Michael J.W. Hall, and Howard M. Wiseman. arXiv:2002.11452, 2020. Concepts of quantum non-markovianity: A hierarchy. [43] Dariusz Chruściński and Filip A Wudarski. Non- Physics Reports, 759:1 – 51, 2018. markovian random unitary qubit dynamics. Physics Let- [33] Shrikant Utagi. Quantum causal correlations and non- ters A, 377(21-22):1425–1429, 2013. markovianity of quantum evolution. Physics Letters A, [44] Sonja Daﬀer, Krzysztof Wódkiewicz, James D Cresser, 386:126983, 2021. and John K McIver. Depolarizing channel as a completely [34] SC Hou, XX Yi, SX Yu, and CH Oh. Singularity of positive map with memory. Phys. Rev. A, 70(1):010304, dynamical maps. Physical Review A, 86(1):012101, 2012. 2004. [35] U. Shrikant, R. Srikanth, and Subhashish Banerjee. Non- [45] Vinayak Jagadish, R Srikanth, and Francesco Petruc- markovian dephasing and depolarizing channels. Phys. cione. Measure of positive and not completely pos- Rev. A, 98:032328, Sep 2018. itive single-qubit pauli maps. Physical Review A, [36] SC Hou, XX Yi, SX Yu, and CH Oh. Alternative non- 99(2):022321, 2019. markovianity measure by divisibility of dynamical maps. [46] Vinayak Jagadish, R. Srikanth, and Francesco Petruc- Physical Review A, 83(6):062115, 2011. cione. Measure of not-completely-positive qubit maps: [37] Michael Marc Wolf, J Eisert, TS Cubitt, and J Igna- The general case. Phys. Rev. A, 100:012336, Jul 2019. cio Cirac. Assessing non-markovian quantum dynamics. [47] Katarzyna Siudzińska. Geometry of pauli maps and pauli Physical review letters, 101(15):150402, 2008. channels. Physical Review A, 100(6):062331, 2019. [38] Dariusz Chruściński, Andrzej Kossakowski, and Saverio [48] Nina Megier, Andrea Smirne, and Bassano Vacchini. Pascazio. Long-time memory in non-markovian evolu- The interplay between local and non-local master equations. Physical Review A, 81(3):032101, 2010. tions: exact and approximated dynamics. New Journal [39] Filip A Wudarski and Dariusz Chruściński. Markovian of Physics, 22(8):083011, aug 2020. semigroup from non-markovian evolutions. Physical Re- [49] SA Uriri, F Wudarski, I Sinayskiy, F Petruccione, and view A, 93(4):042120, 2016. MS Tame. Experimental investigation of markovian and [40] Vinayak Jagadish, R Srikanth, and Francesco Petruc- non-markovian channel addition. Physical Review A, cione. Convex combinations of pauli semigroups: Ge- 101(5):052107, 2020. ometry, measure, and an application. Physical Review A, [50] Simone Cialdi, Claudia Benedetti, Dario Tamascelli, Ste- 101(6):062304, 2020. fano Olivares, Matteo GA Paris, and Bassano Vacchini. [41] Vinayak Jagadish, R Srikanth, and Francesco Petruc- Experimental investigation of the eﬀect of classical noise cione. Convex combinations of cp-divisible pauli channels on quantum non-markovian dynamics. Physical Review that are not semigroups. Physics Letters A, 384:126907, A, 100(5):052104, 2019. 2020. [51] Alexander Carmele and Stephan Reitzenstein. Non- [42] Shrikant Utagi, Vinod N Rao, R Srikanth, and Sub- markovian features in semiconductor quantum optics: hashish Banerjee. A class of quasi-eternal non-Markovian quantifying the role of phonons in experiment and theory. Pauli channels and their measure. arXiv preprint Nanophotonics, 8(5):655–683, 2019.