arXiv:1904.08432v2 [cond-mat.quant-] 31 Oct 2019 ieeouino est arx(e .. es[,9]): Refs.[8, e.g., the (see governs matrix density that of equation evolution master time Lindblad under the in arises been recently, experiments[100–102]. Very have in observed 40]. 88–99]. been NHSE 58, 37, has 37, NHSE 31, 33, of a 25, 32, investigations[25, zone[24, implications bound- of Brillouin bulk-boundary Broader generalized at non-Bloch eigenstates the based theory the on band localized non-Bloch of and suggests 28] correspondence[24, are majority which aries, operator the non-Hermitian that namely where ytm sthe is systems in- experimental[82–87] vestigations. and in- theoretical[71–81] other among teresting classifications[68–70], and topological [57–67], novel and topo- 43–56] of insulators[25, generalizations 35–42], logical 31, topological non-Hermitian 27, 25, on new invariants[24, focused correspondence[22–34], been bulk-boundary have Consid- attentions attentions. growing erable topological attracting and been Re- have non-Hermiticity phases of lifetimes[17–21]. interplay finite the systems[6– with cently, quantum open and gain[1–5], 16], sys- and photonic loss as such with phenomena tems of range wide a for work iaosdsrbn unu up u oculn to coupling to and due environment, jumps the quantum describing sipators prtr eoeteocrec fajm,teshort- the jump, a Hamil- tonian non-Hermitian of effective the occurrence follows evolution the time Before operator. ρH o ml,teeeto onaycniini insignifi- commonly is is condition condition boundary boundary periodic such, of As effect cant. the small, too noe unu ytm,nnHriiiynaturally non-Hermiticity systems, quantum open In n ftermral hnmn fnon-Hermitian of phenomena remarkable the of One frame- natural a provide Hamiltonians Non-Hermitian ti eeal eivdta hntesse iei not is size system the when that believed generally is It eff † dρ dt [1 2 103]. 12, )[11, H = H − eff o-emta kneetadcia apn noe quantum open in damping chiral and effect skin Non-Hermitian steHamiltonian, the is i [ ,ρ H, = ad as hrldmigwt hr aern.Teeph These no wavefront. the to sharp dynamics. belong quantum a they with instead, damping Hamiltonians; non-Hermitian chiral un n a algebraic the cause Moreover, is systems bands condition. quantum boundary open shape open of under dramatically class exponential effect un a in its skin we damping uncover non-Hermitian the Letter, and the effect, this that skin In shown non-Hermitian the exhibit evolution. ma can Lindblad time the described generate for well calls superoperators be dynamics often long-time bou while can the tonians, evolution at short-time localized a exponentially systems, are eigenstates the that o-emta kneffect skin non-Hermitian n fteuiu etrso o-emta aitnasis Hamiltonians non-Hermitian of features unique the of One H + ] − X i µ P 1 L nttt o dacdSuy snhaUiest,Beijin University, Tsinghua Study, Advanced for Institute µ 2 scle h iuila super- Liouvillian the called is L L µ † µ L ρL L µ µ µ † as saeteLnba dis- Lindblad the are ’s { − e Song, Fei dρ/dt L µ † L 2,26](NHSE), [24, µ ρ , = 1 } −  hnuYao, Shunyu i L ≡ ( H eff ρ, ρ (1) − 1 I.1 a S oe ihsagrdhopping staggered with model SSH (a) 1. FIG. ai change basis a hoyi ae ntefl iuila.Although Liouvillian. the full Crucially, the on damping”. based “chiral is unidirectional, the theory is dubbed damping is the which that implies More- NHSE condition. over, boundary alge- open condition is under boundary exponential damping periodic while time under law) long Notable power the (i.e., braic that Liouvillian. found the the are of from examples NHSE derived the to matrix related damping is periodic-boundary the this a Furthermore, sys- that of system. that open-boundary show from an we differ of dramatically paper, tem dynamics this Lindblad In time rele- long more experiments. is to condition vant open-boundary though adopted, experimentally. feasible more is (b) Hamiltonian nseta ee(.. t aigNS rntde not does not fact in or is NHSE it having role, its matter). important (e.g., an here play inessential to expected be may Hamiltonian hopping The different A model. (b) same h framework. the equation of master realization the in [Eq.(2)] ihteoasidctn h ntcls h lc Hamil- Bloch The cells. unit the is indicating tonian ovals the with L osadgi r ecie ytedissipators the by described are gain and loss rainadanhlto prtr tsite at operators annihilation and creation n hn Wang Zhong and ( x l k Model.– ( = ) = √ dr ftesse.Froe quantum open For system. the of ndary -emta hsc ffl-egdopen full-fledged of n-Hermitian γ treutos nwihteLiouvillian the which in equations, ster nHriinsi ffc n non-Bloch and effect skin on-Hermitian t yteeetv o-emta Hamil- non-Hermitian effective the by l 1 c h xetdpyia osqecs tis It consequences. physical expected h o-emta kneet namely effect, skin non-Hermitian the h ogtm yais uhthat such dynamics, long-time the s n htLovlinsuperoperators Liouvillian that find xA h ytmi lutae nFg1a.Our Fig.1(a). in illustrated is system The ( e eidcbudr odto but condition boundary periodic der + k ( = ) nmn r eodteeffective the beyond are enomena t and H 2 σ cos y = ↔ L t ,108,China 100084, g, k 1 x g 1, ) P σ + σ = z ∗ x eas h anadls son-site, is loss and gain the Because . ij t 2 + √ h cos γ t ij g 2 c c sin k xA † i † ) c σ j b seuvln o()via (a) to equivalent is (b) . kσ x where , + z n h isptr are dissipators the and , t 2 sin systems c kσ i † c , y h fermion The . i i r fermion are L (including t l 1 and and H L t eff 2 g , 2 additional degrees of freedom such as spin is straight- (a) L=150 forward). We will consider single particle loss and gain, ˜n(t) l l -2 with loss dissipators Lµ = i Dµici and gain dissipa- 10 g g † tors Lµ = i Dµici , respectively.P For concreteness, we take h to be the Su-Schrieffer-Heeger (SSH) Hamiltonian,

P 10-4

namely hij = t1 and t2 on adjacent links. A site is also A t1=t2=1,¡ g= l=0.4 £ labelled as i = xs, where x refers to the unit cell, and B t1=0.7,t2=1,¢ g= l=0.6

s = A, B refers to the sublattice. For simplicity, let each 10-6 ¥ C t1=1.2,t2=1,¤ g= l=0.05

unit cell contain a single loss and gain dissipator (namely, § D t1=1.2,t2=1,¦ g= l=0.1 µ is just x): 10-8 1 10 100 t 1000 l g † † Lx = γl/2(cxA − icxB),Lx = γg/2(cxA + icxB); p q (2) l l g in other words, Dx,xA = iDx,xB = γl/2,Dx,xA = g p −iDx,xB = γg/2. We recognized in Eq.(2) that the σy = +1 statesp are lost to or gained from the bath. A seemingly different but essentially equivalent realization of the same model is shown in Fig.1(b), which can be obtained from the initial model [Fig.1(a)] after a basis change σy ↔ σz. Accordingly, the dissipators in Fig.1(b) are σz = 1 states. As the gain and loss are on-site, its experimental implementation is easier. Keeping in mind that Fig.1(b) shares the same physics, hereafter we focus on the setup in Fig.1(a). To see the evolution of density matrix, it is convenient to monitor the single-particle correlation † FIG. 2. (a) The damping of fermion number towards the ∆ij (t) = Tr[ci cj ρ(t)], whose time evolution is d∆ij /dt = † steady state of a periodic boundary chain with length L = 150 Tr[c cj dρ/dt]. It follows from Eq.(1) that (see the Sup- i (unit cell). The damping is algebraic for cases A, B with t1 plemental Material) ≤ t2, while exponential for C, D with t1 >t2. The initial state is † d∆(t) the completely filled state Q cxs 0 . (b) The eigenvalues of = i[hT , ∆(t)] −{M T + M , ∆(t)} +2M , (3) x,s | i dt l g g the damping matrix X. Blue: periodic-boundary; Red: open- boundary. The Liouvillian gap of the periodic-boundary chain g∗ g l∗ l where (Mg)ij = µ Dµi Dµj and (Ml)ij = µ DµiDµj , vanishes for A and B, while it is nonzero for C and D. For and both Ml andPMg are Hermitian matrices.P Majorana the open-boundary chain, the Liouvillian gap is nonzero in versions of Eq.(3) appeared in Refs.[8, 16, 104]. We can all four cases. This drastic spectral distinction between open define the damping matrix and periodic boundary comes from the NHSE (see text). T T X = ih − (Ml + Mg), (4) which recasts Eq.(3) as and express Eq.(6) as

d∆(t) † = X∆(t) + ∆(t)X +2Mg. (5) ˜ ∗ ˜ dt ∆(t)= exp[(λn + λn′ )t]|uRnihuLn|∆(0)|uLn′ ihuRn′ |. n,nX′ The steady state correlation ∆s = ∆(∞), to which long time evolution of any initial state converges, is deter- (8) † mined by d∆s/dt = 0, or X∆s + ∆sX +2Mg = 0. In this paper, we are concerned mainly about the dynam- By the dissipative nature, Re(λn) ≤ 0 always holds true. ics, especially the speed of converging to the steady state, The Liouvillian gap Λ = min[2Re(−λn)] is decisive for therefore, we shall focus on the deviation ∆(˜ t) = ∆(t) − the long-time dynamics. A finite gap implies exponential † converging rate towards the steady state, while a vanish- ∆s, whose evolution is d∆(˜ t)/dt = X∆(˜ t) + ∆(˜ t)X , which is readily integrated to ing gap implies algebraic convergence[106]. Periodic chain.–Let us study the periodic boundary ˜ Xt ˜ X†t ∆(t)= e ∆(0)e . (6) chain, for which going to momentum space is more con- We can write X in terms of right and left venient. It can be readily found that h(k) = (t1 + eigenvectors[105], t2 cos k)σx + t2 sin kσy and

X = λn|uRnihuLn|, (7) γl γg Ml(k)= (1 + σy),Mg(k)= (1 − σy). (9) Xn 2 2 3

(a) γg=γl=0.2 also give the steady state. In fact, our Ml and Mg ˜ satisfy M T + M = M γ/γ , which guarantees that n(t) l g g g ∆s = (γg/γ)I2L×2L is the steady state solution. It is -4 10 independent of boundary conditions. periodic Now we show the direct relation between the algebraic open L=30 damping and the vanishing gap of X. The eigenvalues of 10-8 open L=40 X(k) are open L=50 open L=60 -12 2 2 2 10 λ±(k)= −γ/2 ± i (t1 + t2 +2t1t2 cos k − γ /4) − it2γ sin k. (11) q 20 40 60 t 80 Let us consider t = t ≡ t for concreteness (case A in (b) γg=γl=0.2 1 2 0 ˜ L=60 nx(t) Fig.2), then λ−(π) = 0 and the expansion in δk ≡ k − π reads 10-3 t2 λ (π + δk) ≈−it δk − 0 (δk)4. (12) − 0 4γ periodic 10-7 open x=1 Now Eq.(7) becomes X = k,α=± λα(k)|uRkαihuLkα|, open x=10 and Eq.(8) reads P open x=20 -11 ∗ ′ 10 open x=30 λα(k)t+λ ′ (k )t ∆(˜ t)= e α |uRkαihuLkα|∆(0)˜ |uLk′α′ ihuRk′ α′ |. 20 40 t 60 kkX′ ,αα′ FIG. 3. (a) The particle number damping of a periodic bound- (13) ary chain ( curve) and open-boundary chains for several L For the initial state with translational symmetry, chain length . The long time damping of a periodic chain ˜ ˜ follows a power law, while the open boundary chain follows we have huLkα|∆(0)|uLk′α′ i = δkk′ huLkα|∆(0)|uLkα′ i. an exponential law after an initial power law stage. (b) The The long-time behavior of ∆(˜ t) is dominated by the site-resolved damping. The left end (x = 1) enters the expo- α = α′ = − sector, which provides a decay factor 2 nential stage from the very beginning, followed sequentially t0 4 exp (2Re[λ−(π + δk)]t) ≈ d(δk) exp[− (δk) t] ∼ by other sites. For both (a) and (b), the initial state is the δk 2γ † −1/4 ˜ −1/2 completely filled state Q c 0 , therefore, ∆(0) the identity tP . Similarly, for t1 t2 boundary chain are λ±(k + iκ), where λ± are the X(k) [Fig.2(b)]. The damping rate is therefore expected to be eigenvalues given in Eq.(11). We can readily check that, algebraic and exponential in each case, respectively. To for |γ| < 2|t1|, confirm this, we numerically calculate the site-averaged γ fermion number deviation from the steady state, defined λ±(k + iκ)= − ± iE(k), (14) 2 2 asn ˜(t) = x n˜x(t)/L, wheren ˜x(t) = nx(t) − nx(∞) with n (t) = n (t)+ n (t), n ≡ ∆ being the 2 2 x pP xA xB xs xs,xs 2 2 γ 2 γ fermion number at site xs. The results are consistent where E(k)= t1 + t2 − 4 +2t2 t1 − 4 cos k, which r q with the vanishing (nonzero) gap in the t1 ≤ t2 (t1 >t2) is real. We have also numerically diagonalized X for a case [Fig.2(a)]. long open chain [red dots in Fig.2(b)], which confirms Although our focus here is the damping dynamics, we Eq.(14). An immediate feature of Eq.(14) is that the real 4 part is a constant, −γ/2, which is consistent with the nu- merical spectrums [Fig.2(b))]. We note that the analytic results based on generalized Brillouin zone produce the continuum bands only, and the isolated topological edge modes [Fig.2(b), A and B panels] are not contained in Eq.(14), though they can be inferred from the non-Bloch bulk-boundary correspondence[24, 28]. Here, we focus on bulk dynamics, and these topological edge modes do not play important roles[107]. It follows from Eq.(14) that the Liouvillian gap Λ = γ, therefore, we expect an exponential long-time damping of ∆(˜ t). This exponential behavior has been confirmed by numerical simulation [Fig.3(a)]. Before entering the exponential stage, there is an initial period of algebraic damping, whose duration grows with chain length L [Fig.3(a)]. To better understand this feature, we plot the damping in each unit cell [Fig.3(b)]. We find that the left end (x = 1) enters the exponential damping im- mediately, and other sites enter the exponential stage sequentially, according to their distances to the left end. As such, there is a “damping wavefront” traveling from FIG. 4. Time evolution ofn ˜x(t) = nx(t) nx( ), which − ∞ the left (“upper reach”) to right (“lower reach”). This shows damping of particle number nx(t) towards the steady is dubbed a “chiral damping”, which can be intuitively state. (a)n ˜x(t) of the main model with dissipators given by Eq.(2) (referred to as “model I”). Left: periodic boundary; related to the fact that all eigenstates of X are localized Right: open boundary. The chiral damping is clearly seen in at the right end[24]. the open boundary case. The dark region corresponds to the More intuitively, the damping ofn ˜x(t)= nx(t)−nx(∞) exponential damping stage seen in Fig.3. (b)n ˜x(t) of model is shown in Fig.4(a). In the periodic boundary chain it II, whose damping matrix X [Eq.(15)] has no NHSE. The follows a slow power law. In the open boundary chain, Liouvillian gap is nonzero and the same for periodic and open a right-moving wavefront is seen. After the wavefront boundary chains. Common parameters: t1 = t2 = 1; γg = γl = 0.2. passes by x, the algebraically decayingn ˜x(t) enters the exponential decay stage and rapidly diminishes. The wavefront can be understood as follows. Ac- with the wavefront velocity (≈ 1) in Fig.4(a). cording to Eq.(6), the damping of ∆(˜ t) is determined As a comparison, we introduce the “model II” (the by the evolution under exp(Xt), which is just the model studied so far is referred to as “model I”) that evolution under exp(−γt/2) exp(−iH t), where H SSH SSH differs from model I only in Lg , which is now Lg = is the non-Hermitian SSH Hamiltonian[24] (with an x x γg † † unimportant sign difference). Now the propagator 2 (cxA − icxB) [compare it with Eq.(2)]. The damping ′ ′ q hxs| exp(−iHSSHt)|x s i can be decomposed as propaga- matrix is tion of various momentum modes with velocity v = k γl − γg γ ∂E/∂k. Due to the presence of an imaginary part κ in X(k)= i[(t1 + t2 cos k)σx + (t2 sin k − i )σy] − I, (15) 2 2 the momentum, propagation from x′ to x acquires an ′ exp[−κ(x − x )] factor. If this factor can compensate which has no NHSE when γl = γg. Accordingly, the open exp(−γt/2), exponential damping can be evaded, giving and periodic boundary chains have the same Liouvillian way to a power law damping. For simplicity we take γ gap, and chiral damping is absent [Fig.4(b)]. ′ small, so that κ ≈ −γ/2t1, therefore exp[−κ(x − x )] ≈ In realistic systems, there may be disorders, fluctu- ′ exp[vk(γ/2t1)t] and the damping of propagation from x ations of parameters, and other imperfections. Fortu- to x is exp[(−γ/2+ vkγ/2t1)t] for the k mode. By a nately, the main results here are based on the presence straightforward calculation, we have max(vk) = t2 (for of NHSE, which is a quite robust phenomenon unchanged 2 2 t1 > t2) or |t1 − γ /4| ≈ t1 (for t1 ≤ t2). Let us con- by modest imperfections. As such, it is expected that our sider t1 ≤ t2pfirst. When x> max(vk)t, the propagation main predictions are robust and observable. ′ from x = x − max(vk)t to x carries a factor exp[(−γ/2+ Final remarks.–(i) The chiral damping originates from max(vk)γ/2t1)t] = 1; while for x < max(vk)t, we need the NHSE of the damping matrix X rather than the ef- ′ the nonexistent x = x − max(vk)t < 0, therefore, com- fective non-Hermitian Hamiltonian. Unlike the damp- pensation is impossible and we have exponential damp- ing matrix, the effective non-Hermitian Hamiltonian de- ing. This indicates a wavefront at x = max(vk)t. For scribes short time evolution. It is found to be Heff = † t1 = t2 = 1, we have max(vk) ≈ 1, which is consistent ij ci (heff)ij cj − iγgL, where heff, written in momentum P 5

γl−γg space, is heff(k) = (t1 +t2 cos k)σx +(t2 sin k−i 2 )σy − [7] Bo Zhen, Chia Wei Hsu, Yuichi Igarashi, Ling Lu, γl−γg Ido Kaminer, Adi Pick, Song-Liang Chua, John D i 2 I. For γg = γl, heff has no NHSE, though X has. Although damping matrices with NHSE can arise quite Joannopoulos, and Marin Soljaˇci´c, “Spawning rings of exceptional points out of dirac cones,” Nature 525, 354 naturally (e.g., in Fig.1), none of the previous models (2015). (e.g., Ref.[108]) we have checked has NHSE. [8] Sebastian Diehl, Enrique Rico, Mikhail A Baranov, and (ii) The periodic-open contrast between the slow alge- Peter Zoller, “Topology by dissipation in atomic quan- braic and fast exponential damping has important im- tum wires,” Nature Physics 7, 971–977 (2011). plications for experimental preparation of steady states [9] Frank Verstraete, Michael M Wolf, and J Ignacio Cirac, (e.g. in cold atom systems). In the presence of NHSE, “Quantum computation and quantum-state engineering driven by dissipation,” Nature physics 5, 633 (2009). approaching the steady states in open-boundary systems [10] Simon Malzard, Charles Poli, and Henning can be much faster than estimations based on periodic Schomerus, “Topologically protected defect states boundary condition. in open photonic systems with non-hermitian (iii) It is interesting to investigate other rich as- charge-conjugation and parity-time symmetry,” pects of non-Hermitian physics such as PT symmetry Phys. Rev. Lett. 115, 200402 (2015). breaking[109] in this platform (Here, we have focused on [11] Jean Dalibard, Yvan Castin, and Klaus Mølmer, “Wave-function approach to dissipative processes in the cases that the open-boundary iX is essentially PT- quantum optics,” Phys. Rev. Lett. 68, 580–583 (1992). symmetric, meaning that the real parts of X eigenvalues [12] H. J. Carmichael, “Quantum trajec- are constant). tory theory for cascaded open systems,” (iv) When fermion-fermion interactions are included, Phys. Rev. Lett. 70, 2273–2276 (1993). higher-order correlation functions are coupled to the two- [13] James Anglin, “Cold, dilute, trapped bosons as an open quantum system,” Phys. Rev. Lett. 79, 6–9 (1997). point ones, and approximations (such as truncations) [14] Youngwoon Choi, Sungsam Kang, Sooin Lim, are called for. Moreover, the steady states can be Wookrae Kim, Jung-Ryul Kim, Jai-Hyung Lee, multiple[110, 111], in which case the damping matrix de- and Kyungwon An, “Quasieigenstate coales- pends on the steady state approached, leading to even cence in an atom-cavity quantum composite,” richer chiral damping behaviors. These possibilities will Phys. Rev. Lett. 104, 153601 (2010). be left for future investigations. [15] Sebastian Diehl, A Micheli, A Kantian, B Kraus, HP B¨uchler, and P Zoller, “Quantum states and phases This work is supported by NSFC under grant No. in driven open quantum systems with cold atoms,” Na- 11674189. ture Physics 4, 878 (2008). [16] CE Bardyn, MA Baranov, CV Kraus, E Rico, A Imamo˘glu,˙ P Zoller, and S Diehl, “Topology by dis- sipation,” New Journal of Physics 15, 085001 (2013). [17] V. Kozii and L. Fu, “Non-Hermitian Topological The- ∗ [email protected] ory of Finite-Lifetime Quasiparticles: Prediction of Bulk [1] Liang Feng, Ramy El-Ganainy, and Li Ge, “Non- Fermi Arc Due to Exceptional Point,” ArXiv e-prints hermitian photonics based on parity–time symmetry,” (2017), arXiv:1708.05841 [cond-mat.mes-hall]. Nature Photonics 11, 752 (2017). [18] Michal Papaj, Hiroki Isobe, and Liang Fu, “Nodal arc [2] Ramy El-Ganainy, Konstantinos G Makris, Mercedeh of disordered dirac fermions and non-hermitian band Khajavikhan, Ziad H Musslimani, Stefan Rotter, and theory,” Phys. Rev. B 99, 201107 (2019). Demetrios N Christodoulides, “Non-hermitian physics [19] Huitao Shen and Liang Fu, “Quantum oscillation from and pt symmetry,” Nature Physics 14, 11 (2018). in-gap states and a non-hermitian landau level prob- [3] Tomoki Ozawa, Hannah M. Price, Alberto Amo, lem,” Phys. Rev. Lett. 121, 026403 (2018). Nathan Goldman, Mohammad Hafezi, Ling Lu, [20] Hengyun Zhou, Chao Peng, Yoseob Yoon, Chia Wei Mikael C. Rechtsman, David Schuster, Jonathan Simon, Hsu, Keith A. Nelson, Liang Fu, John D. Joannopou- Oded Zilberberg, and Iacopo Carusotto, “Topological los, Marin Soljaˇci´c, and Bo Zhen, “Observation of bulk photonics,” Rev. Mod. Phys. 91, 015006 (2019). fermi arc and polarization half charge from paired ex- [4] B Peng, S¸K Ozdemir,¨ S Rotter, H Yilmaz, M Liertzer, ceptional points,” (2018), 10.1126/science.aap9859. F Monifi, CM Bender, F Nori, and L Yang, “Loss- [21] Tsuneya Yoshida, Robert Peters, and No- induced suppression and revival of lasing,” Science 346, rio Kawakami, “Non-hermitian perspective of 328–332 (2014). the band structure in heavy-fermion systems,” [5] Liang Feng, Ye-Long Xu, William S Fegadolli, Ming- Phys. Rev. B 98, 035141 (2018). Hui Lu, Jos´eEB Oliveira, Vilson R Almeida, Yan-Feng [22] Huitao Shen, Bo Zhen, and Liang Fu, “Topolog- Chen, and Axel Scherer, “Experimental demonstration ical band theory for non-hermitian hamiltonians,” of a unidirectional reflectionless parity-time metamate- Phys. Rev. Lett. 120, 146402 (2018). rial at optical frequencies,” Nature materials 12, 108 [23] Tony E. Lee, “Anomalous edge state in a non-hermitian (2013). lattice,” Phys. Rev. Lett. 116, 133903 (2016). [6] Ingrid Rotter, “A non-hermitian hamilton operator and [24] Shunyu Yao and Zhong Wang, “Edge states and the physics of open quantum systems,” Journal of topological invariants of non-hermitian systems,” Physics A: Mathematical and Theoretical 42, 153001 Phys. Rev. Lett. 121, 086803 (2018). (2009). 6

[25] Shunyu Yao, Fei Song, and Zhong [42] Hui Jiang, Chao Yang, and Shu Chen, “Topological Wang, “Non-hermitian chern bands,” invariants and phase diagrams for one-dimensional two- Phys. Rev. Lett. 121, 136802 (2018). band non-hermitian systems without chiral symmetry,” [26] V. M. Martinez Alvarez, J. E. Barrios Var- Phys. Rev. A 98, 052116 (2018). gas, M. Berdakin, and L. E. F. Foa Torres, [43] Gal Harari, Miguel A Bandres, Yaakov Lumer, Mikael C “Topological states of non-hermitian systems,” Rechtsman, Yi Dong Chong, Mercedeh Khajavikhan, The European Physical Journal Special Topics 227, 1295–1308 (2018).Demetrios N Christodoulides, and Mordechai Segev, [27] Daniel Leykam, Konstantin Y. Bliokh, Chunli Huang, “Topological laser: theory,” Science 359, Y. D. Chong, and Franco Nori, “Edge modes, degen- eaar4003 (2018). eracies, and topological numbers in non-hermitian sys- [44] Cem Yuce, “Topological phase in a non-hermitian pt tems,” Phys. Rev. Lett. 118, 040401 (2017). symmetric system,” Physics Letters A 379, 1213–1218 [28] Flore K. Kunst, Elisabet Edvardsson, Jan Carl Bu- (2015). dich, and Emil J. Bergholtz, “Biorthogonal bulk- [45] Baogang Zhu, Rong L¨u, and Shu Chen, “ boundary correspondence in non-hermitian systems,” symmetry in the non-hermitian su-schrieffer-PT Phys. Rev. Lett. 121, 026808 (2018). heeger model with complex boundary potentials,” [29] Ye Xiong, “Why does bulk boundary correspondence Phys. Rev. A 89, 062102 (2014). fail in some non-hermitian topological models,” Journal [46] Simon Lieu, “Topological symmetry classes of Physics Communications 2, 035043 (2018). for non-hermitian models and connections to [30] V. M. Martinez Alvarez, J. E. Barrios Vargas, and the bosonic bogoliubov–de gennes equation,” L. E. F. Foa Torres, “Non-hermitian robust edge Phys. Rev. B 98, 115135 (2018). states in one dimension: Anomalous localization [47] C. Yuce, “Majorana edge modes with gain and loss,” and eigenspace condensation at exceptional points,” Phys. Rev. A 93, 062130 (2016). Phys. Rev. B 97, 121401 (2018). [48] Henri Menke and Moritz M. Hirschmann, “Topolog- [31] Kazuki Yokomizo and Shuichi Murakami, “Non- ical quantum wires with balanced gain and loss,” bloch band theory of non-hermitian systems,” Phys. Rev. B 95, 174506 (2017). Phys. Rev. Lett. 123, 066404 (2019). [49] C Wang and XR Wang, “Non-quantized edge chan- [32] L. Jin and Z. Song, “Bulk-boundary correspondence in nel conductance and zero conductance fluctuation a non-hermitian system in one dimension with chiral in non-hermitian chern insulators,” arXiv preprint inversion symmetry,” Phys. Rev. B 99, 081103 (2019). arXiv:1901.06982 (2019). [33] Heinrich-Gregor Zirnstein, Gil Refael, and Bernd [50] Timothy M. Philip, Mark R. Hirsbrunner, and Rosenow, “Bulk-boundary correspondence for Matthew J. Gilbert, “Loss of hall conductivity quan- non-Hermitian Hamiltonians via Green func- tization in a non-hermitian quantum anomalous hall in- tions,” arXiv e-prints , arXiv:1901.11241 (2019), sulator,” Phys. Rev. B 98, 155430 (2018). arXiv:1901.11241 [cond-mat.mes-hall]. [51] Yu Chen and Hui Zhai, “Hall conductance of a non- [34] Lo¨ıc Herviou, Jens H. Bardarson, and Nicolas Reg- hermitian chern insulator,” Physical Review B 98, nault, “Defining a bulk-edge correspondence for non- 245130 (2018). hermitian hamiltonians via singular-value decomposi- [52] Mark R. Hirsbrunner, Timothy M. Philip, and tion,” Phys. Rev. A 99, 052118 (2019). Matthew J. Gilbert, “Topology and observ- [35] Kenta Esaki, Masatoshi Sato, Kazuki Hasebe, ables of the non-hermitian chern insulator,” and Mahito Kohmoto, “Edge states and topo- Phys. Rev. B 100, 081104 (2019). logical phases in non-hermitian systems,” [53] Marcel Klett, Holger Cartarius, Dennis Dast, J¨org Phys. Rev. B 84, 205128 (2011). Main, and G¨unter Wunner, “Relation between [36] Zongping Gong, Yuto Ashida, Kohei Kawabata, -symmetry breaking and topologically nontrivial Kazuaki Takasan, Sho Higashikawa, and Masahito phasesPT in the su-schrieffer-heeger and kitaev models,” Ueda, “Topological phases of non-hermitian systems,” Phys. Rev. A 95, 053626 (2017). Phys. Rev. X 8, 031079 (2018). [54] Qi-Bo Zeng, Yan-Bin Yang, and Yong Xu, [37] Tao Liu, Yu-Ran Zhang, Qing Ai, Zongping Gong, “Topological Non-Hermitian Quasicrystals,” Kohei Kawabata, Masahito Ueda, and Franco Nori, arXiv e-prints , arXiv:1901.08060 (2019), “Second-order topological phases in non-hermitian sys- arXiv:1901.08060 [cond-mat.mes-hall]. tems,” Phys. Rev. Lett. 122, 076801 (2019). [55] Longwen Zhou and Jiangbin Gong, “Non- [38] Simon Lieu, “Topological phases in the hermitian floquet topological phases with ar- non-hermitian su-schrieffer-heeger model,” bitrarily many real-quasienergy edge states,” Phys. Rev. B 97, 045106 (2018). Phys. Rev. B 98, 205417 (2018). [39] Chuanhao Yin, Hui Jiang, Linhu Li, Rong L¨u, and [56] Kohei Kawabata, Yuto Ashida, Hosho Katsura, and Shu Chen, “Geometrical meaning of winding num- Masahito Ueda, “Parity-time-symmetric topological su- ber and its characterization of topological phases perconductor,” Phys. Rev. B 98, 085116 (2018). in one-dimensional chiral non-hermitian systems,” [57] Yong Xu, Sheng-Tao Wang, and L.-M. Duan, “Weyl Phys. Rev. A 97, 052115 (2018). exceptional rings in a three-dimensional dissipative cold [40] Tian-Shu Deng and Wei Yi, “Non-bloch topological atomic gas,” Phys. Rev. Lett. 118, 045701 (2017). invariants in a non-hermitian domain wall system,” [58] Huaiqiang Wang, Jiawei Ruan, and Haijun Phys. Rev. B 100, 035102 (2019). Zhang, “Non-hermitian nodal-line semimetals with [41] Ananya Ghatak and Tanmoy Das, “New topological in- an anomalous bulk-boundary correspondence,” variants in non-hermitian systems,” Journal of Physics: Phys. Rev. B 99, 075130 (2019). Condensed Matter 31, 263001 (2019). [59] Alexander Cerjan, Meng Xiao, Luqi Yuan, and Shan- 7

hui Fan, “Effects of non-hermitian perturbations on Phys. Rev. B 98, 085126 (2018). weyl hamiltonians with arbitrary topological charges,” [77] W. B. Rui, Y. X. Zhao, and Andreas P. Schny- Phys. Rev. B 97, 075128 (2018). der, “Topology and exceptional points of massive [60] Jan Carl Budich, Johan Carlstr¨om, Flore K. dirac models with generic non-hermitian perturba- Kunst, and Emil J. Bergholtz, “Symmetry- tions,” Phys. Rev. B 99, 241110 (2019). protected nodal phases in non-hermitian systems,” [78] Konstantin Y. Bliokh and Franco Nori, “Klein- Phys. Rev. B 99, 041406 (2019). gordon representation of acoustic waves and [61] Zhesen Yang and Jiangping Hu, “Non- topological origin of surface acoustic modes,” hermitian hopf-link exceptional line semimetals,” Phys. Rev. Lett. 123, 054301 (2019). Phys. Rev. B 99, 081102 (2019). [79] Xi-Wang Luo and Chuanwei Zhang, “Higher-order [62] Johan Carlstr¨om and Emil J. Bergholtz, “Exceptional topological corner states induced by gain and loss,” links and twisted fermi ribbons in non-hermitian sys- Phys. Rev. Lett. 123, 073601 (2019). tems,” Phys. Rev. A 98, 042114 (2018). [80] Z. Ozcakmakli Turker and C. Yuce, “Open and [63] Ryo Okugawa and Takehito Yokoyama, “Topolog- closed boundaries in non-hermitian topological sys- ical exceptional surfaces in non-hermitian systems tems,” Phys. Rev. A 99, 022127 (2019). with parity-time and parity-particle-hole symmetries,” [81] Naomichi Hatano and David R. Nelson, “Localiza- Phys. Rev. B 99, 041202 (2019). tion transitions in non-hermitian quantum mechanics,” [64] Kristof Moors, Alexander A. Zyuzin, Alexan- Phys. Rev. Lett. 77, 570–573 (1996). der Yu. Zyuzin, Rakesh P. Tiwari, and Thomas L. [82] Julia M. Zeuner, Mikael C. Rechtsman, Yonatan Schmidt, “Disorder-driven exceptional lines and Plotnik, Yaakov Lumer, Stefan Nolte, Mark S. fermi ribbons in tilted nodal-line semimetals,” Rudner, Mordechai Segev, and Alexander Phys. Rev. B 99, 041116 (2019). Szameit, “Observation of a topological tran- [65] A. A. Zyuzin and A. Yu. Zyuzin, “Flat band sition in the bulk of a non-hermitian system,” in disorder-driven non-hermitian weyl semimetals,” Phys. Rev. Lett. 115, 040402 (2015). Phys. Rev. B 97, 041203 (2018). [83] L. Xiao, X. Zhan, Z. H. Bian, K. K. Wang, X. Zhang, [66] Tsuneya Yoshida, Robert Peters, Norio Kawakami, and X. P. Wang, J. Li, K. Mochizuki, D. Kim, N. Kawakami, Yasuhiro Hatsugai, “Symmetry-protected exceptional Wei Yi, H. Obuse, B. C. Sanders, and Peng Xue, rings in two-dimensional correlated systems with chiral “Observation of topological edge states in parity–time- symmetry,” Phys. Rev. B 99, 121101 (2019). symmetric quantum walks,” Nature Physics 13, 1117 [67] Hengyun Zhou, Jong Yeon Lee, Shang Liu, and (2017). Bo Zhen, “Exceptional surfaces in pt-symmetric non- [84] Charles Poli, Matthieu Bellec, Ulrich Kuhl, Fabrice hermitian photonic systems,” Optica 6, 190–193 (2019). Mortessagne, and Henning Schomerus, “Selective en- [68] Kohei Kawabata, Sho Higashikawa, Zongping hancement of topologically induced interface states in a Gong, Yuto Ashida, and Masahito Ueda, “Topo- dielectric resonator chain,” Nature communications 6, logical unification of time-reversal and particle- 6710 (2015). hole symmetries in non-Hermitian physics,” [85] S Weimann, M Kremer, Y Plotnik, Y Lumer, S Nolte, Nature Communications 10, 297 (2019). KG Makris, M Segev, MC Rechtsman, and A Sza- [69] Hengyun Zhou and Jong Yeon Lee, “Periodic table meit, “Topologically protected bound states in photonic for topological bands with non-hermitian symmetries,” parity–time-symmetric crystals,” Nature materials 16, Phys. Rev. B 99, 235112 (2019). 433 (2017). [70] Kohei Kawabata, Ken Shiozaki, Masahito Ueda, and [86] Alexander Cerjan, Sheng Huang, Mohan Wang, Kevin P Masatoshi Sato, “Symmetry and Topology in Non- Chen, Yidong Chong, and Mikael C Rechtsman, “Ex- Hermitian Physics,” arXiv e-prints , arXiv:1812.09133 perimental realization of a weyl exceptional ring,” Na- (2018), arXiv:1812.09133 [cond-mat.mes-hall]. ture Photonics , 1 (2019). [71] M. S. Rudner and L. S. Levitov, “Topologi- [87] Xiang Zhan, Lei Xiao, Zhihao Bian, Kunkun cal transition in a non-hermitian quantum walk,” Wang, Xingze Qiu, Barry C. Sanders, Wei Yi, Phys. Rev. Lett. 102, 065703 (2009). and Peng Xue, “Detecting topological invari- [72] A. McDonald, T. Pereg-Barnea, and A. A. Clerk, ants in nonunitary discrete-time quantum walks,” “Phase-dependent chiral transport and effective non- Phys. Rev. Lett. 119, 130501 (2017). hermitian dynamics in a bosonic kitaev-majorana [88] Hui Jiang, Li-Jun Lang, Chao Yang, Shi-Liang Zhu, chain,” Phys. Rev. X 8, 041031 (2018). and Shu Chen, “Interplay of non-hermitian skin effects [73] Wenchao Hu, Hailong Wang, Perry Ping Shum, and and anderson localization in nonreciprocal quasiperiodic Y. D. Chong, “Exceptional points in a non-hermitian lattices,” Phys. Rev. B 100, 054301 (2019). topological pump,” Phys. Rev. B 95, 184306 (2017). [89] Ching Hua Lee and Ronny Thomale, “Anatomy of [74] M´ario G. Silveirinha, “Topological the- skin modes and topology in non-hermitian systems,” ory of non-hermitian photonic systems,” Phys. Rev. B 99, 201103 (2019). Phys. Rev. B 99, 125155 (2019). [90] Ching Hua Lee, Linhu Li, and Jiangbin Gong, “Hy- [75] Kohei Kawabata, Takumi Bessho, and Masatoshi brid higher-order skin-topological modes in nonrecipro- Sato, “Classification of exceptional points cal systems,” Phys. Rev. Lett. 123, 016805 (2019). and non-hermitian topological semimetals,” [91] Flore K. Kunst and Vatsal Dwivedi, “Non-hermitian Phys. Rev. Lett. 123, 066405 (2019). systems and topology: A transfer-matrix perspective,” [76] Jos´e A. S. Louren¸co, Ronivon L. Eneias, Phys. Rev. B 99, 245116 (2019). and Rodrigo G. Pereira, “Kondo effect in a [92] Elisabet Edvardsson, Flore K. Kunst, and -symmetric non-hermitian hamiltonian,” Emil J. Bergholtz, “Non-hermitian extensions PT 8

of higher-order topological phases and their [109] Bo Peng, S¸ahin Kaya Ozdemir,¨ Fuchuan Lei, Faraz biorthogonal bulk-boundary correspondence,” Monifi, Mariagiovanna Gianfreda, Gui Lu Long, Shan- Phys. Rev. B 99, 081302 (2019). hui Fan, Franco Nori, Carl M Bender, and Lan Yang, [93] Dan S. Borgnia, Alex Jura Kruchkov, and “Parity–time-symmetric whispering-gallery microcavi- Robert-Jan Slager, “Non-Hermitian Boundary ties,” Nature Physics 10, 394 (2014). Modes,” arXiv e-prints , arXiv:1902.07217 (2019), [110] Victor V. Albert, “Lindbladians with multiple steady arXiv:1902.07217 [cond-mat.mes-hall]. states: theory and applications,” arXiv e-prints , [94] B. X. Wang and C. Y. Zhao, “Topological arXiv:1802.00010 (2018), arXiv:1802.00010 [quant-ph]. in one-dimensional non- [111] Leo Zhou, Soonwon Choi, and Mikhail D. Lukin, hermitian silicon carbide nanoparticle chains,” “Symmetry-protected dissipative preparation of ma- Phys. Rev. B 98, 165435 (2018). trix product states,” arXiv e-prints , arXiv:1706.01995 [95] Motohiko Ezawa, “Non-hermitian boundary (2017), arXiv:1706.01995 [quant-ph]. and interface states in nonreciprocal higher- [112] Moos van Caspel, Sergio Enrique Tapias Arze, and order topological metals and electrical circuits,” Isaac Prez Castillo, “Dynamical signatures of topo- Phys. Rev. B 99, 121411 (2019). logical order in the driven-dissipative Kitaev chain,” [96] Motohiko Ezawa, “Non-hermitian higher-order SciPost Phys. 6, 26 (2019). topological states in nonreciprocal and recipro- [113] Michael J. Kastoryano and Mark S. Rudner, “Topolog- cal systems with their electric-circuit realization,” ical transport in the steady state of a quantum particle Phys. Rev. B 99, 201411 (2019). with dissipation,” Phys. Rev. B 99, 125118 (2019). [97] Motohiko Ezawa, “Braiding of majorana-like corner states in electric circuits and its non-hermitian gener- alization,” Phys. Rev. B 100, 045407 (2019). [98] Xiaosen Yang, Yang Cao, and Yunjia Zhai, “Non- hermitian weyl semimetals: Non-hermitian skin effect and non-bloch bulk-boundary correspondence,” arXiv preprint arXiv:1904.02492 (2019). DERIVATION OF THE DIFFERENTIAL [99] Zi-Yong Ge, Yu-Ran Zhang, Tao Liu, Si-Wen Li, EQUATION OF CORRELATION FUNCTIONS Heng Fan, and Franco Nori, “Topological band the- ory for non-hermitian systems from the dirac equation,” Phys. Rev. B 100, 054105 (2019). We shall derive Eq.(3) in the main article, which is [100] Tobias Helbig, Tobias Hofmann, Stefan Imhof, Mo- reproduced as follows: hamed Abdelghany, Tobias Kiessling, Laurens W. Molenkamp, Ching Hua Lee, Alexander Szameit, Mar- tin Greiter, and Ronny Thomale, “Observation of d∆(t) = i[hT , ∆(t)] −{M T + M , ∆(t)} +2M .(16) bulk boundary correspondence breakdown in topolectri- dt l g g cal circuits,” arXiv e-prints , arXiv:1907.11562 (2019), arXiv:1907.11562 [cond-mat.mes-hall]. [101] Lei Xiao, Tianshu Deng, Kunkun Wang, Gaoyan Zhu, In fact, after inserting the Lindblad master equation † Zhong Wang, Wei Yi, and Peng Xue, “Observa- into d∆ij /dt = Tr[ci cj dρ/dt] and reorganize the terms, tion of non-Hermitian bulk-boundary correspondence in we have ,” arXiv e-prints , arXiv:1907.12566 (2019), arXiv:1907.12566 [cond-mat.mes-hall]. d∆ij (t) † [102] Ananya Ghatak, Martin Brandenbourger, Jasper van = iTr [H,ci cj ]ρ(t) + Wezel, and Corentin Coulais, “Observation of non- dt   Hermitian topology and its bulk-edge correspon- † † † † Tr 2Lµ[ci cj,Lµ] + [LµLµ,ci cj ] ρ(t(17)) . dence,” arXiv e-prints , arXiv:1907.11619 (2019), Xµ h  i arXiv:1907.11619 [cond-mat.mes-hall]. [103] Masaya Nakagawa, Norio Kawakami, and Masahito Ueda, “Non-hermitian kondo effect in ultracold alkaline- By a straightforward calculation, we have earth atoms,” Phys. Rev. Lett. 121, 203001 (2018). [104] Toma ˇz Prosen and Enej Ilievski, “Nonequilibrium † † † in a periodically driven xy spin chain,” [H,ci cj ]= hmn[cmcn,ci cj ] Phys. Rev. Lett. 107, 060403 (2011). Xmn [105] As a non-Hermitian matrix, X can have exceptional † † = hmn(−δmjci cn + δincmcj ) points, and we have checked that our main results are Xmn qualitatively similar therein. † † [106] Zi Cai and Thomas Barthel, “Algebraic versus exponen- = (−hjnci cn + hnicncj ), (18) tial decoherence in dissipative many-particle systems,” Xn Phys. Rev. Lett. 111, 150403 (2013). [107] Topological edge modes have been investigated recently therefore, the Hamiltonian commutator term in Eq.(17) in Ref. [112, 113] in models without NHSE. is reduced to i[hT , ∆(t)] , which is the first term of [108] Yuto Ashida and Masahito Ueda, “Full-counting many- ij particle dynamics: Nonlocal and chiral propagation of Eq.(16). l correlations,” Phys. Rev. Lett. 120, 185301 (2018). The commutators terms from the loss dissipators Lµ = 9

l i Dµici are The corresponding terms in Eq.(17) sum to T P −{Ml , ∆(t)}ij , l† † l l∗ † † l 2 Lµ [ci cj ,Lµ]=2 Dµmcm[ci cj , Dµncn] Similarly, for the commutators from the gain dissipa- Xµ Xµm Xn tors, we have l∗ l † † =2 DµmDµncm[ci cj ,cn] µmnX 2 Lg†[c†c ,Lg ]=2 (M ) c c†, l∗ l † µ i j µ g mj m i (21) =2 DµmDµn(−δincmcj ) Xµ Xm µmnX † = −2 (Ml)micmcj , (19) and Xm and g† g † † † [Lµ Lµ,ci cj ]= −(Mg)nj cnci + (Mg)incj cn . l† l † l∗ l † † Xµ Xn   [Lµ Lµ,ci cj ]= DµmDµn[cmcn,ci cj ] Xµ µmnX (22) † † l∗ l † † Writing cnci = δni − ci cn, we see that the corresponding = DµmDµn(−δmj ci cn + δincmcj ) terms in Eq.(17) sum to 2(Mg)ij Tr(ρ) −{Mg, ∆(t)}ij = µmnX 2(Mg)ij −{Mg, ∆(t)}ij . Therefore, all terms at the right † † = −(Ml)jnci cn + (Ml)nicncj (20). hand side of Eq.(17) sum to that of Eq.(16). Xn