Arxiv:1901.06914V4

Home , Hubbard model

arXiv:1901.06914v4 [cond-mat.str-el] 28 Oct 2020 O n eieteeutoswihmk the make with which models equations Hubbard the of derive class and a SOC to generalized be can ory prtra ieoeao fteHmloin h ex- The Hamiltonian. the of act eigenoperator an operator sa xc eutdet Yang to due result exact an is etgto.I atclr ubro hoeia un- correlated SOC strongly with theoretical systems for in- of electronic obtained number intense been a phys- under have currently particular, and derstandings In are phases SOC topological vestigation. with of systems wealth ical a produce can uecnutn rudsae nacaso togycor- strongly of models class electronic a related in states ground superconducting pnobtculditrcigeetoi ytm would systems for valuable. electronic solutions be interacting or results coupled exact spin-orbit Therefore, deserves systems studies. these further and of rigorous understanding more physical and theory clearer field mean as such approximate methods various used investigations theoretical these and physics models Hubbard-like thermalization in quantum scars of many-body quantum study the and of entropy entanglement of culation SU the by inspired Studies theory. BCS of work the importantly, More ino aiu ymtybekn res(n related (and unifica- orders the study) behavior symmetry-breaking including multicritical various pursued, of been tion since have theory n ffdaoa ogrneodr(DR)adreveals and (ODLRO) possess- order an eigenstates long-range exact off-diagonal of ing construction the for allows prcae.Oeipratavnei tdigthe studying so-called in the advance is important model highly Hubbard One and rare are dimension appreciated. spatial one beyond models ti o eonzdta pnobtculn (SOC) coupling spin-orbit that recognized now is It nti ok efis hwta Yang’s that show first we work, this In xc eut rsltoso neatn many-electron interacting of solutions or results Exact 2 SU 2 ag hoyo h ubr model Hubbard the of theory gauge (2) ainlLbrtr fSldSaeMcotutrsadSch and Microstructures State Solid of Laboratory National SU 2 suopnsymmetry pseudospin (2) 2 suopnsmer a ete preserved then be can symmetry pseudospin (2) hc xiisa xc suopnsmer ihfine-tuned i with an on symmetry focus pseudospin the and exact for construct an then We exhibits which systems. these of topology iesmmtl n hr nuao.The insulator. Chern a and semimetal, line xc eut a eapidt ait fHbadmdl wi Dir models a Hubbard be of can variety limit a noninteracting to whose applied lattices, bipartite be can results exact oe,ee huhthe though even model, The results. xcl ovbemlirtclln.I h n,w discuss we end, symm the In distinct these line. Between multicritical solvable established. exactly are states ground the which under conditions the obtain and egnrlz the generalize We η .INTRODUCTION I. η piigi orltdfrinmdl ihsi-ri coup spin-orbit with models fermion correlated in -pairing arn hsrvasa exact an reveals thus pairing exact η piigsae a ev sexact as serve can states -pairing 3 η 1 – piiggon tts(ihmomentum (with states ground -pairing colo hsc,ZeghuUiest,Zeghu400,C 450001, Zhengzhou University, Zhengzhou Physics, of School 6 hc sbyn h frame- the beyond is which , 2 23 , 7 1 – – η 25 The . 11 2 piigter nHbadmdl oteoe ihspin-orbi with ones the to models Hubbard in theory -pairing η eetees otof most Nevertheless, . h osrcino an of construction the , fteHbadmodel. Hubbard the of piigter,which theory, -pairing SU η piigstates -pairing η 2 pnsmer sepiil rknb h O.I particu In SOC. the by broken explicitly is symmetry spin (2) piigoperator -pairing η 12 piigthe- -pairing – 14 SU Dtd coe 0 2020) 30, October (Dated: η η h cal- the , -pairing -pairing 2 suopnsmer norsi-ri ope Hubbard coupled spin-orbit our in symmetry pseudospin (2) 17 15 η a Li Kai – piigoeao sa ieoeao fteHamiltonian. the of eigenoperator an is operator -pairing η , 22 16 arn odtosas moecntanso h band the on constraints impose also conditions pairing . , o fPyis ajn nvriy ajn 103 China 210093, Nanjing University, Nanjing Physics, of ool ,2, 1, rdc u oe n eeaiethe generalize and model our troduce td n eetwrsaedsusd ocnetthe Hubbard convert the To besides teractions discussed. present are the works η pseu- between recent the the relation and the of in study and consequences insulator, symmetry Some Chern dospin semimetal, a Weyl limit. and a noninteracting semimetal, semimetal, sys- these nodal-line Dirac coupled a exemplify a spin-orbit then concrete including We four tems, SOC. with results the exact by broken plicitly nSec. In eetbihteexact the the making regions, By establish parameter we some lattice. in Dirac- square frustration-free interacting the Hamiltonian an on on model focus semimetal and construct we ticular, the while systems, these in xc hredniywv CW rudsae,where states, the ground (CDW) charge-density-wave exact osbeeprmna elztoso u eut noptical in discuss results lattices. we our of end, realizations the the experimental between In possible line phases. multicritical symmetry-breaking solvable three there exactly Moreover, an a exists and respectively. phase superconductor, (PDW) pair-density-wave uniform a with identified utciia iei Sec. in line multicritical nteappendices. the in Sec. in discussion Sec. and in realizations imental r ihfu oceemdl.A neatn Dirac- interacting An exact with models. model concrete semimetal four with ory rsne nSec. in presented piigegnttsit rudsae,adtoa in- additional states, ground into eigenstates -pairing hsppri raie sflos nSec. In follows: as organized is paper This ∗ η π piiggon ttswt momentum with states ground -pairing csmmtl elsmmtl nodal- a semimetal, Weyl a semimetal, ac osbeeprmna elztoso our of realizations experimental possible or III 0 tybekn hss hr xssan exists there phases, etry-breaking n the and ) hSCo ihrbprieo non- or bipartite either on SOC th trcigDrcsmmtlmodel, Dirac-semimetal nteracting eeepiytegeneralized the exemplify we aaees h tblt regions stability The parameters. IV exact n eietf neatysolvable exactly an identify we and VII η piiggon ttsadthe and states ground -pairing charge-density-wave V eti eal a efound be can details Certain . hina edsuspsil exper- possible discuss We . opig(SOC) coupling t SU VI η piiggon ttsis states ground -pairing 2 pnsmer sex- is symmetry spin (2) n rsn summary a present and U a,these lar, ling r edd npar- In needed. are η piigtheory. -pairing η piigthe- -pairing π and II ein- we 0 are 2

II. MODEL AND GENERALIZATION OF η (note the different meaning of an eigenoperator intro- PAIRING duced in the context of the rigged Hilbert space26,27) of the total Hamiltonian, namely The generic Hamiltonian for our spin-orbit coupled Hubbard models in arbitrary spatial dimensions is given [H, ηQ† ] = (λ + U)ηQ† . (6) by Therefore, one can construct many exact eigenstates H = H0 + HU m m H(ηQ† ) |0i = m(λ + U)(ηQ† ) |0i possessing ODLRO. H0 = Ck† (αkσx + βkσy + γkσz + εkσ0)Ck The SU(2) spin symmetry is explicitly broken by the Xk SOC terms in Eq.(1). However, from Eq.(6), we see that our spin-orbit coupled Hubbard model possesses an exact HU = U ni ni . (1) ↑ ↓ SU(2) pseudospin symmetry when λ + U = 0. The term Xi pseudospin is attributed to the SU(2) algebra formed by The noninteracting part H0 is written in momentum η-pairing operators, i.e., [Jα,Jβ] = iǫαβγJγ (α,β,γ = space, where C† = (c† ,c† ). In H , the three 1 1 k k k 0 x,y,z), where J = (η† + ηQ),J = (η† − ηQ), J = ↑ ↓ x 2 Q y 2i Q z momentum-dependent real coefficients αk,βk,γk with 1 (Nê −N), Nê = niσ, and N is the number of lattice Pauli matrices σx,y,z represent the SOC, and the term εk 2 iσ sites. P with the identity matrix σ0 is the usual energy band without SOC. The interacting part H written in real space Physically, the pseudospin symmetry can be under- U stood via a particle-hole transformation for the spin- is the on-site Hubbard interaction, where niσ = ciσ† ciσ. iQ j down electrons: fj† = cj† ,fj† = e− · cj . The pseu- In the absence of SOC (αk = βk = γk = 0), Yang finds ↑ ↑ ↓ ↓ iπ j dospin operators can now be rewritten in terms of the f that the so-called η-pairing operator ηπ† = j e · cj† cj† 1 ↑ ↓ fermions as Jα = 2 j fj†ταfj, where τα are the Pauli obeys the commutation relations [H0, ηπ† ]P = −2µηπ† and matrices and fj† = (fPj† ,fj† ). Thus, the pseudospin can [HU , ηπ† ] = Uηπ† , where µ is the chemical potential and ↑ ↓ π = (π,π,π) or (π, π) for the cubic or square lat- be interpreted as the spin of f fermions. This is rem- iniscent of the SU(2) gauge structure in the fermionic- tice. The total Hamiltonian and η† thus obey [H, η† ]= π π spinon representation of spin operators28–30. It is impor- (U − 2µ)ηπ† , and one can construct many exact eigen- m m tant to emphasize that the η-pairing states do not have states H(ηπ† ) |0i = m(U − 2µ)(ηπ† ) |0i (m = 0, 1, ...) possessing ODLRO. the pseudospin symmetry. In fact, they correspond to In Yang’s observation, the key point that makes the the ferromagnetic states of f fermions with maximal total 2 m N N m ”spin” N/2, i.e., J (η† ) |0i = ( + 1)(η† ) |0i and possible η pairing is the fact that the energy band εk Q 2 2 Q m N m 2 2 (for the cubic or square lattice without SOC) has the Jz(ηQ† ) |0i = (m − 2 )(ηQ† ) |0i (where J = α Jα), property εk + επ k = −2µ (independent of k), which which break the f fermion’s spin symmetry and hence − P leads to the commutation relation [H0, ηπ† ]= −2µηπ† . break the pseudospin symmetry. Therefore, to generalize the η-pairing theory to Hub- bard models with SOC, e.g., Eq.(1), we need some new constraints on the coefficients αk,βk, and γk. To achieve III. APPLICATION OF THE GENERALIZED these new constraints, we first define the η-pairing oper- η-PAIRING THEORY ator with a generic momentum Q as Based on the above exact results for model (1), we find iQ j ηQ† = ck† cQ† k = e · cj† cj† . (2) that the the η pairing can exist in a variety of spin-orbit ↑ − ↓ ↑ ↓ Xk Xj coupled systems. In particular, we apply these exact results to four concrete examples, which include three gap- It can then be shown that (see Appendix A) less and one gapped noninteracting topological phases due to SOC. [H0, η† ]= λη† , (3) Q Q The first example is a Dirac-semimetal Hubbard model if αk,βk,γk,εk, and Q satisfy the following equations on a square lattice. Its Hamiltonian is given by Eq.(1) with αk = 2t′ sin kx,βk = 2t′ sin ky,γk = 0, and αk = αQ k, βk = βQ k, − − εk = −2t(cos kx + cos ky) − µ. The lattice spacing εk + εQ k + γk − γQ k = λ, (4) has been set to unity throughout our paper. There − − are four Dirac points K = (n π,n π) in momentum where λ is a constant independent of k. The commutator 1 2 space at U = 0, where n = 0 or 1. Specifically, of the Hubbard interaction and the η-pairing operator 1,2 near the doubly degenerate point (e.g., K = (0, 0)) reads of the energy bands, the Bloch Hamiltonian H(k) = α σ + β σ + γ σ + ε σ to the leading order reads [HU , ηQ† ]= UηQ† , (5) k x k y k z k 0 H(k) ≃ 2t′kxσx + 2t′kyσy + (−4t − µ), which is a 2- which holds for arbitrary Q. It follows from Eqs.(3) and dimensional massless Dirac Hamiltonian31–33. Substitut- (5) that the η-pairing operator is now an eigenoperator ing the expressions of αk,βk,γk and εk into Eq.(4), we 3

find that the η-pairing momentum Q = π and the con- where hAi denotes the ensemble average of the operator stant λ = −2µ. A at any temperature. For the above four examples of Similarly, the next example we consider is a Weyl- spin-orbit coupled Hubbard model (1), the exact SU(2) semimetal Hubbard model on a cubic lattice34. Its pseudospin symmetry is respected when λ + U = 0 or simply µ = U/2, so the electron density is fixed at half Hamiltonian is given by Eq.(1) with αk =2t′ sin kx,βk = filling (independent of the temperature) when µ = U/2. 2t′ sin ky,γk =2t′ sin kz, and εk = −2t(cos kx + cos ky + Another consequence of the exact pseudospin symmetry cos kz) − µ. There are totally eight Weyl points K = is that the ODLRO and CDW long-range order must (n π,n π,n π) at U = 0, where n , , =0 or 1. As can 1 2 3 1 2 3 2,39 be seen from Eq.(4), the η-pairing momentum Q = π either be both absent or coexist . In particular, the and the constant λ = −2µ. Notice that for our Dirac- or PDW phase (see below) possesses these two orders, whose Weyl-semimetal Hubbard model, the original η-pairing parent Hamiltonian Eq.(7) exhibits an exact pseudospin 1 symmetry (see Appendix F). theory is recovered in the absence of SOC, e.g., t′ = 0. Lastly, beyond these four concrete topological Hub- Our third example is a nodal-line-semimetal Hubbard bard models, the exact SU(2) pseudospin symmetry has model on a cubic lattice35. Its Hamiltonian is given by also been found in other Hubbard-like models, such as Eq.(1) with α = t(cos k +cos k −b),β = t sin k ,γ = k x y k ′ z k the extended Falicov-Kimball model40 and the Hubbard 0, and ε = −µ. At U = 0, upon expanding α and β k k k model with a Zeeman term41, which are in fact spe- near the k = 0 point, we obtain a nodal ring with radius cial examples of our general spin-orbit coupled Hubbard 2(2 − b) pinned at the k = 0 plane. We notice that, z model (1). Therefore, the η pairing structure and the from Eq.(4), the η-pairing momentum Q = (0, 0, π) and p associated SU(2) pseudospin symmetry of the models in the constant λ = −2µ. Refs.40,41 can be readily obtained by using our general- The last example we consider is a Chern-insulator Hub- ized η-pairing theory for Hubbard models with SOC. bard model on a triangular lattice36. Its Hamiltonian is kx √3ky given by Eq.(1) with αk = −2t cos( 2 − 2 ),βk = IV. CONSTRUCTION OF THE EXACT kx √3ky −2t cos( 2 + 2 ),γk = −2t cos kx, and εk = −µ. η-PAIRING GROUND STATES This model breaks time-reversal symmetry and H0 has a nonzero band Chern number ±2. In the large-U limit, In general, the exact η-pairing eigenstates of the Hub- this system is described effectively by the triangular lat- 1 tice Kitaev-Heisenberg spin model36,37. We see that, bard model (1) are not ground states . In fact, beyond from Eq.(4), the η-pairing momentum Q = 0 and the one spatial dimension, the Hubbard model has no ex- constant λ = −2µ. Note that the η-pairing operator with act solution to this day. However, some extended Hub- Q = 0 is simply the s-wave Cooper-pairing operator in bard models containing additional interactions are ex- the usual BCS theory. In fact, the first generalization of actly solvable by constructing the ground state wave functions explicitly3,4,6,42. Following the method used in η pairing to a triangular lattice was obtained by adding 42 a staggered π/2 flux through each triangle plaquette38, Ref. , we would further show that the η-pairing states in the absence of SOC. can be constructed as the exact ground states of extended Hubbard models with SOC. Now, a few remarks are in order. Firstly, as shown in For the sake of concreteness, let us consider a square Appendix B, the η pairing conditions Eq.(4) impose con- lattice Dirac-semimetal model similar to the one dis- straints on the topological properties of spin-orbit cou- cussed above. Besides the on-site Hubbard U, we now pled systems. For example, the number of Dirac points introduce additional bond interactions, and the resulting in a 2-dimensional Dirac-semimetal must be a multiple new Hamiltonian can be expressed as of four, and the band Chern number of a Chern insulator must be an even integer. These have been embodied in x y H′ = hij + hij , (7) the above concrete models. x-bondsX y-bondsX Secondly, we notice that the SU(2) pseudospin symmetry group contains a finite subgroup, namely the charge- where the summation x-bonds ( y-bonds) runs over all conjugation symmetry (see Appendix D), which leads to the bonds along the x (Py) directionP of the square lattice. a vanishing U(1) charge hQii ≡ hni − 1i = 0 on each site, The bond Hamiltonians are given by 4

x µ U hij = t(ci†σxcj + cj†σxci) − (ni + nj )+ (ni ni + nj nj ) 4 4 ↑ ↓ ↑ ↓

+B (ciσ† cjσ¯ + cj†σ¯ ciσ)(niσ¯ + njσ)+ Vninj + P (ci† ci† cj cj + cj† cj† ci ci ) ↓ ↑ ↓ ↑ Xσ ↑ ↓ ↑ ↓ y µ U hij = t(ci†σycj + cj†σyci) − (ni + nj )+ (ni ni + nj nj ) 4 4 ↑ ↓ ↑ ↓

+B (¯σiciσ† cjσ¯ + σicj†σ ciσ)(niσ¯ + njσ)+ Vninj + P (ci† ci† cj cj + cj† cj† ci ci ), (8) ¯ ↓ ↑ ↓ ↑ Xσ ↑ ↓ ↑ ↓

where ci† = (ci† ,ci† ), ni = ni + ni , andσ ¯ = −σ. The noninteracting↑ part↓ t represents↑ SOC↓ which is respon- TABLE I. This table summarizes the 16 local eigenstates and their respective eigenvalues of the bond Hamiltonians (8), for sible for the Dirac points in momentum space. The B 43 B = −t. The 16 local bases are defined as follows: |00i term in Eq.(8) is known as the bond-charge interaction , † † † † † † denotes the vacuum, |22i = ci↑ci↓cj↑cj↓|00i, |20i = ci↑ci↓|00i, which completely breaks the SU(2) spin-rotation symme- † † ′ † † † 44 |02i = cj↑cj↓|00i, |σσ i = ciσcjσ′ |00i, |σ0i = ciσ|00i, |0σi = try and can be thought of as due to SOC . Note that it † † † † † † † differs from the usual SU(2) spin symmetric bond-charge cjσ|00i, |σ2i = ciσcj↑cj↓|00i, and |2σi = ci↑ci↓cjσ |00i. interaction (c† cjσ + c† ciσ )(niσ + njσ ) by spin flips x y σ iσ jσ ¯ ¯ hij hij on the bonds.P In addition, the terms V and P denote Eigenvalue Eigenstate Eigenstate the nearest-neighbor Coulomb interaction and the pair- 0 |00i |00i U hopping term, respectively. 2 − µ + 4V |22i |22i U µ 4 − 2 ± P |20i ± |02i |20i ± |02i µ ′ ′ We note that the η-pairing operator with momentum π V − 2 |σσ i |σσ i µ can be an eigenoperator of our extended Hubbard model ±t − 4 |σ0i ± |0¯σi |σ0i± σi|0¯σi U 3µ with SOC, and hence Eq.(7) exhibits an exact pseu- ±t + 4 − 4 + 2V |σ2i ∓ |2¯σi |σ2i∓ σi|2¯σi dospin symmetry with fine-tuned parameters, as shown in Appendix F. In the absence of the bond interactions (B = V = P = 0), Eq.(7) reduces to the spin-orbit values, which yields the following constraints coupled Hubbard model (1), and one can show that, us- P U ing the generalized η-pairing theory discussed above, the V = − < 0, µ = +4V η-pairing operator with zero momentum is an eigenoper- 2 2 ator. U < min(−8|t|− 8V, −4V ). (9)

To construct the exact ground states of Eq.(7), the Therefore, inside this parameter region together with B = −t, the exact ground states of our interacting Dirac- basic idea is to identify some parameter regions where the m Hamiltonian (7) is frustration-free: The ground states of semimetal model (7) are (ηπ† ) |0i (m =0, 1, ..., N). The zero temperature ODLRO for this phase can be read- Eq.(7) are simultaneous ground states of each and every † † x,y c c cj cj m local bond Hamiltonian h in Eq.(8). h i↑ i↓ ↓ ↑i ij ily evaluated: hci† ci† cj cj iT =0 = m N+1 = ↑ ↓ ↓ ↑ 1 iπ (j i) Pm(N m) iπ (j i) e · − , where hci† ci† cj cj im = − e · − is 6 ↓ ↑ N(N 1) To find the parameter region (i.e., restrictions on U, ↑ ↓ −m 1 B, V , and P in the form of equalities and inequalities) the ODLRO for each ground state (ηπ† ) |0i . It is where the η-pairing states are frustration-free ground this non-vanishing ODLRO hci† ci† cj cj iT =0 at large dis- ↑ ↓ ↓ ↑ states, let us diagonalize the local bond Hamiltonians tance |i−j| → ∞ that characterizes the superconducting x,y 45 (8). Here, each bond Hamiltonian hij has 16 local eigen- phase inside the region of Eq.(9). This phase breaks states and their respective energies, which are summa- the U(1) particle-number symmetry, say, the number m rized in Table I. Notice that we have set B = −t in the of electrons 2m in each ground state (ηπ† ) |0i is not calculation3,4,6,42. For B 6= −t, the local eigenstates be- fixed with changing m. In addition, the lattice transla- come complicated and it is difficult to see whether there tion symmetry is also broken due to the nonzero pair- exists a global frustration-free ground state42. ing momentum π: Each ground state transforms as m m m (ηπ† ) |0i → (−1) (ηπ† ) |0i under the translation by one We note that the η-pairing states with momentum lattice spacing along the x or y direction, but the factor m m π = (π, π) , i.e., (ηπ† ) |0i, can be built completely from (−1) depends on the parity of m. We thus conclude m the local states |00i, |22i, and |20i−|02i. Thus, (ηπ† ) |0i that the above superconducting phase is in fact a PDW will be the common ground states of all the bond Hamil- phase. x y tonians hij and hij , if |00i, |22i, and |20i − |02i are In the same way, one can establish the parameter re- local ground states. This requires that, from Table I, gion in which the η-pairing states with momentum 0 be- U U µ m 0 = 2 − µ +4V = 4 − 2 − P are the minimum eigen- come the exact ground states. (η0† ) |0i is now built com- 5 pletely from the local states |00i, |22i, and |20i + |02i, A complementary approach to confirm this is by check- which have to be made local ground states. Again from ing the analyticity of ground-state energy near the line Table I, we get the following constraints (12). As can be seen from Table I, the ground-state P U energy (per bond) vanishes continuously from each of V = < 0, µ = +4V the three phases toward the line (12). However, its first 2 2 min derivatives (with respect to, e.g., U and µ) have a jump U < (−8|t|− 8V, −4V ). (10) at line (12), namely a kink, implying a quantum phase Similar to the η-pairing states with momentum π, the transition at this line. phase inside this region also possesses a non-vanishing More interestingly, the multicritical line (12) is also an 1 ODLRO hci† ci† cj cj iT =0 = and is thus a supercon- exactly solvable line: Within this regime, we see that, ↓ ↑ 6 ducting phase.↑ ↓ The global U(1) symmetry is sponta- from Table I, the local ground states are |00i, |20i, |02i, neously broken, due to the fluctuating electron numbers and |22i. In this case, the ground-state space of Eq.(7) m li in the degenerate ground states (η0† ) |0i. However, this is spanned by the bases i(ci† ci† ) |0i (li = 0, 1) and phase preserves the lattice translation symmetry since is highly degenerate, whichQ contains↑ ↓ both the η-pairing the pairing momentum is 0. So it is a spatially uniform states (with arbitrary momentum) and the CDW states. superconductor. In fact, as shown in Appendix C, it is closely related to the conventional s-wave superconductor. It is interesting to see that the CDW state |CDWi = VI. EXPERIMENTAL CONSIDERATIONS i ci† ci† |0i can also be an exact ground state, where AQ ∈Adenotes↑ ↓ one sublattice of the square lattice. Notice that |CDWi can be constructed completely from the local A few remarks are now in order concerning possible ex- states |20i and |02i, we get the following constraints perimental realizations of our results in optical lattices. P =0 The essential ingredients of the present study are SOC U < min(−4|t| + µ, 4V, 2µ) and interaction. In general, the SOC can be implemented artificially in an optical lattice by introducing synthetic |t| µ U µ V > max( + , − + ). (11) gauge fields47 and the interaction can be widely tuned 2 8 16 8 through Feshbach resonances48,49. In particular, con- The CDW phase obviously breaks the lattice translation crete schemes to realize spin-orbit coupled Hubbard mod- symmetry, accompanied by a charge density modulation els with controlled band structures and interactions have with the same periodicity as that of the PDW phase. We been proposed50,51. In fact, the Fermi-Hubbard model notice that some important aspects of the physics of this has been realized in an optical lattice52–54, where the ki- simple CDW order in topological semimetals have been netic energy, the interaction strength, and the chemical explored recently46. potential can be varied independently55–60. Given these, Eqs.(9)-(11) are exact results that establish the stabil- it is plausible that the SU(2) pseudospin symmetry can ity regions of superconducting (i.e., the η-pairing states) be realized via fine tuning of the model parameters, e.g., and CDW ground states. We thus see that, for suffi- µ or U so that µ = U/2 as discussed previously, in cold- ciently small U, superconducting ground states are sta- atom systems47–49,61. bilized by attractive V and finite P , whereas a repulsive On the other hand, the exactly solvable interacting V (without P ) favors a CDW ground state. Dirac-semimetal model (7) contains off-site interactions, such as the B, V , and P terms in Eq.(8), which can arise from the matrix elements of the low-energy effective po- V. IDENTIFICATION OF AN EXACT 62,63 MULTICRITICAL LINE tential between fermionic atoms in an optical lattice . Recent experiments in cold fermionic Rydberg atoms in an optical lattice have achieved a high degree of Given the three exactly solvable regions (9)-(11), it control over these interactions, such as adjusting the is desirable to determine the phase transitions between sign, strength, and range of the interactions48,49,61,64–68. these three symmetry-breaking phases. We note that the Therefore, the ability to control and tune interactions of no three regions (9)-(11) have overlaps with each other, extended Hubbard models in optical lattices has paved however, overlaps exist in their boundaries. Specifically, the way for the realization of Hamiltonian (7) over a wide as shown in Appendix G, any two of the three boundaries range of parameters. Moreover, the η-pairing states with of regions (9)-(11) share a common phase boundary, say zero momentum are closely related to the s-wave super- B = −t, V = P =0,U =2µ ≤−8|t|. (12) conductivity (see Appendix C), whose experimental ev- idence in an optical lattice has already been established As a consequence, Eq.(12) is in fact an exact multicrit- and well studied69,70. We are therefore optimistic that ical line, where the PDW, uniform superconductor, and the exact η-pairing ground states of Eq.(7) could be re- CDW phases meet. alized in optical lattices in the near future. 6

VII. SUMMARY AND DISCUSSION Now, let us consider the ﬁrst summation term S =

k αkck† cQ† k in Eq.(A2), and we have the following In summary, we have generalized Yang’s η-pairing the- Pidentities↓ − ↓ ory to a class of Hubbard models with SOC, where a S = − αkcQ† k ck† hidden SU(2) symmetry is uncovered. We apply these − ↓ ↓ Xk exact results to four concrete spin-orbit coupled Hub- ′ bard models, which include three gapless and one gapped S = αQ k cQ† k′ ck† ′ = αQ kcQ† k ck† ,(A3) − − ↓ ↓ − − ↓ ↓ topological phases in the noninteracting limit. We then Xk′ Xk construct and focus on an interacting Dirac-semimetal where the first line follows from the anti-commutation exact model and establish the stability regions for the relation of fermion operators, and the substitution k = π 0 ′ η-pairing ground states (with momentum or ) and Q − k is used in the second line. As can be seen from the exact CDW ground states. Between these distinct Eq.(A3), if αk = αQ k, then we have S = −S and hence symmetry-breaking phases, there exists an exactly solv- S = 0. The second− summation term in Eq.(A2) also able multicritical line. vanishes in the same way when αk = αQ k. Similarly, The present work may provide new guidance in explor- the third and the fourth summation terms− in Eq.(A2) ing exotic quantum phases, for instance a PDW phase vanish separately when βk = βQ k. Finally, the last here, in a wider range of physical systems, such as spin- − summation term in Eq.(A2) is equal to λ k ck† cQ† k = orbit coupled systems with SU(2) pseudospin symmetry. ↑ − ↓ The exactly solvable interacting Dirac-semimetal model ληQ† , if εk +εQ k +γk −γQ k = λ, wherePλ is a constant − − we constructed deserves further studies: This model independent of k. hosts various exact symmetry-breaking ground states and In conclusion, when αk = αQ k, βk = βQ k, and εk + − − an exact multicritical line between them, which should εQ k + γk − γQ k = λ, i.e., Eq.(4), we have [H0, ηQ† ] = serve as a clean platform to study intertwined orders8 and − − λη† . quantum phase transitions between the Dirac-semimetal Q and symmetry-breaking phases10,71–75. Appendix B: Implications of η-pairing theory on the topological properties of spin-orbit coupled systems ACKNOWLEDGMENTS Based on our generalized η-pairing theory, in the fol- I am especially grateful to Jian-Xin Li and Zhenyu lowing we will show that Eq.(4) can have nontrivial re- Zhang for helpful discussions and encouragement during strictions on the properties of band topology of spin-orbit the course of this work. I also thank Meng Cheng, Zhao- coupled systems. Long Gu, Zhao-Yang Dong, Feng Tang, Rui Wang, and Since we are interested in the topological properties, Xiaoyu Zhu for enlightening conversations. This work the usual energy dispersion term εkσ0 in H0 can be was supported by the NSF of China under Grant No. ignored because it does not affect the band topology. 11704341 and the Open Research Program of the Labora- Eq.(4) is thus reduced to tory of Solid State Microstructures (Nanjing University) under Grant No. M30020. αk = αQ k, βk = βQ k, γk − γQ k = λ =0, (B1) − − − where the last equation vanishes because λ is independent Appendix A: Derivation of Eq.(4) in the main text Q of k and we can take k = 2 such that γ Q − γQ Q = 2 − 2 γ Q − γ Q = 0. Eq.(B1) implies that the corresponding Here, we show how to obtain Eq.(4) in the main text. 2 2 Bloch Hamiltonian H(k)= α σ + β σ + γ σ has the We first rewrite the noninteracting part H in Eq.(1) as k x k y k z 0 property

H0 = [αk(ck† ck + ck† ck )+ βk(−ick† ck + ick† ck ) ↑ ↓ ↓ ↑ ↑ ↓ ↓ ↑ H(k)= H(Q − k), (B2) Xk

+(εk + γk)ck† ck + (εk − γk)ck† ck ]. (A1) and the Bloch wave-functions also satisfy ψ(k)= ψ(Q − ↑ ↑ ↓ ↓ It is now convenient to use the momentum space expres- k) (up to an unphysical phase factor). From the point of view of topology, this implies that each point k is sion of η-pairing operator ηQ† = k ck† cQ† k to calcu- ↑ − ↓ topologically identical to the point Q − k in momentum late its commutator with H0. AfterP a direct and lengthy calculation, we get the following result space. Let us now focus on 2-dimensional systems. If we Q [H0, ηQ† ]= [αkck† cQ† k − αQ kcQ† k ck† translate the Brillouin zone (BZ) by vector 2 and choose ↓ − ↓ − − ↑ ↑ Xk its center as the new origin of coordinate in momentum space, then the above arguments mean that each point +iβkck† cQ† k + iβQ kcQ† k ck† ↓ − ↓ − − ↑ ↑ k is topologically identical to the point −k. We now +(εk + εQ k + γk − γQ k)ck† cQ† k ](A2). bipartite the BZ into two halves, then each half BZ is − − ↑ − ↓ 7 topologically identical to a sphere which is a closed man- tors Jx,Jy and Jz. The global pseudospin-rotation by 36 iπJy ifold (See Appendix A 3 in Ref. for the proof). π about the Jy axis, say C = e , ﬂips the sign of 1 Therefore, we arrive at the following conclusions: Jz: CJzC− = −Jz. We recall that the pseudospin op- (1). If the spin-orbit coupled system is a 2-dimensional erator Jz is proportional to the U(1) charge operator Dirac-semimetal, then the number of Dirac points must ˆ ˆ 1 ˆ Q ≡ Ne − N, say Jz = 2 Q. Therefore, the pseudospin- be a multiple of four. rotation C = eiπJy is in fact a charge conjugation14,77, Proof: Firstly, the two half BZs host equal number i.e., of Dirac points due to the symmetry property of Bloch 1 Hamiltonian, e.g., Eq.(B2). Secondly, each half BZ is a CQˆC− = −Q.ˆ (D1) closed manifold where the Dirac points must appear in pairs. In fact, C = eiπJy ﬂips the sign of the local operator (2). If the system is a Chern insulator, then the band z 1 1 Ji ≡ 2 Qi ≡ 2 (ni − 1) on each site. So, if the model Chern number must be an even integer. [See Appendix Hamiltonian has the SU(2) pseudospin symmetry (and 36 A 3 in Ref. for the proof.] hence the charge-conjugation symmetry), then hQii = z hJi i = 0 on each site, where hAi denotes the ensemble average of the operator A at any temperature. Appendix C: The η-pairing states are topologically It is interesting to see that the charge-conjugation op- trivial erator C squares to the charge parity77

ˆ In this section, we argue that the η-pairing states, C2 = (−1)Q, (D2) m e.g., (η0† ) |0i (m = 0, 1, ..., N), are closely related to the conventional BCS superconductivity and hence 2 i2πJy i2πJz 2Jz 1 ˆ since C = e = e = (−1) and Jz = 2 Q. they are topologically trivial. Specifically, let us con- An alternative way to understand Eq.(D2) is to make sider the s-wave BCS mean-field Hamiltonian HBCS = an analogy between the charge-conjugation C and the kσ ξkck† σ ckσ + k ∆k(ck† c† k +c k ck ), where ∆k < ”time reversal” T . Specifically, let us focus on the f ↑ − ↓ − ↓ ↑ f P0. The intra-pairP distance between the two electrons in fermions introduced by a particle-hole transformation for the usual Cooper pair is finite, whereas it is zero (i.e., 1 the spin-down electrons and we have Jα = f †ταfj , on-site singlet pairing) in the η-pairing states. There- 2 j j as shown in the main text. The charge-conjugationP C = fore, the ground state of HBCS is in general not a linear eiπJy rotates the spin of f fermions Cf C 1 = −f , m j − j combination of (η0† ) |0i. 1 ↑ ↓ Cfj C− = fj . The ”time reversal” Tf acts on these However, H is topologically equivalent to a simpler ↓ ↑ BCS f-fermion operators the same as C, except that Tf is an Hamiltonian 1 antiunitary operator, e.g., Tf iTf− = −i. So, we can Hη = ∆k(ck† c† k + c k ck ), (C1) define Tf to be ↑ − ↓ − ↓ ↑ Xk T = CK, (D3) since the energy spectrum remains fully gapped during f the deformation ξ → 0 and there should be no topologi- k where K denotes the complex conjugation operator. To cal phase transition. The ground state wave function (up be precise, consider the f-fermion occupation number to a normalization factor) of H is now readily written η N lj↑ lj↓ N 1 m basis j=1(fj† ) (fj† ) |0i, where |0i is the vacuum as |Gi = k(1+ck† c† k )|0i = m=0 m! (η0† ) |0i, which ↑ ↓ ↑ − ↓ stateQ of f fermions and ljσ ∈ {0, 1}. Now the op- m 76 is a superpositionQ of η-pairingP states (η0† ) |0i . Since erator K is defined in this basis, e.g., the base kets m N HBCS is topologically trivial, Hη and hence (η0† ) |0i are lj↑ lj↓ j=1(fj† ) (fj† ) |0i are invariant under the action of also topologically trivial. KQ. In this↑ representation,↓ the charge-conjugation C = An alternative way to see that the η-pairing states are eiπJy acts as topologically trivial may be as follows. As shown in the N main text, the η-pairing states can be mapped to the iτ 0 ⊗ C = y , (D4) ferromagnetic states through a local particle-hole trans- 0 I formation. A local transformation should not change the global topological property of a state. Therefore, the where τy is the second Pauli matrix and I is the 2 × 2 η-pairing states are topologically trivial since the ferro- identity matrix. The unitary operator C thus turns out to magnetic states are topologically trivial. be a real orthogonal matrix in the above representation, 2 2 say C∗ = C. Using Eq.(D3), we have Tf = CC∗ = C . Now we recall that the ”time reversal” operator Tf Appendix D: Charge-conjugation symmetry as a 2 Nˆf subgroup of the SU(2) pseudospin symmetry squares to the fermion number parity Tf = (−1) , with ˆ Nf = j,σ fjσ† fjσ. We finally conclude that As shown in the main text, the generators of the SU(2) P ˆ pseudospin symmetry group are the pseudospin opera- C2 = (−1)Nf . (D5) 8

Note that Eq.(D5) is indeed equivalent to Eq.(D2), Eq.(E2), say ˆ ˆ ˆ since Q and Nf have the same parity: Nf = −2ρ ˆ D+(ω)= , (E3) 2Jz + 2 j fj† fj = Q + 2 j fj† fj , where the term ω + (λ + U)+ iδ ↓ ↓ ↓ ↓ 2 j fj† Pfj always has an evenP parity. where δ is a positive inﬁnitesimal. Similarly, we have ↓ PFor the↓ sake of completeness, we outline the action of i D (t,t′)= − Θ(t − t′)h[J (t), ∆z† (t′)]i the charge-conjugation operator C on the original elec- − N − tron operators, say 2ρ D (ω)= ∗ , (E4) − ω − (λ + U)+ iδ 1 iQ j CcjσC− = −σe · cj,† σ, (D6) − and for which reason C is also called particle-hole transforma- i Dz(t,t′)= − Θ(t − t′)h[Jz(t), ∆+(t′)]i tion. N ρ D (ω)= . (E5) z ω + iδ Appendix E: Implications of η-pairing theory on the Therefore, as can be seen from Eqs.(E3), (E4), and superconducting properties in Hubbard models with (E5), if the ground state of our spin-orbit coupled Hub- SOC bard model is in the superconducting phase (i.e., ρ 6= 0), then there must exist a triplet of collective modes with energies −(λ + U), λ + U, and 0, respectively. Note that Following Zhang’s idea in Ref.2, we will show that the SU(2) pseudospin symmetry of our spin-orbit coupled the collective mode with energy 0 in Eq.(E5) is simply the gapless Goldstone mode resulting from the spontaneous Hubbard models would lead to a pair of new collective modes together with the usual Goldstone mode, if the U(1) symmetry breaking of superconductivity. In addition, we ﬁnd that there are a pair of new collective modes system is in the superconducting phase. with energies ±(λ + U), which is also a consequence of The crucial point in Zhang’s observation is the fact superconductivity. that the transverse components of the pseudospin operators look like a superconducting order, while the z component looks like a CDW order, up to a local phase Appendix F: The SU pseudospin symmetry of iQ j (2) factor e · . So the multiplet of the particle-particle and Eq.(7) in the main text particle-hole operators should transform as a vector under the pseudospin rotation As shown in the main text, the η-pairing operator can be an eigenoperator of a class of Hubbard models with [J+, ∆z]= −2∆+, [J , ∆+]= −∆z, [Jz, ∆+] = ∆+, − SOC. In fact, the η-pairing operator can also be an eigen- (E1) operator of an extended Hubbard model with SOC, such rotate meaning that the pseudospin operators can the su- as the interacting Dirac-semimetal model Eq.(7) we stud- perconducting order and the CDW order into each other, ied in the main text. where the pseudospin operators J+ = Jx + iJy = ηQ† , To see this, we calculate the commutator of the Hamil- 1 ˆ J = J+† , Jz = (Ne − N), the s-wave Cooper-pairing tonian (7) and the η-pairing operator ηπ† , which reads − 2 operator ∆+ = j cj† cj† , ∆ = ∆+† , and the CDW iπ j − [H′, η† ]= −2(t + B) e · (c† c† − c† c† ) i↑Q j↓ π i j i j operator ∆z = Pjσ e− · njσ with the wave-vector Q. x-bondsX ↑ ↑ ↓ ↓ To study theP collective modes, let us consider the fol- iπ j +2i(t + B) e (c† c† + c† c† ) lowing response function · i j i j y-bondsX ↑ ↑ ↓ ↓ i iπ j +(2V + P ) e · (cj† cj† ni − ci† ci† nj ) D+(t,t′)= − Θ(t − t′)h[J+(t), ∆z(t′)]i. (E2) ↑ ↓ ↑ ↓ N Xi,j h i Since the Hamiltonian is time-independent, we have +(U − 2µ − 4P )ηπ† . (F1) i D+(t,t′) = D+(t − t′) = − N Θ(t − t′)h[J+(t − t′), ∆z]i. Therefore, the η-pairing operator ηπ† becomes an eigenop- m From Eq.(6) in the main text, the time dependence erator of H′ and (ηπ† ) |0i become exact eigenstates with of pseudospin operators in the Heisenberg representa- P eigenvalues m(U − 2µ − 4P ), when B = −t, V = − 2 . tion can be determined explicitly, in particular, J (t − P U + Moreover, if B = −t, V = − , and µ = +4V , H′ i(λ+U)(t t′ ) 2 2 t′)= e − J+. Using the commutator in Eq.(E1), commutes with ηπ† and hence possesses the SU(2) pseu- Eq.(E2) can be thus calculated exactly, i.e., D+(t − t′)= ′ dospin symmetry. Accordingly, within the regime of the i(λ+U)(t t ) ∆+ m 2iρΘ(t − t′)e − , where ρ = h N i denotes the su- PDW phase where (ηπ† ) |0i are the exact ground states perconducting order parameter. (see Eq.(9) in the main text), the corresponding Hamil- Information about collective modes are now embed- tonian (7) exhibits an exact pseudospin symmetry which ded in the analytic properties of the Fourier transform of gets spontaneously broken. 9

Appendix G: Derivation of Eq.(12) in the main text

Eq.(12) describes the common phase boundary shared by the three exactly solvable regions (9), (10), and (11), which respectively host the PDW, uniform superconductor, and CDW phases. To obtain Eq.(12), we need ﬁrst to work out the boundary of each region. From Eq.(9), we see that its boundary ∂PDW consists of two disjoint sets

P U ∂PDW = {V = P =0,U =2µ ≤−8|t|} ∪ {V = − < 0,µ = +4V,U = min(−8|t|− 8V, −4V )}. (G1) 2 2 Similarly, the boundary of Eq.(10) reads

P U ∂SC = {V = P =0,U =2µ ≤−8|t|} ∪ {V = < 0,µ = +4V,U = min(−8|t|− 8V, −4V )}. (G2) 2 2 Lastly, the boundary of Eq.(11) reads |t| µ U µ ∂CDW = {P =0,U ≤ min(−4|t| + µ, 4V, 2µ), V = max( + , − + )} 2 8 16 8 |t| µ U µ ∪{P =0,U = min(−4|t| + µ, 4V, 2µ),V > max( + , − + )}. (G3) 2 8 16 8 Note that we have omitted B = −t in all of the above equations. From these equations, we ﬁnd that any two of the boundaries ∂PDW, ∂SC, and ∂CDW have equal intersections

∂PDW ∩ ∂SC = ∂SC ∩ ∂CDW = ∂CDW ∩ ∂PDW = {B = −t, V = P =0,U =2µ ≤−8|t|}. (G4)

As a consequence, Eqs.(9),(10),(11) in the main text share a common phase boundary

∂PDW ∩ ∂SC ∩ ∂CDW = {B = −t, V = P =0,U =2µ ≤−8|t|}, (G5) which is a multicritical line between these distinct symmetry-breaking phases.

∗ [email protected] 3, 043 (2017). 1 C. N. Yang, Phys. Rev. Lett. 63, 2144 (1989). 17 R. Nandkishore and D. A. Huse, Annual Review of Con- 2 S. Zhang, Phys. Rev. Lett. 65, 120 (1990). densed Matter Physics 6, 15 (2015). 3 F. H. L. Essler, V. E. Korepin, and K. Schoutens, Phys. 18 J. R. Garrison, R. V. Mishmash, and M. P. A. Fisher, Rev. Lett. 68, 2960 (1992). Phys. Rev. B 95, 054204 (2017). 4 F. H. L. Essler, V. E. Korepin, and K. Schoutens, Phys. 19 T. Veness, F. H. L. Essler, and M. P. A. Fisher, Phys. Rev. Lett. 70, 73 (1993). Rev. B 96, 195153 (2017). 5 S.-Q. Shen and Z.-M. Qiu, Phys. Rev. Lett. 71, 4238 20 X. Yu, D. Luo, and B. K. Clark, Phys. Rev. B 98, 115106 (1993). (2018). 6 J. de Boer, V. E. Korepin, and A. Schadschneider, Phys. 21 S. Moudgalya, N. Regnault, and B. A. Bernevig, Phys. Rev. Lett. 74, 789 (1995). Rev. B 102, 085140 (2020). 7 S.-C. Zhang, Science 275, 1089 (1997). 22 D. K. Mark and O. I. Motrunich, Phys. Rev. B 102, 075132 8 E. Fradkin, S. A. Kivelson, and J. M. Tranquada, Rev. (2020). Mod. Phys. 87, 457 (2015). 23 D. Pesin and L. Balents, Nat. Phys. 6, 376 (2010). 9 B. Roy and M. S. Foster, Phys. Rev. X 8, 011049 (2018). 24 W. Witczak-Krempa, G. Chen, Y. B. Kim, and L. Balents, 10 R. Boyack, H. Yerzhakov, and J. Maciejko, Annu. Rev. Condens. Matter Phys. 5, 57 (2014). arXiv:2004.09414 (2020). 25 J. G. Rau, E. K.-H. Lee, and H.-Y. Kee, Annu. Rev. Con- 11 A. Szabo and B. Roy, arXiv:2009.05055 (2020). dens. Matter Phys. 7, null (2016). 12 K.-S. Kim, Phys. Rev. Lett. 97, 136402 (2006). 26 J. E. Roberts, Journal of Mathematical Physics 7, 1097 13 K.-S. Kim, Phys. Rev. B 75, 245105 (2007). (1966). 14 M. Hermele, Phys. Rev. B 76, 035125 (2007). 27 J. E. Roberts, Comm. Math. Phys. 3, 98 (1966). 15 H. Fan and S. Lloyd, Journal of Physics A: Mathematical 28 I. Aﬄeck and J. B. Marston, Phys. Rev. B 37, 3774 (1988). and General 38, 5285 (2005). 29 E. Dagotto, E. Fradkin, and A. Moreo, Phys. Rev. B 38, 16 O. Vafek, N. Regnault, and B. A. Bernevig, SciPost Phys. 2926 (1988). 10

30 X.-G. Wen, Phys. Rev. B 65, 165113 (2002). 57 M. F. Parsons, A. Mazurenko, C. S. Chiu, G. Ji, D. Greif, 31 R. Jackiw, Phys. Rev. D 29, 2375 (1984). and M. Greiner, Science 353, 1253 (2016). 32 F. D. M. Haldane, Phys. Rev. Lett. 61, 2015 (1988). 58 M. Boll, T. A. Hilker, G. Salomon, A. Omran, J. Nespolo, 33 A. Melikyan and Z. Teˇsanović, Phys. Rev. B 76, 094509 L. Pollet, I. Bloch, and C. Gross, Science 353, 1257 (2016). (2007). 59 L. W. Cheuk, M. A. Nichols, K. R. Lawrence, M. Okan, 34 W. Witczak-Krempa, M. Knap, and D. Abanin, Phys. H. Zhang, E. Khatami, N. Trivedi, T. Paiva, M. Rigol, and Rev. Lett. 113, 136402 (2014). M. W. Zwierlein, Science 353, 1260 (2016). 35 B. Roy, Phys. Rev. B 96, 041113 (2017). 60 P. T. Brown, D. Mitra, E. Guardado-Sanchez, P. Schauß, 36 K. Li, S.-L. Yu, Z.-L. Gu, and J.-X. Li, Phys. Rev. B 94, S. S. Kondov, E. Khatami, T. Paiva, N. Trivedi, D. A. 125120 (2016). Huse, and W. S. Bakr, Science 357, 1385 (2017). 37 K. Li, S.-L. Yu, and J.-X. Li, New J. Phys. 17, 043032 61 I. Bloch, J. Dalibard, and W. Zwerger, Rev. Mod. Phys. (2015). 80, 885 (2008). 38 H. Zhai, Phys. Rev. B 71, 012512 (2005). 62 F. Werner, O. Parcollet, A. Georges, and S. R. Hassan, 39 F.-C. Pu and S.-Q. Shen, Phys. Rev. B 50, 16086 (1994). Phys. Rev. Lett. 95, 056401 (2005). 40 R. Fujiuchi, T. Kaneko, Y. Ohta, and S. Yunoki, Phys. 63 A. Rosch, D. Rasch, B. Binz, and M. Vojta, Phys. Rev. Rev. B 100, 045121 (2019). Lett. 101, 265301 (2008). 41 M. Fava, B. Ware, S. Gopalakrishnan, R. Vasseur, and 64 X. Li and S. D. Sarma, Nature Communications 6, 7137 S. A. Parameswaran, Phys. Rev. B 102, 115121 (2020). (2015). 42 J. de Boer and A. Schadschneider, Phys. Rev. Lett. 75, 65 T. J. Carroll, K. Claringbould, A. Goodsell, M. J. Lim, 4298 (1995). and M. W. Noel, Phys. Rev. Lett. 93, 153001 (2004). 43 S. Kivelson, W.-P. Su, J. R. Schrieffer, and A. J. Heeger, 66 M. Viteau, M. G. Bason, J. Radogostowicz, N. Malossi, Phys. Rev. Lett. 58, 1899 (1987). D. Ciampini, O. Morsch, and E. Arimondo, Phys. Rev. 44 A. Montorsi and M. Rasetti, Phys. Rev. Lett. 66, 1383 Lett. 107, 060402 (2011). (1991). 67 P. Schauss, M. Cheneau, M. Endres, T. Fukuhara, S. Hild, 45 C. N. Yang, Rev. Mod. Phys. 34, 694 (1962). A. Omran, T. Pohl, C. Gross, S. Kuhr, and I. Bloch, 46 D. Sehayek, M. Thakurathi, and A. Burkov, Nature 491, 87 (2012). arXiv:2007.07256 (2020). 68 H. Labuhn, D. Barredo, S. Ravets, S. de Leseleuc, 47 J. Dalibard, F. Gerbier, G. Juzeli¯unas, and P. Ohberg,¨ T. Macri, T. Lahaye, and A. Browaeys, Nature 534, 667 Rev. Mod. Phys. 83, 1523 (2011). (2016). 69 48 C. Chin, R. Grimm, P. Julienne, and E. Tiesinga, Rev. J. K. Chin, D. E. Miller, Y. Liu, C. Stan, W. Setiawan, Mod. Phys. 82, 1225 (2010). C. Sanner, K. Xu, and W. Ketterle, Nature 443, 961 49 S. Giorgini, L. P. Pitaevskii, and S. Stringari, Rev. Mod. (2006). 70 Phys. 80, 1215 (2008). D. Mitra, P. T. Brown, E. Guardado-Sanchez, S. S. Kon- 50 L.-M. Duan, E. Demler, and M. D. Lukin, Phys. Rev. dov, T. Devakul, D. A. Huse, P. Schaub, and W. S. Bakr, Lett. 91, 090402 (2003). Nature Physics 14, 173 (2017). 71 51 W. V. Liu, F. Wilczek, and P. Zoller, Phys. Rev. A 70, L. Classen, I. F. Herbut, L. Janssen, and M. M. Scherer, 033603 (2004). Phys. Rev. B 92, 035429 (2015). 72 52 U. Schneider, L. Hackermüller, S. Will, T. Best, I. Bloch, B. Roy, V. Juricic, and I. Herbut, J. High Energy Phys. T. A. Costi, R. W. Helmes, D. Rasch, and A. Rosch, 04, 018 (2016). 73 Science 322, 1520 (2008). T. Sato, M. Hohenadler, and F. F. Assaad, Phys. Rev. 53 R. Jordens, N. Strohmaier, K. Gunter, H. Moritz, and Lett. 119, 197203 (2017). 74 T. Esslinger, Nature 455, 204 (2008). L. Janssen, I. F. Herbut, and M. M. Scherer, Phys. Rev. 54 N. Strohmaier, Y. Takasu, K. Günter, R. Jördens, M. Köhl, B 97, 041117 (2018). 75 H. Moritz, and T. Esslinger, Phys. Rev. Lett. 99, 220601 E. Torres, L. Weber, L. Janssen, S. Wessel, and M. M. (2007). Scherer, Phys. Rev. Research 2, 022005 (2020). 76 55 R. A. Hart, P. M. Duarte, T.-L. Yang, X. Liu, T. Paiva, T. Shirakawa, S. Miyakoshi, and S. Yunoki, Phys. Rev. B E. Khatami, R. T. Scalettar, N. Trivedi, D. A. Huse, and 101, 174307 (2020). 77 R. G. Hulet, Nature 519, 211 (2015). Y.-Z. You, A. W. W. Ludwig, and C. Xu, Phys. Rev. B 56 A. Mazurenko, C. S. Chiu, G. Ji, M. F. Parsons, 95, 115150 (2017). M. Kanasz-Nagy, R. Schmidt, F. Grusdt, E. Demler, D. Greif, and M. Greiner, Nature 545, 462 (2017).