Fate of measurement-induced phase transition in long-range interactions

Takaaki Minato1, Koudai Sugimoto1, Tomotaka Kuwahara2, and Keiji Saito1 1Department of Physics, Keio University, Hiyoshi, Kohoku-ku, Yokohama 223-8522, Japan and 2 Mathematical Science Team, RIKEN Center for Advanced Intelligence Project (AIP), 1-4-1 Nihonbashi, Chuo-ku, Tokyo 103-0027, Japan (Dated: September 28, 2021) We consider quantum many-body dynamics under quantum measurements, where the so-called measurement-induced phase transitions (MIPs) occur, which depend on the frequency of the mea- surement. In this work, we consider the robustness of the MIP for long-range interaction that decays as r−α with distance r. The effects of long-range interactions are classified into two regimes: i) for α > αc, the MIP is observed. ii) for α < αc, the MIP is absent, even for arbitrarily strong measure- ments. Using fermion models, we demonstrate both regimes in integrable and non-integrable cases. We identify the underlying mechanism and propose sufficient conditions to observe the MIP, that is, α > d + 1/2 for general bilinear systems and α > d + 1 for general non-integrable systems (d: spatial dimension). Numerical calculation indicates that these conditions are optimal.

Introduction .— Understanding the general properties ↵c and finding new phenomena regarding the time evolution Area law phase of quantum entanglement in quantum many-body sys- tems is a critical subject in physics. Recently, novel dy- No namic phase transitions in quantum entanglement have MIP (↵) been discovered in the presence of quantum measure- c ments [1–22]. In general, the bipartite entanglement en- tropy of isolated systems grows over time and eventually reaches the order of the system size. Conversely, pro- (Sub)Volume law phase jective quantum measurements suppress entanglement ↵ growth, such as in the quantum Zeno effect under contin- uous measurement [23]. As a result of this competition, with increasing measurement amplitude (frequency) in non-integrable systems, the bipartite entanglement en- tropy in the long-time limit shows a transition from the ↵ order of the system size (the volume law phase) to the r order of the boundary area (the area law phase), which is called the measurement-induced phase transition (MIP). FIG. 1. Schematic of findings. We consider fermion systems with long-range interactions, where each site is constantly MIPs have been intensively studied in various systems, measured by the amplitude (frequency) γ. The function γc(α) such as quantum circuit models [1–15], cold atomic sys- is the critical measurement amplitude (see Fig.3 (b) for the tems [16–19], and quantum spin systems [21, 22]. We free fermion case). In the regime α < αc, the MIP does not emphasize that the MIP generally occurs irrespective exist. of integrability/non-integrability. It has been recently found that free-fermion systems also show the MIP, a transition between the phase of the entanglement entropy 32], and dynamic properties [33–39]. Thus, it is natu- with the order of the logarithmic system size (the sub- ral to ask whether long-range interactions influence the arXiv:2104.09118v2 [quant-ph] 27 Sep 2021 volume law phase) and the area law phase [19]. The MIP physics of MIP. is believed to be a ubiquitous phenomenon in isolated The long-range interaction immediately propagates the many-body quantum systems. quantum information to particles with arbitrary dis- In this paper, we consider the MIP in long-range in- tances; hence, the entanglement growth should be en- teracting systems to understand the mechanism more hanced. From this viewpoint, one anticipates nontrivial deeply. Here, long-range interaction means that the am- competition between quantum measurement and long- α plitude of interaction decays as r− , where r is the dis- range interaction strength. To explore this, we address tance between particles (see also [24]) As conceptually the following questions: (i) is there a limitation on the established in statistical mechanics [25], phase transi- existence of the MIP ?; and (ii) what are the conditions tions generally depend on the interaction range, dimen- for the existence of the MIP ? sionality, and types of interactions. Various studies have The primary obstacle to addressing these questions lies demonstrated that the physics can qualitatively change in the fact that the dynamics under quantum measure- under long-range interactions, including the static prop- ments are highly nonlinear [23], which makes the anal- erties of the equilibrium phase [26–28], ground state [29– yses difficult even numerically. To overcome this diffi- 2 culty, simple Clifford circuit models have been employed α α in many studies [1,2,4–6, 12–15]. Hence, in this pa- γ γ per, as a simple toy model to grasp the essence, we start γ γ with a simple fermion model to address question (i). We γ γ γ γ then identify the physical mechanism to make a state- S γ γ ment applicable to generic systems addressing question γ γ (ii), where sufficient conditions to observe the MIP in generic systems are proposed. Then, we obtain several physical pictures, as summarized schematically in Fig.1. Model.— To obtain the essential physics of the effect sin( ) sin( ) of long-range interaction in the MIP, we consider the fol- lowing simple long-range Hamiltonian: α α

L/2 L h i L L X X 1 L L H = cj†+rcj cj†cj+r + V nj+rnj , (1) rα − − L L =1 =1 j r ( γ,α ) I “a” L † 3L 8 3L where cj and c are the annihilation and creation op- “b” 8 “d” j 8 L 8 erators of the spinless fermion at site j, and n = c†c , j j j “c” γc respectively. The parameter α is the degree of long-range interaction. We impose the periodic boundary condition γ γ for the total system size L, that is, cj+L = cj. We set ¯ the Neel state to the initial state, that is, ψ(t = 0) = FIG. 2. (a) and (b): S` as a function of sin(π`/L)(L = 200) QL/2 for α = 1.8 in (a) and for α = 0.8 in (b), where the x-axis is log i=1 c2†i 1 vac , where vac is the vacuum state. Note − | i | i scale. The CFT behavior can be observed around ` ∼ L/2 as that the total number of fermions is fixed at L/2 at all indicated in (a), where the solid lines are guides for the eyes. times. We perform a quantum measurement uniformly (c) and (d): The mutual information as a function of γ. for all sites with a finite measurement amplitude (fre- quency) γ. Then, the time evolution of the wave func- tion is described by the standard quantum jump process system (hence, the regions “a” and “d” contact with [18, 23], that is, each other on the ring geometry). We then consider d ψ(t) = iH ψ(t) dt the mutual information between the region “a” and “c”, | i − | i I(γ, α) = S + S S , where S , S and S are L " # a c − ac a c ac X cj†cj ψ(t) the entanglement entropy for the regions “a”, “c”, and + | i ψ(t) dw (t) , (2) p − | i j “a”+“c”, respectively. We denote the average values of j=1 ψ(t) nj ψ(t) h | | mutual information by I¯(γ, α). where dwj(t) takes 0 or 1 obeying the Poisson process, Free fermion case.— We first consider the free fermion that is, dwj(t) = γdt, and dwj(t)dwj0 (t0) = 0 for case, that is, V = 0. The wave function can be expressed hh ii hh ii L/2 L j = j0. Here, ... is the noise average. As a result of Q P 6 hh ii in the form of ψ(t) = m=1 j=1 uj,m(t) cj† vac . the measurement process, there are many trajectories of The many-body wave| i function is expressed through| thei the wave functions starting from the fixed initial state. PL function uj,m(t), where the relation j=1 uj,m∗ uj,m0 = Hence, we need to take an average over many trajectories δm,m0 is imposed to guarantee normalization. The to examine the statistical properties of any observables. correlation function is calculated using the relation The main physical quantity we address is entanglement PL/2 ψ(t) ci†cj ψ(t) = m=1 uj,m(t)ui,m∗ (t). In addition, the entropy. Let us divide the system into two subsystems A hentanglement| | entropyi can be computed once the correla- and B, which have sizes ` and L ` (` L/2), respec- − ≤ tion functions are obtained [40]. See the supplementary tively. Then, the entanglement entropy is defined as: material (SM) for these established methods [41].

S` := Tr(ρA log ρA). (3) In Fig.2 (a) and (b), we show the numerical data − for the entanglement entropy, which is a long-time av- where ρA is the reduced density matrix of subsystem A erage starting from the Neel state. See footnote [42] for for a given wave function ψ , that is, ρ = Tr ( ψ ψ ), the numerical details used to obtain the data. These | i A B | ih | where TrB is the partial trace with respect to part B. figures show the `-dependence of the entanglement en- We discuss the trajectory average S¯`, employing a suffi- tropy for α = 1.8 and α = 0.8, respectively, for a cient number of trajectories. We also compute the mu- fixed length L = 200. As indicated in Fig.2 (a) with tual information as another indicator to detect the MIP. solid lines, the entanglement entropy around ` L/2 is To this end, we divide the total system into four subsys- well fitted by the functional form of the conformal∼ field tems in the order of “a”, “b”, “c” and “d” along the theory (CFT) with the effective central charge c(γ, α): 3

p L exp( ν/ γ γ (α)). Here, γ (α) is the crossing point, (a) − − c c

) 0.4 L which depends on α. In Fig.3 (a), we verified that the α L scaling ansatz. These scaling data strongly suggest that ( γ , γc can be regarded as critical points. In Fig.3 (b), we

) γI 0.2 show the behavior of γc(α) as a function of α. In the

g ( L α α α α gray shaded area (α < 1.5), we can find no critical points ∞ (that is, no crossing point, as shown in Fig.2 (d) [41]). 0.0 Thus, we find that the critical value α 1.5 separates 0 0 0 0 c ' log(L) ν/√γ γc the absence and existence of the MIP (see Fig.1). − − Analysis based on the entanglement growth rate.— 1.0 (b) Next, we discuss the underlying physical mechanism for the numerical findings in Fig.3 (b). We argue that the key component is the growth rate of the entanglement en- α )

( 0.5 tropy in pure quantum dynamics without measurement. c

γ Note the following expression for the entropy growth rate under pure quantum dynamics:

αc 0.0 S˙ = H λ(ρ) , ` k ABk (4) ∞ λ(ρ) := Tr (h [ρ, ρ 1 ]) , α AB A ⊗ B where ρ = ψ(t) ψ(t) , h := H / H ( ... is the FIG. 3. (a): Finite-size scaling for the mutual in- | ih | AB AB k ABk k k formation with the BKT scenario γg(L)I¯(γ, α) versus operator norm), and 1B is the identity operator for sub- p log(Lξ(γ, γc(α))) = log L − ν/ γ − γc(α), where ν = system B. The Hamiltonian HAB denotes the boundary 6.0, 5.5, 4.5, and 3.0 for α = 1.6, 1.8, 2.3, and α = ∞, re- interaction between subsystems A and B: spectively and g(L) = [1 + 1/(2 log L − 4)]−1. (b): Critical X X amplitude γc(α) as a function of α. No critical amplitude H = h , (5) exists for α < 1.5. AB i,j i A j B ∈ ∈

where hi,j is an interaction operator acting on sites i and ¯ S` = (c(γ, α)/3) log2 [(L/π) sin(π`/L)] + const . For the j. Of interest is the case where ` = L/2. While the func- short-range interaction limit α , this behavior was tion λ(ρ) highly depends on the states [47–49], the value → ∞ reported in Ref. [19]. In Fig.2 (a), the entanglement en- is finite for less entangled states (it is zero, especially for tropies becomes constants for large γ, implying the area a decoupled state such as the Neel state). Suppose that law, while Fig.2 (a) has no indication of the area law. HAB is finite. Then, a sufficiently large measurement Figs.2 (c) and (d) show the mutual information I¯(γ, α) amplitudek k can suppress the entanglement growth, and between the regime “a” and “c” depicted schematically in hence the MIP should occur. Conversely, when HAB Fig.2 (c), where the regions “ a” and “c” with the length diverges in the thermodynamic limit, the finite measure-k k L/8 are separated by the regime “b” and “d” with the ment amplitude can no longer suppress entanglement length 3L/8. Fig.2 (c) is the result for α = 1.8, where growth, leading to the absence of the MIP. Therefore, there is a crossing point γc, while Fig.2 (d) for α = 0.8 the operator norm HAB should play a central role in does not exhibit such a crossing phenomenon. determining the presencek k or absence of MIPs. We con- For the free fermion model, the finite-size effect is sig- sider the size dependence of the operator norm H k ABk nificant [19], and hence it is not trivial to obtain critical for ` = L/2 in the free fermion case. We numerically find 1 α points which separate the sub-volume law phase (non- that H L − (α < 1), log L (1 < α < 1.5) and k ABk ∝ ∝ zero central charge in S¯`) and the area law phase (con- constants (α > 1.5) (see also Fig. S7 in SM [41]). This ¯ stants in SL/2). To suppress the size-effects, we use explains the absence of MIP for α < 1.5 in Fig.3 (b). crossing points as illustrated in Fig.2 (c) to detect the Sufficient condition for the MIP in generic systems.— critical points. In addition, we consider the Berezinskii- The free fermion model indicates that the behavior of the Kosterlitz-Thouless (BKT) scenario that was valid for boundary interaction Hamiltonian is a key component for the short-range limit [19]. We remark that a similar observing the MIP, as it governs the entanglement growth technique using crossing points has been employed for rate under pure quantum dynamics, and we now use this several equilibrium systems [43, 44]. Note also that the key component to make a statement applicable to generic mutual information has been employed as a good indica- systems. We discuss the sufficient conditions to observe tor to detect the MIP in many systems [4, 10, 18]. We the MIP for generic fermion systems. That is, we seek use an ansatz of the finite-size scaling for the BKT sce- the value α ( α ), where for α > α , the MIP exists sc ≥ c sc nario [19, 45, 46]: γg(L)I¯(γ, α) = F [log(Lξ(γ, γc(α)))], for generic many-body quantum systems. To this end, where ξ is the BKT-type correlation length ξ(γ, γ (α)) we rigorously derive the following lemma: c ∼ 4

0.05 0.08 8 10 ® = 0.50 ® = 2.00 (a) ® = 3 (b) ® = 0.5 (c) (d) ® = 3 ® = 1.00 ® = 2.25 0.04 8 0.06 6 ® = 1.25 ® = 2.50 ® = 1.50 ® = 3.00 °p °p ® = 1.75 ® = ¯ 0.03 1 6 0.04 p 4 L = 8 ° ( ° , ® ) ¹ I

0.02 L = 10 ( ° , ® ) L 4 ¹ L = 12 I L = 8 L = 10 L = 14 0.02 2 L = 16 L = 12 0.01 2 L = 14 L = 16 0.00 0.00 0 0 0 2 4 6 8 10 0 2 4 6 8 10 8 10 12 14 16 18 20 22 -4 0 4 8 12 16 20 1/ ° ° L (° ° )L º ¡ p FIG. 4. (a) and (b): The mutual information I¯(γ, α) between the farthest two sites for α = 3.0 in (a) and for α = 0.5 in (b). The values γp indicated by arrows are amplitudes that give the peaks. (c): The system-size dependence of γp. (d): The finite-size −β 1/ν scaling for α = 3.0 with the ansatz I¯(γ, α) = L f((γ − γp)L ) with the exponents β = 2.59 ± 0.02 and ν = 1.4 ± 0.1.

Lemma 1 Let us consider a Hamiltonian on the d- between between the farthest two sites “a” and “c” that P dimensional hypercubic lattice as H = i αsc We show the results for α = 3.0 and α = 0.5 in  d/2 + 1 Bilinear Systems, Figs.4 (a) and (b), respectively, as functions of γ for α = (6) sc d + 1 Interacting Systems, different system sizes. To discuss the figures, we recall that many studies so far [4, 10, 18] have established that the operator norm HAB is upper-bounded by the bound- in non-integrable systems, the mutual information shows ary area between Ak and Bk . a peak as a function of γ, where the amplitude giving a peak denoted by γp is identified as a critical value of See the SM for rigorous proof [41]. Using this lemma and the measurement amplitude separating the volume law the underlying physics, we can make a physical statement phase for γ < γp, and the area law phase for γ > γp in regarding the existence of the MIP: the thermodynamic limit. Figs.4(a) and (b) also show the peak structure as a function of γ. However, a crucial Statement 1 α > αsc is a sufficient condition for observation here is that the values of γp (indicated by the existence of the MIP for generic uniform and non- arrows) generally depend on the system size. In the case commuting systems. of α = 3.0, the values of γp are not affected by the sys- Note that the statement is the sufficient condition for tem size, especially for large L. For α = 0.5, the values any system to have the MIP, and hence, this does not of γp are strongly affected by the system size, that is, γp exclude that specific models can show the MIP for α < systematically increases with increasing system size. αsc. True critical value for a specific model, αc, is equal This systematic change for α = 0.5 indicates that γp to or smaller than the value αsc. Intriguingly, in the eventually diverges in the thermodynamic limit, leading bilinear fermion system that we have discussed, the above to the absence of the MIP. From this observation, the statement is optimal, since αc = αsc = 3/2. system-size dependence of γp should be an indicator of Interacting fermion systems.— In the remainder of this the existence of the MIP. In Fig.4 (c), we plot the val- paper, we demonstrate that the above statement is satis- ues of γp as a function of the system size L for various α fied in non-integrable models. We consider the interact- values. This figure shows that for α > 2, γp is robustly fi- ing fermion system with V = 1 in (1). We first calculate nite for sufficiently large system sizes, which means that the operator norm H up to the size L = 256 us- the critical measurement amplitude exists even in the k ABk ing the density-matrix renormalization group technique thermodynamic limit; hence, the MIP shows up for (at 2 α [50–52]. We find clear evidence that H L − for least) α > 2. This observation is consistent with State- k ABk ∝ α < 2, while they are constants for α > 2 (see Fig. S8 in ment1, which states that α > 2 is sufficient to observe SM [41]), from which we anticipate that the MIP appears the MIP in one-dimensional generic interacting systems. for α > 2, which is consistent with Statement1. As an additional check on the existence of the MIP for To assess this statement in greater detail, we per- α > 2, we consider the finite-size scaling with the ansatz β 1/ν form the time-evolution calculation up to the system size I¯(γ, α) = L− f((γ γp)L ). In Fig.4 (d), we show L = 22 using the time-dependent variational principle that the finite-size scaling− works well with the exponents method [53–55]. See the footnote for the numerical de- β = 2.59 0.02 and ν = 1.4 0.1. Note that for α < 2, ± ± tails [56]. We focus on the mutual information I¯(γ, α) this scaling is not available since γp varies as increasing 5 the size. Available analysis for given data are consistent ity in random quantum circuits,” Phys. Rev. B 101, with Statement1. In the present interacting system, the 104302 (2020). [8] Yimu Bao, Soonwon Choi, and Ehud Altman, “Theory sufficient condition is optimal, since αsc = αc = 2. Summary.— We have revealed the effects of long-range of the phase transition in random unitary circuits with measurements,” Phys. Rev. B 101, 104301 (2020). interactions on the measurement-induced phase transi- [9] Soonwon Choi, Yimu Bao, Xiao-Liang Qi, and Ehud Alt- tion (MIP), which is summarized in Fig.1. The key man, “Quantum error correction in scrambling dynamics component for the existence of the MIP is the boundary and measurement-induced phase transition,” Phys. Rev. interaction Hamiltonian under pure quantum dynamics Lett. 125, 030505 (2020). in the thermodynamic limit. Based on this, we have ar- [10] M. Szyniszewski, A. Romito, and H. Schomerus, “En- rived at sufficient conditions to observe the MIP, as de- tanglement transition from variable-strength weak mea- scribed in Statement1. The numerical results for the surements,” Phys. Rev. B 100, 064204 (2019). [11] Ruihua Fan, Sagar Vijay, Ashvin Vishwanath, and Yi- specific models indicate that this condition is optimal. Zhuang You, “Self-organized error correction in random We hope that this criterion is useful in real experimental unitary circuits with measurement,” Phys. Rev. B 103, setup with long-range interaction [57–69]. 174309 (2021). [12] Sagar Vijay, “Measurement-driven phase transition within a volume-law entangled phase,” (2020), arXiv:2005.03052 [quant-ph]. ACKNOWLEDGEMENT [13] Ali Lavasani, Yahya Alavirad, and Maissam Barkeshli, “Measurement-induced topological entanglement transi- tions in symmetric random quantum circuits,” (2021). We are grateful to Yohei Fuji, Michael Buchhold, [14] Shengqi Sang and Timothy H. Hsieh, “Measurement- and Sebastian Diehl for providing details regarding their protected quantum phases,” Phys. Rev. Research 3, papers and useful suggestions. We also thank Seiji 023200 (2021). Miyashita for his useful comments on the BKT scal- [15] Matteo Ippoliti, Michael J. Gullans, Sarang Gopalakrish- ing. K.Su. was supported by Grants-in-Aid for Scien- nan, David A. Huse, and Vedika Khemani, “Entangle- tific Research (JP19K14644, JP20H01849).T.K. was sup- ment phase transitions in measurement-only dynamics,” ported by the RIKEN Center for AIP and JSPS KAK- Phys. Rev. X 11, 011030 (2021). [16] Qicheng Tang and W. Zhu, “Measurement-induced phase ENHI (Grant No. 18K13475). K.Sa. was supported transition: A case study in the nonintegrable model by Grants-in-Aid for Scientific Research (JP19H05603, by density-matrix renormalization group calculations,” JP19H05791). Phys. Rev. Research 2, 013022 (2020). Note added: After completing this work, we became [17] Shimpei Goto and Ippei Danshita, “Measurement- aware of complementary works on the measurement- induced transitions of the entanglement scaling law in induced phase transition using long-range quantum cir- ultracold gases with controllable dissipation,” Phys. Rev. cuits [70] and on the field theoretical argument for the A 102, 033316 (2020). [18] Yohei Fuji and Yuto Ashida, “Measurement-induced free fermion systems [71]. The latter work [71] is related quantum criticality under continuous monitoring,” Phys. to our sufficient condition in Statement1 and some clas- Rev. B 102, 054302 (2020). sification of phase transitions is shown. [19] Ori Alberton, Michael Buchhold, and Sebastian Diehl, “Entanglement transition in a monitored free-fermion chain: From extended criticality to area law,” Physical Review Letters 126, 170602 (2021). [20] Lukasz Fidkowski, Jeongwan Haah, and Matthew B. [1] Yaodong Li, Xiao Chen, and Matthew P. A. Fisher, Hastings, “How Dynamical Quantum Memories Forget,” “Quantum zeno effect and the many-body entanglement Quantum 5, 382 (2021). transition,” Phys. Rev. B 98, 205136 (2018). [21] Xhek Turkeshi, Rosario Fazio, and Marcello Dalmonte, [2] Amos Chan, Rahul M. Nandkishore, Michael Pretko, “Measurement-induced criticality in (2+1) -dimensional and Graeme Smith, “Unitary-projective entanglement hybrid quantum circuits,” Phys. Rev. B 102 (2020), dynamics,” Phys. Rev. B 99, 224307 (2019). 10.1103/physrevb.102.014315. [3] Brian Skinner, Jonathan Ruhman, and Adam Nahum, [22] Nicolai Lang and Hans Peter B¨uchler, “Entanglement “Measurement-induced phase transitions in the dynamics transition in the projective transverse field ising model,” of entanglement,” Phys. Rev. X 9, 031009 (2019). Physical Review B 102, 094204 (2020). [4] Yaodong Li, Xiao Chen, and Matthew P. A. Fisher, [23] Howard M Wiseman and Gerard J Milburn, Quantum “Measurement-driven entanglement transition in hybrid measurement and control (Cambridge university press, quantum circuits,” Phys. Rev. B 100, 134306 (2019). 2009). [5] Michael J. Gullans and David A. Huse, “Dynamical pu- [24] As in Ref. [72], a quantum circuit model with randomly rification phase transition induced by quantum measure- selected networks can also be called a long-range interact- ments,” Phys. Rev. X 10, 041020 (2020). ing model. However, it is not relevant to ours because the [6] Michael J. Gullans and David A. Huse, “Scalable probes system we address here is a Hamiltonian system, where of measurement-induced criticality,” Phys. Rev. Lett. each particle simultaneously interacts with the rest of the 125, 070606 (2020). particles depending on the distance. [7] Chao-Ming Jian, Yi-Zhuang You, Romain Vasseur, and [25] Evgenii Mikhailovich Lifshitz and Lev Petrovich Andreas W. W. Ludwig, “Measurement-induced critical- Pitaevskii, Statistical physics: theory of the condensed 6

state, Vol. 9 (Elsevier, 2013). [45] Anders W Sandvik, “Computational studies of quantum [26] Freeman J Dyson, “Existence of a phase-transition in a spin systems,” in AIP Conference Proceedings, Vol. 1297 one-dimensional ising ferromagnet,” Communications in (American Institute of Physics, 2010) pp. 135–338. Mathematical Physics 12, 91–107 (1969). [46] Kenji Harada and Naoki Kawashima, “Universal jump in [27] DJ Thouless, “Long-range order in one-dimensional ising the helicity modulus of the two-dimensional quantum xy systems,” Physical Review 187, 732 (1969). model,” Physical Review B 55, R11949 (1997). [28] JM Kosterlitz, “Phase transitions in long-range ferro- [47] Karel Van Acoleyen, Micha¨elMari¨en, and Frank Ver- magnetic chains,” Physical Review Letters 37, 1577 straete, “Entanglement rates and area laws,” Physical (1976). review letters 111, 170501 (2013). [29] Tomotaka Kuwahara and Keiji Saito, “Area law of non- [48] Charles H Bennett, Aram Wettroth Harrow, Debbie W critical ground states in 1d long-range interacting sys- Leung, and John A Smolin, “On the capacities of bipar- tems,” Nature communications 11, 1–7 (2020). tite hamiltonians and unitary gates,” IEEE Transactions [30] Thomas Koffel, M Lewenstein, and Luca Tagliacozzo, on Information Theory 49, 1895–1911 (2003). “Entanglement entropy for the long-range ising chain in [49] Sergey Bravyi, “Upper bounds on entangling rates of a transverse field,” Physical review letters 109, 267203 bipartite hamiltonians,” Physical Review A 76, 052319 (2012). (2007). [31] Davide Vodola, Luca Lepori, Elisa Ercolessi, Alexey V [50] Steven R. White, “Density Matrix Formulation for Quan- Gorshkov, and Guido Pupillo, “Kitaev chains with tum Renormalization Groups,” Physical Review Letters long-range pairing,” Physical review letters 113, 156402 69, 2863–2866 (1992). (2014). [51] Steven R. White, “Density-matrix algorithms for quan- [32] Davide Vodola, Luca Lepori, Elisa Ercolessi, and Guido tum renormalization groups,” Physical Review B 48, Pupillo, “Long-range ising and kitaev models: phases, 10345–10356 (1993). correlations and edge modes,” New Journal of Physics [52] Ulrich Schollw¨ock, “The density-matrix renormalization 18, 015001 (2015). group in the age of matrix product states,” Annals of [33] Chi-Fang Chen and Andrew Lucas, “Finite speed of Physics 326, 96–192 (2011). quantum scrambling with long range interactions,” Phys- [53] Jutho Haegeman, J. Ignacio Cirac, Tobias J. Osborne, ical review letters 123, 250605 (2019). Iztok Piˇzorn,Henri Verschelde, and Frank Verstraete, [34] Minh C Tran, Chi-Fang Chen, Adam Ehrenberg, An- “Time-Dependent Variational Principle for Quantum drew Y Guo, Abhinav Deshpande, Yifan Hong, Zhe- Lattices,” Physical Review Letters 107, 070601 (2011). Xuan Gong, Alexey V Gorshkov, and Andrew Lucas, [54] Jutho Haegeman, Tobias J. Osborne, and Frank Ver- “Hierarchy of linear light cones with long-range interac- straete, “Post-matrix product state methods: To tan- tions,” Physical Review X 10, 031009 (2020). gent space and beyond,” Physical Review B 88, 075133 [35] Tomotaka Kuwahara and Keiji Saito, “Strictly linear (2013). light cones in long-range interacting systems of arbitrary [55] Mingru Yang and Steven R. White, “Time-dependent dimensions,” Physical Review X 10, 031010 (2020). variational principle with ancillary Krylov subspace,” [36] Tianci Zhou, Shenglong Xu, Xiao Chen, Andrew Guo, Physical Review B 102, 094315 (2020). and Brian Swingle, “Operator l´evyflight: Light cones in [56] The truncation error is set to be smaller than 10−10, and chaotic long-range interacting systems,” Physical review after each measurement, we enlarge the bond dimension letters 124, 180601 (2020). of the matrix-product state using the global-subspace ex- [37] Shuji Tamaki and Keiji Saito, “Energy current correla- pansion method [55]. For each system size, we use the tion in solvable long-range interacting systems,” Physical same number of trajectories as in the free fermion case Review E 101, 042118 (2020). [42]. [38] Tomotaka Kuwahara and Keiji Saito, “Absence of fast [57] Vera Bendkowsky, Bj¨orn Butscher, Johannes Nipper, scrambling in thermodynamically stable long-range in- James P Shaffer, Robert L¨ow, and Tilman Pfau, “Ob- teracting systems,” Phys. Rev. Lett. 126, 030604 (2021). servation of ultralong-range rydberg molecules,” Nature [39] Rahul M Nandkishore and Shivaji Lal Sondhi, “Many- 458, 1005–1008 (2009). body localization with long-range interactions,” Physical [58] Immanuel Bloch, Jean Dalibard, and Wilhelm Zwerger, Review X 7, 041021 (2017). “Many-body physics with ultracold gases,” Reviews of [40] Jens Eisert, Marcus Cramer, and Martin B Plenio, “Col- modern physics 80, 885 (2008). loquium: Area laws for the entanglement entropy,” Re- [59] Mark Saffman, Thad G Walker, and Klaus Mølmer, views of Modern Physics 82, 277 (2010). “Quantum information with rydberg atoms,” Reviews of [41] Supplementary Material. modern physics 82, 2313 (2010). [42] Numerical details: We set the relaxation time to t = 30. [60] Bo Yan, Steven A Moses, Bryce Gadway, Jacob P Covey, We use 105 trajectories for averaging. For each trajectory, Kaden RA Hazzard, Ana Maria Rey, Deborah S Jin, and we shift the coordinate of sites along the ring to take more Jun Ye, “Observation of dipolar spin-exchange interac- average. Totally, we use L × 105 samples to obtain S¯. tions with lattice-confined polar molecules,” Nature 501, [43] Juan Carrasquilla and Marcos Rigol, “Superfluid to nor- 521–525 (2013). mal phase transition in strongly correlated bosons in two [61] K Aikawa, A Frisch, M Mark, S Baier, A Rietzler, and three dimensions,” Physical Review A 86, 043629 R Grimm, and F Ferlaino, “Bose-einstein condensation (2012). of erbium,” Physical review letters 108, 210401 (2012). [44] Masamichi Nishino and Seiji Miyashita, “Termination of [62] Joseph W Britton, Brian C Sawyer, Adam C Keith, the berezinskii-kosterlitz-thouless phase with a new crit- C-C Joseph Wang, James K Freericks, Hermann Uys, ical universality in spin-crossover systems,” Physical Re- Michael J Biercuk, and John J Bollinger, “Engineered view B 92, 184404 (2015). two-dimensional ising interactions in a trapped-ion quan- tum simulator with hundreds of spins,” Nature 484, 489– 7

492 (2012). [63] R Islam, C Senko, WC Campbell, S Korenblit, J Smith, A Lee, EE Edwards, C-CJ Wang, JK Freericks, and C Monroe, “Emergence and frustration of magnetism with variable-range interactions in a quantum simulator,” science 340, 583–587 (2013). [64] Johannes Zeiher, Rick Van Bijnen, Peter Schauß, Se- bastian Hild, Jae-yoon Choi, Thomas Pohl, Immanuel Bloch, and Christian Gross, “Many-body interferome- try of a rydberg-dressed spin lattice,” Nature Physics 12, 1095–1099 (2016). [65] Johannes Zeiher, Jae-yoon Choi, Antonio Rubio-Abadal, Thomas Pohl, Rick Van Bijnen, Immanuel Bloch, and Christian Gross, “Coherent many-body spin dynamics in a long-range interacting ising chain,” Physical Review X 7, 041063 (2017). [66] Hannes Bernien, Sylvain Schwartz, Alexander Keesling, Harry Levine, Ahmed Omran, Hannes Pichler, Soon- won Choi, Alexander S Zibrov, Manuel Endres, Markus Greiner, et al., “Probing many-body dynamics on a 51- atom quantum simulator,” Nature 551, 579–584 (2017). [67] Jiehang Zhang, Guido Pagano, Paul W Hess, Antonis Kyprianidis, Patrick Becker, Harvey Kaplan, Alexey V Gorshkov, Z-X Gong, and Christopher Monroe, “Obser- vation of a many-body dynamical phase transition with a 53-qubit quantum simulator,” Nature 551, 601–604 (2017). [68] Brian Neyenhuis, Jiehang Zhang, Paul W Hess, Jacob Smith, Aaron C Lee, Phil Richerme, Zhe-Xuan Gong, Alexey V Gorshkov, and Christopher Monroe, “Obser- vation of prethermalization in long-range interacting spin chains,” Science advances 3, e1700672 (2017). [69] Fangli Liu, Rex Lundgren, Paraj Titum, Guido Pagano, Jiehang Zhang, Christopher Monroe, and Alexey V Gor- shkov, “Confined quasiparticle dynamics in long-range interacting quantum spin chains,” Physical review letters 122, 150601 (2019). [70] Maxwell Block, Yimu Bao, Soonwon Choi, Ehud Altman, and Norman Yao, “The measurement-induced transi- tion in long-range interacting quantum circuits,” arXiv preprint arXiv:2104.13372 (2021). [71] Michael Buchhold Thomas M¨uller, Sebastian Diehl, “Measurement-induced dark state phase transitions in long-ranged fermion systems,” arXiv preprint arXiv:2105.08076 (2021). [72] Adam Nahum, Sthitadhi Roy, Brian Skinner, and Jonathan Ruhman, “Measurement and entanglement phase transitions in all-to-all quantum circuits, on quan- tum trees, and in landau-ginsburg theory,” Physical Re- view X Quantum 2, 010352 (2021). [73] Matthew Fishman, Steven R White, and E Miles Stoudenmire, “The itensor software library for tensor network calculations,” arXiv preprint arXiv:2007.14822 (2020). Supplemental Material for “Fate of measurement-induced phase transition in long-range interactions”

Takaaki Minato1, Koudai Sugimoto1, Tomotaka Kuwahara2, and Keiji Saito1 1Department of Physics, Keio University, Hiyoshi, Kohoku-ku, Yokohama 223-8522, Japan 2Mathematical Science Team, RIKEN Center for Advanced Intelligence Project (AIP), 1-4-1 Nihonbashi, Chuo-ku, Tokyo 103-0027, Japan

PROOF OF THE LEMMA 1

We rigorously show the lemma in the main text, which is restated as follows P Lemma 1 Let us consider a Hamiltonian on a d-dimensional hypercubic lattice as H = i α with k i,jk ≤ i,j sc  d/2 + 1 Bilinear Systems, α = (S.1) sc d + 1 Interacting Systems, the operator norm H is upper-bounded by the boundary area between A and B. k ABk Below, we present the proof of the lemma. We consider a d-dimensional hypercubic lattice with linear size L and unit spacing. We impose the open boundary condition. We can write the operator norm of the boundary interaction Hamiltonian per boundary surface area as follows: A X X X X H / := h / h / , (S.2) k ABk A k i,jk A ≤ k i,jk A i A j B i A j B ∈ ∈ ∈ ∈ d 1 where is the boundary surface area between A and B, that is, = L − . We consider the r.h.s. in (S.2). We first considerA the case of bilinear systems. In the bilinear HamiltonianA case, the local Hamiltonian can be generically written as:

α hi,j = (gi,jci†cj + gi,j0 cicj + h.c.)/ri,j, (S.3) where gi,j and gi,j0 are O(1). Then, we estimate the bound in (S.2). To this end, we define

1 X α 1 X α di := µi− gi,jcj/ri,j , di0 := µi0 − gi,j0 cj/ri,j,, j B j B ∈ ∈  1/2  1/2 (S.4) X 2 2α X 2 2α µi :=  gi,j /r  , µ0 :=  g0 /r  . | | i,j i | i,j| i,j j B j B ∈ ∈ Then, we can estimate the bound in (S.2) as follows   X X 1 X X α X α hi,j / 2 −  c† gi,jcj/r + ci g0 cj/r  k k A ≤ A k i i,jk k i,j i,jk i A j B i A j B j B ∈ ∈ ∈ ∈ ∈  1/2 1 X h i 1 X 1 X X 2α = 2 − c†µidi + ciµ0 d0 2 − µi + µ0 4 − gmax  1/r  A k i k k i ik ≤ A i ≤ A i,j i A i A i A j B ∈ ∈ ∈ ∈ 1/2 L/2 L/2  1 X X X X 2α = 4 − gmax  1/r  A i,j x=1 i:dist(i:B)=x y=1 j:dist(A:j)=y 1/2 L/2 L/2  X X X 2α = 4gmax  1/ri,j  , (S.5) x=1 y=1 j:dist(A:j)=y where gmax = max( gi,j, gi,j0 ), dist(i : B) := minj B ri,j, and dist(A : j) := mini A ri,j. We have used = P d 1{ } ∈ ∈ A i:dist(i:B)=x = L − , irrespective of x. Finally, we consider the r.h.s. in (S.5). For d = 1, we can estimate the bound as follows:

1/2 1/2 1/2 1/2 L/2 L/2  L/2 L/2  L/2  L/2 L/2  X X X 2α X X 2α X 2α X X 2α  1/ri,j  =  1/(x + y)  =  1/(1 + y)  +  1/(x + y)  x=1 y=1 j:dist(A:j)=y x=1 y=1 y=1 x=2 y=1 1/2 1/2 1 3/2 (2α 1)− + (2α 1)− (3/2 α)− ((L/2) 1) , (S.6) ≤ − − − − where in advance, we assumed α > 1/2 to obtain the final inequality. For d 2, we can calculate the r.h.s. in (S.5) as follows: ≥

1/2 1/2 1/2 L/2 L/2  1 L/2  L/2 L/2  X X X 2α X X X 2α X X X 2α  1/ri,j  =  1/ri,j  +  1/ri,j  x=1 y=1 j:dist(A:j)=y x=1 y=1 j:dist(A:j)=y x=2 y=1 j:dist(A:j)=y  1/2  1/2 L/2 Z L/2 L/2 Z X ∞ d 2  2 2α X X ∞ d 2  2 2α Γd dξξ − / (1 + y) + ξ  + Γd dξξ − / (x + y + 1) + ξ  ≤ y=1 0 x=1 y=1 0 1/2 1/2  L/2  L/2  L/2  X d 1 2α X X d 1 2α = (ΓdB/2) (y + 1) − −  + (ΓdB/2) (x + y + 1) − −  y=1 x=1 y=1

1/2 1/2 h α+d/2+1 i [(Γ B/2)/(2α d)] + [(Γ B/2)/(2α d)] (L/2)− 1 /( α + d/2 + 1) , ≤ d − d − − − (S.7)

where Γd = 2 for d = 2 and 2π for d = 3, which is the coefficient of the (d 1)-dimensional spherical surface area, and B is the beta function, B( 1/2 + d/2, α + 1/2 d/2). In the calculation,− we assumed 2α d > 0 in advance. Thus, the bilinear case in (S.1)− is proven. − − In generic Hamiltonian systems, we cannot use the trick (S.4); hence, we calculate the bound as follows:

L/2 L/2 X X 1 X X α X X X α h / g − 1/r g 1/r . (S.8) k i,jk A ≤ A i,j ≤ i,j i A j B i A j B x=1 y=1 j:dist(A:j)=y ∈ ∈ ∈ ∈

For the d = 1 case, we have

L/2 L/2 L/2 L/2 X X X α X X α  α+2  1/r = 1/(x + y) 1/(α 1) + (L/2)− 1 / [(α 1)( α + 2)] , (S.9) i,j ≤ − − − − x=1 y=1 j:dist(A:j)=y x=1 y=1 where, in advance, we have assumed α 1 > 0. For d 2, the bound is computed as follows: − ≥

L/2 L/2 1 L/2 L/2 L/2 X X X α X X X α X X X α 1/ri,j = 1/ri,j + 1/ri,j x=1 y=1 j:dist(A:j)=y x=1 y=1 j:dist(A:j)=y x=2 y=1 j:dist(A:j)=y L/2 Z L/2 L/2 Z X ∞ d 2  2 2α/2 X X ∞ d 2  2 2α/2 Γ dξξ − / (x + y) + ξ + Γ dξξ − / (x + y + 1) + ξ ≤ d d y=1 0 x=2 y=1 0  α+d+1  Γ B0/ [2(α d)] + Γ B0/ [2(α d)] (L/2)− 1 /( α + d + 1) , (S.10) ≤ d − d − − − where Γd is the same as in the bilinear case, and B0 is the beta function B0 = B((d 1)/2, (α d + 1)/2). In the calculation, we assumed α d > 0 in advance. Thus, the condition for generic systems− in (S.1) is− proven. − NUMERICAL METHOD FOR FREE FERMION SYSTEMS

Let L be the size of a one-dimensional lattice, and let N be the total number of fermions. We consider the free fermion system as:

L L/2 X X 1 h i H = c† c c†c + . (S.11) rα − j+r j − j j r j=1 r=1

Because the Hamiltonian is bilinear, the many-body wave function is always cast in the following form:

N L Y X ψ(t) = u (t) c† vac , (S.12) | i j,m j | i m=1 j=1

PL where V is the vacuum state. The relation u∗ u 0 = δ 0 guarantees the normalization ψ(t) ψ(t) = 1 | i j=1 j,m j,m m,m h | i PL n o because bm† := j=1 uj,mcj† defines a new fermion operator satisfying the anti-commutation relations bm, bm† 0 = δ 0 and b , b 0 = 0 m,m { m m } The correlation function c†c 0 := ψ(t) c†c 0 ψ(t) is computed using the L N matrix u. Note that c†c 0 = h i i i h | i i | i × h i i i δ 0 ψ(t) c 0 c† ψ(t) , and b0 = c 0 and b† = c†. Then, we have i,i − h | i i | i 0 i 0 i

ci†ci0 = V bN bN 1 b1b0b0†b1† bN† 1bN† V = det A , (S.13) h i (h | − ··· ··· − | i V b0bj† V i = 0, & j = 0, ,N Ai,j = h | | i ··· , (S.14) V b b† V otherwise h | i j| i where A is the (N + 1) (N + 1) matrix. Note that V bibj† V = δi,j for i, j = 1, ,N. Then, after simple algebra using elementary mathematical× properties on the determinant,h | | i we obtain the relation···

N X c†c 0 = u∗ u 0 . (S.15) h i i i i,m i ,m m=1 The entanglement entropy is computed following the standard argument [40], once the correlation matrix is obtained. Suppose that we consider subsystem A consisting of ` sites j = 1, , `. Let ρA be a reduced density matrix for subsystem A, which is generically written in the following Gaussian form:···

` ! ` X Y am ρ = exp a d† d / (1 + e− ) , (S.16) A − m m m m=1 m=1 where dm is a fermion operator connected to the original fermion operator cj through a unitary transform dm = P` { } j=1 vm,jcj. Then, the entanglement entropy S` is given as

` X   S = d† d log d† d . + d d† log d d† , (S.17) ` − h m mi h m mi h m mi h m mi m=1 where d† d can be computed by diagonalizing the correlation matrix D, the element of which is given by D = h m mi i,j ci†cj (i, j = 1, , `). h Wei explain the··· protocol for computing the time evolution of the wave function ψ(t), that is, equivalently, the time evolution of the L N matrix u(t)[19] (i): We determine the jump time τ = log(r)/(γN) with a uniformly distributed random number× r [0, 1]. (ii): During the time duration τ, the unitary time− evolution is performed with ∈ ihτ the one-particle Hamiltonian h for the matrix u as u(t + τ) = e− u(t). (iii) Site j is measured according to the probability Pj = ψ(t + τ) cj†cj ψ(t + τ) /N. (iv) The post-measurement state ψ0 after the measurement of site j is formally given ash | | i | i

cj†cj ψ ψ0 = | i . (S.18) | i q c†c h j ji Using Wick’s theorem, one finds the relation for the correlation function

 1 i = i = j  0 D 0 := ψ0 c†c 0 ψ0 = 0 (i = j, i0 = i) or (i = i0, i0 = j) . (S.19) i,i h | i i | i 6 6  c†c 0 c†c 0 c†c / c†c otherwise h i i i − h j i ih i ji h j ji

The new matrix u is obtained through an SVD decomposition D = uSu†, where Si,i = 1 (1 i 1) and 0 (N + 1 i L). ≤ ≤ ≤ ≤

SUPPLEMENTARY NUMERICAL DATA FOR FREE FERMION SYSTEMS

Here, we present supplementary numerical data for the free fermion system. We first show the crossing behavior of the mutual information as a function of γ in Fig. S5. See the main text for the definition of mutual information. In the figures for α = 5.0, 2.0, and 1.6, we clearly see that the data cross each other. However, for α = 1.4, 1.3 and 1.1, curves overlap each other for the large γ regime and hence, we cannot see the crossing phenomena. In the main text, we showed the finite scaling analysis for the mutual information. We here show that the central charges show finite-size scaling. Getting the the mutual information data does not need any fitting procedures, and hence the mutual information is accurately computable. The central charge is, however, obtained through the fitting procedure around ` = L/2 for the entanglement entropies S¯` as shown in Fig.1 (a) in the main text. Hence, we should have in mind that values of central charges are less accurate than the mutual information due to the fitting procedure. We show the results of finite-size scaling for the central charges in S6. Within available data, we can see that the central charges are well scaled with the BKT scenario. Finally, we show the size dependence of the boundary interaction Hamiltonian HAB, which is defined as

`  L j+L/2  X X X X 1 X 1 h i HAB := hi,j =  +  c†ck c† cj , (S.20) L + j k α k j α − j − k i A j B j=1 k=L/2+j k=`+1 ∈ ∈ | − | | − | where we set the subsystem A with length L/2. We numerically calculate the operator norm HAB as a function of the system size up to L = 28000. The results are shown in Fig. S7. The figures strongly suggestk k the following

α α α

L L L ) ) L ) L L L L L ( γ,α ( γ,α ( γ,α I I I

γ γ γ α α α

L L L

) L ) L ) L L L L ( γ,α ( γ,α ( γ,α I I I

γ γ γ

FIG. S5: Mutual informaations as a function of γ for many cases of α. 3.5 L 3.0 L ) 2.5 L

c( γ 2.0 L ) γ 1.5

g ( L 1.0 α α α 0.5 ∞ 0.0 0 0 0 log(L) ν/√γ γ − − c

FIG. S6: The finite-size scaling for the central charges with the BKT scenario. γg(L) versus p log(Lξ(γ, γc(α))) = log L ν/ γ γc(α), where ν = 5.0, 4.3 and 3.5 for α = 1.8, 2.3 and , − − respectively. ∞

(a) L L (b) α α (c) α α L L α α α α p L L µ α α L0.8 ∝ α L0.6 α

AB ∝ AB AB H H H L0.4 ∝ L0.2 ∝

0 < α < 1.0 1.0 < α < 1.5 1.5 < α

α L L

FIG. S7: Operator norm of the boundary-interaction Hamiltonian HAB for ` = L/2 and V = 0. We calculated up to the system size L = 28000. (a): The operator normk fork various system-sizes as a function of α. We show the data up to L = 20000. We see that the norms are constants for α > 1.5, which is guaranteed by the Lemma 1. (b): The operator norm as a function of the system size for the regime α < 1 in the log-log plot. The lines are fittings by the function aLµ + b. (c): The operator norm as a function of the system size for the regime 1 < α < 1.5 in the semi-log plot. The lines are fittings by 1 α the function p log L + q. From these data, the system-size dependence indicates that HAB L − for α < 1, log L for 1 < α < 1.5 and constants for α > 1.5. k k ∝ ∝ behavior:  1 α  L − α < 1 , HAB log L 1 < α < 1.5 , (S.21) k k ∝  const. α > 1 .

SUPPLEMENTARY NUMERICAL DATA FOR THE INTERACTING SYSTEMS

In this section, we consider the Hamiltonian

L L/2 X X 1 h i H = c† c c†c + + V n + n , (S.22) rα − j+r j − j j r j r j j=1 r=1 where we set V = 1. We consider the boundary-interaction Hamiltonian HAB:

`  L j+L/2  X X X X 1 X 1 h i HAB := hi,j =  +  c†ck c† cj + V njnk . (S.23) L + j k α k j α − j − k i A j B j=1 k=L/2+j k=`+1 ∈ ∈ | − | | − | 1000 2

¹ 1 ® = 0.5

100 0 0 1 2 3 k ® AB k H 10

® = 3.0

1 16 32 64 128 256 L

FIG. S8: Size dependence of HAB at α = 0.5, 0.75, 1.0, , 3.0. The numerical calculation shows k power-lawk behavior (S.24··· ).

2.0 1.0 (a) ® = 3.0 (b) ® = 0.5 0.8 1.5 ° = 1.0 ° = 8.0 0.6 `

¹ 1.0 S 0.4 0.5 0.2 ° = 20.0 ° = 4.0 0.0 0.0 0 2 4 6 8 10 0 2 4 6 8 10 ` `

FIG. S9: `-dependence of the entanglement entropy for various measurement amplitudes γ at L = 20. (a): α = 3.0 and γ = 1.0, 1.5, 2.0, , 4.0. (b): α = 0.5 and γ = 8.0, 10.0, 12.0, , 20.0. ··· ···

We calculate the operator norm HAB using the density-matrix renormalization group technique [50–52] implemented in the ITensor Library [73]. Thek resultsk are shown in Fig. S8, which shows the power-law dependence on the system µ size L. We fitted the results HAB = aL + b at L 64. The exponents µ are shown as a function of α in the inset, which indicates that k k ≥  L2 α α < 2 , H − (S.24) k ABk ∝ const. α > 2 . From this dependence, we anticipate that the MIP exists (at least) for α > 2. We perform time-evolution calculations of the interacting systems using the one-site version of the time-dependent 10 variational principle method [53–55]. The truncation error is set to be smaller than 10− , and after each measurement, we enlarge the bond dimension of the matrix-product state using the global-subspace expansion method [55]. For each system size, we use the same number of trajectories as in the free fermion case [42]. In Fig. S9, we show the entanglement entropy for α = 3.0 and α = 0.5, as typical `-dependence for different measurement amplitudes γ. For the case of α = 3.0, sufficiently strong measurements suppress the entanglement growth, leading to the area law, that is, they become flat in small ` regimes. Conversely, for α = 0.5, the entanglement entropies never become flat before L/2, even for large measurement amplitudes. We remark that the entanglement entropy at L/2 is always flat, as seen in the Page curve for the random matrix.