arXiv:1708.00924v2 [cond-mat.stat-mech] 19 Feb 2018 eul rknit iceeone discrete sponta- a is into symmetry broken translational neously space crystals In continuous the- in symmetry. the Lorentz effective especially the low-energy break physics, the completely where solid-state ories crystals, in of role study a the physics play high-energy not in important does essentially is which try unexplored. is property the such whether has one, symmetry discrete Lorentz a into mass. continu- broken can fundamental of spontaneously symmetry some translational be origin space that the the like known symmetries or well ous is crystals, it of tran- While existence phase the ferromagnetic-paramagnetic used sition, been the has mechanism explain This to system. a of states energy few. a just superconductors topolog- topological different and and classify insulators to ical reversal used time are symmetry the chiral the particle-hole, the and interaction, n otersmer ne oain rnlto and translation rotation, under symmetry accord- reflection group their space are the to crystals and prerequisite the ing group examples, point the For the is it. by of classified system theory a a building of for symmetry re- the nature. then describing and Knowing theories laws candidate physical of number the the on duces constraint a imposes It h ona´ ymtyaetebsco h relativistic the of theory basic field the quantum are Poincar´e symmetry the aiysymmetry parity s hte h rsascnhv n fntalthe all not the constrain if to any people have helps may can One which crystals symmetry crystals. Lorentz the of theory whether a ask down writing in freedom crystals. of group symmetry the not but theory itneadLrnztasomtoso rirr veloc- arbitrary of ity arbitrary transformations of Lorentz translations and Poincar´e group temporal distance times continuous and integer The spatial of an constant. consists being lattice distance the the spe- of of some with along translation is direction spatial translation cific the arbitrary if only an but under coordinates, same the look thsbe ogblee ht h oet symme- Lorentz the that, believed long been has It low- at broken spontaneously be can symmetry The ymtypasa motn oei oenphysics. modern in role important an plays Symmetry irgrigteLrnzsmer evstomuch too leaves symmetry Lorentz the Disregarding 7 hc stesmer ru farltvsi field relativistic a of group symmetry the is which , 1 h oet ymtyadisgeneralization its and symmetry Lorentz the , iceeLrnzsmer n iceetm translational time discrete and symmetry Lorentz Discrete osbedsrt oet n iceetm translational i time broken discrete is and a symmetry Lorentz into discrete translational broken possible time be continuous also the should that fol symmetry that, Lorentz argue We conti continuous the one. the of discrete manifestation a into the broken are spontaneously Crystals nature. of tries n hwhwt ul edter raltiefil hoytha theory field lattice a or theory field a build to how show and h oet ymtyadtesaeadtm rnltoa sy translational time and space the and symmetry Lorentz The .INTRODUCTION I. 3,4 mrvsorudrtnigo weak of understanding our improves 2 1 h icvr ftevoainof violation the of discovery the , eateto hsc,Zein omlUiest,Jnu 3 Jinhua University, Normal Zhejiang Physics, of Department 6 rsa osnot does crystal A . Dtd eray2,2018) 20, February (Dated: 5 oname to , e Wang Pei o obidafil hoyo atc edter that theory field lattice a or Poincar´e symmetry. discuss discrete theory We the field to- has a Poincar´e group. translations build discrete to spatial a how and up discrete make temporal the gether discrete with compatible the transformations be and Lorentz discrete to The symmetry. one broken Lorentz discrete spontaneously trans- a be also into time should continuous symmetry the spatial lational Furthermore, discrete transforma- the under Lorentz invariant translations. are discrete they as these just tions, under should La- crystals invariant the describing theory be that effective suggest an We of of grangian group. subgroup Lorentz discrete continuous a the up corresponding make The transformations values. Lorentz discrete spe- universal some of at sequence moving Lorentz are discrete who velocities stay observers cific a laws two physical for But the same which the under exist. exist, cannot may sym- symmetry then translational space and sym- discrete Lorentz metry conse- continuous the the the contradicts that is argue what metry Lorentz We and the it. crystal, extent of a which quence in to exist study may we bro- symmetry paper spontaneously this is In symmetry can Lorentz ken. something the whether after words, left other be in them, of theories uhaboe ymtyi ubda”iecrystal”. debate ”time under a now still is to with dubbed crystals” up matter ”time is exist The symmetry there Whether broken one. discrete a a such into transla- time symmetry continuous tional the of breaking spontaneous the ainlsmer sarsl ftepicpeo relativity trans- of time principle symmetry). broken the Lorentz of of the result (the theories theory, a is our previous symmetry In lational the it But from crystals”. ”time distinguished symmetry. translational be time should continuous the of oet ymtyo pctm atc a hnbe- then has problem lattice important spacetime the an a maintain dis- come to on is how symmetry spacetime and Lorentz continuous, the of that instead assumes cretized approaches gravity Some quantum contexts. different to possi- in could dis- interesting finding be the This bly under transformations. invariant Lorentz keeps crete which rectangular centered lattice, a or spacetime rectangular a by represented be 1, n21,Wlzket Wilczek 2012, In h iceetasainlsmer htw n can find we that symmetry translational discrete The ∗ ymtisi +-iesoa spacetime, 1+1-dimensional in symmetries uu pc rnltoa ymtybeing symmetry translational space nuous oigtesaetasainlsymmetry, translational space the lowing t iceeoe eddc l the all deduce We one. discrete a nto 10–17 iceeoe hc ute implies further which one, discrete a hs symmetries. these has t mtyaefnaetlsymme- fundamental are mmetry v u hoyas rdcstebreaking the predicts also theory Our . eaiet ahother. each to relative 10,China 21004, al. 8,9 18–23 rpsdater about theory a proposed symmetry ntecua e the- set causal the In . v a nytk a take only can 2 ory, the spacetime is discretized into a random lattice by the physical laws stay the same in the coordinate systems the Poisson process, which keeps invariant under contin- which are at rest relative to each other but differ by a spa- uous Lorentz transformations statistically, in the sense tial or temporal translation of the origin. The continuous that one realization of the lattice has no Lorentz sym- Lorentz and translational transformations together make metry but the ensemble of them has18,19. In the loop up the Poincar´egroup7, which is the symmetry group of quantum gravity, the distance and the time interval are a relativistic quantum field theory. treated as quantum operators. The spacetime is dis- cretized in the sense that the corresponding operators have discrete eigenvalues. And the Lorentz transforma- The crystals do not have the continuous space transla- tion becomes a unitary transformation acting on the op- tional symmetry. The breaking of the continuous space erators, with the Lorentz symmetry being explained as translational symmetry into a discrete one signals the the invariance of the eigenvalues under unitary transfor- freezing phase transition from liquids to crystals. We mations21. In this paper, the Lorentz symmetry on the imagine the whole spacetime being occupied by a per- spacetime lattice has a different meaning. On our space- fect crystal which is infinitely large and has no bound- time lattice, the Lorentz symmetry is explicit but not sta- ary, thereafter, the spacetime has a discrete space trans- tistical. The cost is that only a discrete symmetry is left. lational symmetry for the observers living inside. We And the spacetime is classical and continuous with zero propose three hypotheses about such a spacetime. curvature (a Minkowski spacetime). We do not try to quantize the spacetime, nor consider any theory of quan- tum gravity. What we want to discuss is the symmetry The first hypothesis is a weaker version of the principle group of the Minkowski spacetime after a spontaneous of relativity. It says that, for any observer in this space- symmetry breaking. time, there exists another observer moving at nonzero The paper is organized as follows. Sec. II lists the ba- velocity relative to him and the physical laws stay the sic hypotheses of our theory. Sec. III demonstrates why same for them. The principle of relativity states that all the continuous Lorentz symmetry and the discrete space the inertial reference frames moving at arbitrary velocity translational symmetry cannot coexist, and how the lat- are equivalent to each other in describing the physical ter puts restrictions on the velocity in the Lorentz trans- laws. Our first hypothesis is different from the princi- formation. In Sec. IV, we deduce the discrete Lorentz ple of relativity which is found to contradict the discrete group that is compatible with the discrete space trans- space translational symmetry, as explained in Sec. III. lational symmetry. In Sec. V, we construct the discrete We only suppose that there exists at least one velocity Poincar´egroup which includes both the translations and v = 0 so that two reference frames are equivalent as one the Lorentz transformations. We then present a corol- is6 moving at v relative to the other. lary of our theory - the continuous time translational symmetry is broken into a discrete one. Sec. VI contin- ues to discuss the discrete spacetime translational sym- In the second hypothesis, we suppose an invariant metry, and shows how its representation (the spacetime ”light speed” which is denoted by a constant c in this lattice) keeps invariance under discrete Lorentz transfor- paper. Note that c is not the speed of light in vacuum mations. We also discuss how to understand the time since the spacetime is occupied by a crystal. Instead, c is dilation and space contraction on the lattice. Sec. VII the supremum limit of the propagation speed of informa- shows the causality between events on the spacetime lat- tion and matter in the crystal. Such a limit always exists, tice. In Sec. VIII, we construct the field theory that has even if it may differ from the speed of light in vacuum. the discrete Poincar´esymmetry. The possible features We suppose that c is the same in any reference frames. and problems when quantizing this theory are discussed in Sec. IX. Sec. X is the conclusion and outlook. The third hypothesis is that the spacetime has a dis- crete space translational symmetry. For example, in the II. HYPOTHESES OF THE THEORY 3+1-dimensional spacetime, there exist three spatial vec- tors a1, a2 and a3. The physical laws keep invariant un- Let us recall the continuous Lorentz and Poincar´esym- der a pure spatial translation of vector r if and only if metry. According to the principle of relativity in special r = n1a1 + n2a2 + n3a3 with n1,n2 and n3 being inte- relativity24, the physical laws must stay the same for the gers. Here a pure spatial translation means that the two observers in different reference frames which are moving coordinate systems are at rest relative to each other and at a constant velocity relative to each other. The trans- their temporal coordinates keep the same. a1, a2 and a3 formation connecting the space and time coordinates of are in fact the three primitive vectors of a crystal1. Note an event as measured in different frames is the Lorentz that the third hypothesis not only says that the spatial transformation, which can be derived from the principle translation of vector n1a1 + n2a2 + n3a3 is a symme- of relativity and the principle of invariant light speed. try transformation, but also says that the other spatial Again, the spacetime has translational symmetry, that is translations are not. 3

III. CONTINUOUS LORENTZ SYMMETRY It is straightforward to verify that the multiplication of AND DISCRETE TRANSLATIONAL Λ has the associative property. SYMMETRY DO NOT COEXIST Now let us consider the first hypothesis. According to it, the physical laws stay the same for two observers as A. 3+1 dimensions one is moving at some velocity v relative to the other. The value of v is not given in the first hypothesis which Any n+1-dimensional spacetime can be equipped with only states that v exists. The corresponding Lorentz the above three hypotheses. We will consider the 3+1- transformation Λ(Lv, 0) is an element of the symmetry dimensional spacetime in this subsection and then turn group of the spacetime. Note that the translation is ab- to 1+1-dimensional spacetime in next. sent in Λ(Lv, 0) because we do not yet know which trans- We consider two coordinate systems, namely Jun and lations keep the physical laws invariant. Let us consider Tao for convenience. Or one can imagine Jun and Tao as two coordinate systems K and K′, while K′ is moving two observers who are located at the origins of the corre- at v relative to K. K′ and K are then equivalent for sponding coordinate systems, respectively. Tao is mov- describing the physical laws. Since they are equivalent, ing at a constant velocity v relative to Jun. And their K is not privileged over K′. If K is equivalent to a coor- origins differ by a four-vector r = (r0, r1, r2, r3)T . The dinate system moving at v relative to itself, so must be coordinate of an event measured by Tao is denoted by K′. Therefore, the coordinate system K′′ that is moving the four-vector y′ = (y′0,y′1,y′2,y′3)T and that by Jun at v relative to K′ must also be equivalent to K′ and is denoted by y. Here the zeroth component of a four- then be equivalent to K, while the coordinate transfor- ′′ vector denotes the time coordinate and the others denote mation between K and K is Λ2(Lv, 0). Furthermore, the space coordinates. The second hypothesis says that the equivalence relation is not only transitive but also the ”light speed” is the same in any reference frames. As reflective. For K′, its equivalent partner K is moving at is well known in special relativity24, an invariant ”light the velocity v relative to it, thereafter, one coordinate speed”, whatever its value is, leads to the Lorentz trans- system moving− at v relative to another must also be formation between the coordinates of an event measured equivalent to it. The− above statements can be translated in different reference frames. The relation between y′ and into the language of group. If Λ(Lv, 0) is an element y is expressed as of any symmetry group (not necessarily the Poincar´e − group), then Λ2(Lv, 0) and Λ 1(Lv, 0) must also be the ′ y = Lvy + r, (1) elements of the symmetry group, due to the property j of a group. In fact, Λ (Lv, 0) for arbitrary integer j where Lv is the 4-by-4 matrix of Lorentz transformation must be an element of the group. By using the prod- 2 2 and r is the translation vector. Note that in the expres- uct rule Eq. (3), we obtain Λ (Lv, 0) = Λ(Lv, 0) and v −1 −1 sion of L , the speed of light in vacuum must be replaced Λ (Lv, 0) = Λ(Lv , 0) = Λ(L−v, 0). by c, i.e. the speed limit of the propagation of matter or The third hypothesis can also be expressed in the lan- information in crystals, since it is c that is invariant in guage of group. A pure translation between two co- our hypothesis. For example, if v is along the x-axis with ordinate systems can be denoted as Λ(1, r) where 1 is an amplitude v, the corresponding Lorentz matrix should the identity matrix which is the Lorentz transformation be between two coordinate systems being at rest relative 1 v/c2 to each other. We distinguish the temporal and spa- − 0 0 2 2 2 2 tial components of the four-vector r by expressing it as 1 v /c 1 v /c 0 T  − v −1  r = (r , r) with r denoting a three-dimensional spatial Lv = − 0 0 . (2) vector. Λ(1, r) with r = (0, r)T represents a pure spa-  p 2 2 p 2 2   1 v /c 1 v /c  tial translation. The third hypothesis in fact says that  −0− 0 10    Λ(1, r) is an element of the symmetry group if and only  p 0p 0 01    if r = n1a1 + n2a2 + n3a3. Since the product of two   ′ ′ Eq. (1) can be reexpressed in a compact form as y′ = translations is Λ(1, r)Λ(1, r ) = Λ(1, r + r ), the spatial Λ(Lv, r)y where Λ(Lv, r) denotes the combination of a translations of vector n1a1 + n2a2 + n3a3 by themselves Lorentz boost and a translation. The operator Λ(Lv, r) make up a discrete group. It must be a subgroup of the is not a matrix as r = 0. But one can still define the overall symmetry group of the spacetime. multiplication of Λ. Let6 us suppose the third observer Pei Up to now, we know that the symmetry group of the who is moving at a velocity v′ relative to Tao and their spacetime has an element Λ(Lv, 0) with v = 0, and its 6 origins differ by r′. The coordinate of the event measured subgroup for spatial translations contains only transla- ′′ ′′ ′ ′ by Pei is denoted by y which is y = Λ(Lv′ , r )y = tions of vector n1a1 + n2a2 + n3a3. Surprisingly, one can ′ Λ(Lv′ , r )Λ(Lv, r)y. Repeatedly applying Eq. (1) leads deduce from these properties that v cannot take contin- ′′ ′ to y = (Lv′ Lv) y + (Lv′ r + r ), the multiplication of uous values! In detail, let us consider seven observers (or two Λ operators can then be expressed as seven coordinate systems), namely K1,K2, ,K6 and ··· K7. K2 is moving at the velocity v relative to K1. K3 ′ ′ − Λ(Lv′ , r )Λ(Lv, r)=Λ(Lv′ Lv,Lv′ r + r ). (3) differs from K2 by a spatial translation of vector a1. K4 4 is moving at the velocity v relative to K3. K5 is moving spacetime. Next we focus on 1+1-dimensions in which T at the velocity v relative to K4. K6 differs from K5 by the coordinate of an event is a two-vector y = (t, x) a translation of vector a1. Finally, K7 is moving at the where t and x denote the time and space coordinates, velocity v relative to K6. Obviously, due to the transi- respectively. The reason that we choose 1+1 dimensions tivity of− equivalence relation, all these seven coordinate is due to its simplicity. Especially, the Wigner rotation25 systems are equivalent to each other. The coordinate of is lack in 1+1 dimensions so that we can easily construct an event measured by K1 and K7 is denoted by y1 and the discrete Lorentz symmetry group. The presence of y7, respectively. And we use the notation a1 = (0, a1). Wigner rotation in 2+1 and 3+1 dimensions makes the The transformation from y1 to y7 is then construction of the discrete Lorentz group more com- plicated. Also in experiments, a one-dimensional (1D) −1 2 −1 Λ (Lv, 0)Λ(1,a1)Λ (Lv, 0)Λ(1,a1)Λ (Lv, 0) crystal can be realized in quantum wires. Therefore, it −1 (4) = Λ(1,Lva1 + Lv a1). is reasonable to first explore the 1+1-dimensional space- time. The construction of the discrete Lorentz symmetry −1 Obviously, Λ(1,Lva1+Lv a1) is a translation and it must in 2+1 and 3+1 dimensions is left in future study. be an element of the symmetry group. Lv acting on Note that for 1D crystals a single real number a (the a1 = (0, a1) usually generates a four-vector with nonzero lattice constant) determines the discrete space transla- temporal component. But this temporal component ex- tional symmetry. The crystal looks the same after a −1 actly cancels the temporal component of Lv a1, so that spatial translation of distance ma with m being an ar- −1 Lva1 + Lv a1 is in fact a four-vector with only spatial bitrary integer. The generator of the spatial translation −1 components and then Λ(1,Lva1+Lv a1) describes a pure is Λ(1, a¯) wherea ¯ = (0,a)T . The third hypothesis in spatial translation with the time coordinate keeping in- 1+1-dimensional spacetime becomes that any pure spa- variant under this transformation. To see the cancel- tial translation in the symmetry group can be expressed lation between the temporal components of Lva1 and as Λ(1,ma¯). −1 Lv a1, one can study the example of Lv in Eq. (2). It The speed limit c is another important constant in our is straightforward to verify that (Lv + L−v)a1 has no theory, thereafter, it is natural to choose a = c = ~ = 1 temporal component for any v and a1. − as the unit, which will be used throughout the left paper. The element Λ(1,Lva +L 1a ) in the symmetry group 1 v 1 In 1+1 dimensions, the Lorentz matrix relating the co- is a pure spatial translation. But we already know that ordinate of an event observed in different reference frames for any spatial translation in the symmetry group the becomes translation vector must be n1a1+n2a2+n3a3. Therefore, we establish an equation 1 v − −1 T 2 2 Lva + L a = (0,n a + n a + n a ) . (5) √1 v √1 v 1 v 1 1 1 2 2 3 3 Lv =  −v 1−  , (6) − 2 2 This equation puts strong restrictions on the velocity v.  √1 v √1 v  −1 − − Since Lva1 +Lv a1 changes continuously with v, Eq. (5)   indicates that v can only take some discrete values. The where the relative velocity v is a signed real number sat- Lorentz transformation Λ(Lv′ , 0) is not an element of the isfying v 1 since we already set c = 1 to the unit of symmetry group if v′ does not satisfy Eq. (5), otherwise, velocity.| | According ≤ to the first hypothesis, there exists the third hypothesis would be violated. If two observers v = 0 so that Λ(Lv, 0) is an element of the symmetry are equivalent to each other, i.e., the physical laws stay group,6 i.e., two observers are equivalent in describing the the same for them, their relative velocity must be a so- physical laws if one is moving at the velocity v relative lution of Eq. (5). In Eq. (5), different integer arrays to the other. (n1,n2,n3) give different v. Since there are infinite num- Λ(Lv, 0) and Λ(1, ¯a) are two elements of the symme- ber of choice for (n1,n2,n3), the number of solutions of try group. As same as what we did in 3+1-dimensional Eq. (5) is also infinite. spacetime, we use Λ(Lv, 0) and Λ(1, ¯a) to construct a −1 The continuous Lorentz symmetry with a continuously symmetry transformation Λ(1,Lva¯ + Lv a¯), which will changing v contradicts the discrete space translational help us to obtain an equation of v. In detail, we suppose symmetry (the third hypothesis). If there exists any seven observers which are equivalent to each other. K2 Lorentz symmetry in the inhomogeneous spacetime of a is moving at the velocity v relative to K . K differs − 1 3 crystal, it must be a discrete symmetry. from K2 by a spatial translation of distance a. K4 is moving at the velocity v relative to K3. K5 is moving at the velocity v relative to K4. K6 differs from K5 by B. 1+1 dimensions a translation of distance a. And K7 is moving at the velocity v relative to K . The relation between the co- − 6 We have shown the contradiction between the contin- ordinates of an event observed by K7 and that by K1 is −1 uous Lorentz symmetry and the discrete space transla- y7 = Λ(1,Lva¯ + Lv a¯)y1. The symmetry transformation −1 tional symmetry in 3+1-dimensional spacetime. In fact, Λ(1,Lva¯ + Lv a¯) is a pure spatial translation, indicating −1 this contradiction exists in arbitrary n+1-dimensional that Lva¯ + Lv a¯ must be an integer times ofa ¯. We then 5

′ obtain Pei is Lv(4)Lv(3) = Lv′ where v denotes the velocity of Pei relative to Jun. By using the velocity-addition for- 0 mula in special relativity which can also be derived from L a¯ + L−1a¯ = 2 v v a Eq. (6), we find that v′ = (v(3) + v(4)) / (1 + v(3)v(4)).  2  √1 v ′ − However, 2/√1 v′2 is not an integer, so that v cannot  0  − = m , (7) be an equivalence velocity because this violates the third a hypothesis. Therefore, the assumption of v(m = 3) and   v(m = 4) being both the equivalence velocity must be which can be further simplified into false. 2 A question then arises as to which Lorentz transfor- = m, (8) √1 v2 4 mations Λ(Lv, 0) with the velocity v = 1 can − ± − m2 where m = 2, 3, 4, is an integer larger than one. In be in the symmetry group which must ber closed under ··· a 1+1-dimensional spacetime that obeys our three hy- multiplication. The answer appears to be simple. First, potheses, the physical laws stay the same for two ob- the identity transformation at v =0or m = 2 must be an servers only if the relative velocity between them is element of the group. For the other elements, we strictly 4 prove that (see App. A for the detail) only the transfor- v = 1 . m = 2 corresponds to v = 0, that is two ± − m2 mations that are generated by a single integer g> 2 can observersr are at rest relative to each other. And v in- make up a group. All the Lorentz transformations in the | | creases monotonically with m. Since m has no supremum group can be expressed as Λ(Lvj (g), 0) with the relative limit, v can take infinite number of values even if v can- velocity being not exceed the speed limit c = 1. Let us list a few possible| | values of v, which are v =0, √5/3, √3/2, . Recall 4 ± ± ··· v (g)= sgn(j) 1 , (9) that the unit of v is c, and then the smallest nonzero j 2 s − mj (g) value of v is √5/3c. Two observers moving at a rela- | | tive speed lower than √5/3c are always not equivalent where sgn(j) denotes the sign of the integer j. And except that they are at rest relative to each other. In mj (g) is an integer sequence generated by g. For j 0, next text, we call the relative velocity v at which two ≥ m0 = 2 and m1 = g are the first two integers in the observers are equivalent the equivalence velocity. An sequence, and the left ones are iteratively generated ac- interesting observation is that the equivalence velocity cording to is independent of a. It is the same in 1D crystals with different lattice constants. Once if the continuous space mj+1 = gmj mj−1. (10) translational symmetry is broken into a discrete one, no − matter how small a is, the equivalence velocity immedi- For j < 0, mj is obtained by using the property that ately loses its continuity. it is an even function, i.e. mj = m−j . The Lorentz

transformations Λ(Lvj (g), 0) for j = 0, 1, make up a cyclic group - the discrete Lorentz group± ··· which is de- IV. DISCRETE LORENTZ SYMMETRY noted as L. L is uniquely determined by the integer g which is called the generator of the group. The closure As shown in above, the 1+1-dimensional spacetime of the group under multiplication can be proved by using with discrete space translational symmetry can only have the relations Λ(Lvj (g), 0)Λ(Lvi(g), 0) = Λ(Lvi+j (g), 0) and j a discrete Lorentz symmetry if not none at all. Eq. (8) Λ (Lv1(g), 0) = Λ(Lvj (g), 0). The velocity-addition for- gives the necessary condition for the equivalence velocity mula reads vi+j (g) = (vi(g)+ vj (g)) / (1 + vi(g)vj (g)). v in the Lorentz transformation. But it is not the suffi- Note that the Lorentz matrix in terms of mj is expressed cient condition. In fact, it is impossible for the observers as 4 moving at the relative velocity v(m)= 1 for ar- sgn(j) 2 2 mj /2 − m 4 ±r − m 2 j − bitrary m to be equivalent to each other. In other words, Lvj =   . sgn(j) 2 q the symmetry group cannot contain all the Lorentz trans- − mj 4 mj /2  2 −  formations Λ(Lv(m), 0) for m 2, because such a set of q ≥  (11) Λ(Lv(m), 0) are not closed under multiplication! To see this point, let us suppose that Λ(Lv(m), 0) for m = 3 and Table I enumerates the first few elements in the se- m = 4 are both the elements of the symmetry group. quence mj (g) generated by g = 3 or g = 4. The cor- This is to say that three observers Jun, T ao and Pei are responding equivalence velocities are also displayed. mj equivalent to each other if T ao is moving at the veloc- increases exponentially with j, according to Eq. (10). A ity v(3) = √5/3 relative to Jun and Pei is moving at more elegant expression of mj can be found in App. B. v(4) = √3/2 relative to T ao. The Lorentz transforma- The cyclic Lorentz groups appear to be the inevitable tion relating the coordinate observed by Jun to that by consequence of our three hypotheses. In a spacetime 6

to the above discussions, the subgroup of for pure boost with no translation must be the discreteP Lorentz group m0 m1(g) m2 m3 m4 m5 m6 Λ(Lvj (g), 0) . And the subgroup of for pure spatial translation with no boost or temporalP translation must 2 3 7 18 47 123 322 be Λ(1,ma¯) witha ¯ = (0, 1)T and m being an integer (recall{ that a}= 1 is the unit of length). In other words, v0 v1 v2 v3 v4 v5 v6 two coordinate systems who have the same origins are equivalent to each other in describing the physical laws if √5 3√5 8√5 21√5 55√5 144√5 and only if they are moving at the velocity vj (g) relative 0 3 7 18 47 123 322 to each other. And two observers who are at rest relative to each other and use the same clock are equivalent if and m0 m1(g) m2 m3 m4 m5 m6 only if the spatial distance between them is an integer times of a = 1. We prove that (see App. B for the detail) the group 2 4 14 52 194 724 2702 exists and is determined by the generator g. The el- P ements of can be generally expressed as Λ(Lvj (g), Y ), v0 v1 v2 v3 v4 v5 v6 P where Lvj (g) is a Lorentz matrix of velocity vj (g) with j =0, 1, 2, being an arbitrary integer. And Y is a √3 4√3 15√3 56√3 209√3 780√3 ± ± ··· 0 discrete translation of spacetime. It is a combination of 2 7 26 97 362 1351 spatial and temporal translations, being expressed as

TABLE I. The integer sequence mj (g) generated by g = 3 1 2 0 g 4 (top) and g = 4 (bottom). The corresponding equivalence Y = N + N 2 − , (12) 1 1 2  1  velocities vj (g) are displayed below mj (g).   p g  2  where N ,N = 0, 1, 2, are arbitrary integers. 1 2 ± ± ··· P where our three hypotheses stand, when an observer is can be written as writing down the equations of physical laws, he knows = Λ L , YN N (g) j, N ,N =0, 1, 2, . that the only observers who are using the same equa- P vj (g) 1 2 1 2 ± ± ···   tions as him must be those who are moving at the equiv-
(13) 2 alence velocity 1 4/mj(g) relative to him. The The group is a discrete subgroup of the continuous other observers moving± − at different velocities have differ- P p Poincar´egroup. ent equations for the physical laws. And Λ(Lvj (g), 0) is It is easy to verify that the subgroup of for pure the transformation relating the coordinate of an event ob- boost is the discrete Lorentz group L. NotP only the served by him to those by the other equivalent observers. Lorentz boost but also the translation depends on g. It is worth emphasizing that each spacetime (each crys- The subgroup of for pure translations includes the ele- tal) has a unique generator g. But different spacetimes P ments Λ (1, YN N (g)) which are the translations of vector (different crystals) may have different generators. 1 2 YN1N2 (g). We use Y to denote the translational group Λ (1, YN N (g)) which is closed under multiplication { 1 2 } since Λ (1, Y )Λ 1, Y ′ ′ = Λ 1, Y ′ ′ . N1,N2 N1,N2 N1+N1,N2+N2 V. DISCRETE POINCARE´ SYMMETRY Y includes not only the pure spatial translation, but also the temporal translation and the
combination of spatial

and temporal translations. The vectors YN1N2 (g) form We derive Eq. (8) by studying the coordinate trans- a lattice including the origin in 1+1-dimensional space- formation between seven well-designed equivalent coor- time, which is the characteristic lattice of Y. This char- dinate systems, namely K , ,K and K . One may 1 ··· 6 7 acteristic lattice is generated by two primitive vectors: wonder whether it is possible to design some equivalent T 1 1 coordinate systems that finally lead to a paradox and (0, 1)T and g2 4, g . The first vector (0, 1)T 2 − 2 then falsifies our three hypotheses. The answer is no! In   fact, an overall symmetry group which includes the dis- corresponds to ap minimum spatial translation, while the crete Lorentz group and the discrete space translational second one corresponds to a combination of spatial and 1 1 group as its subgroups does exist. Our three hypotheses temporal translations since g2 4 and g are both are self-consistent. We will discuss this overall symmetry 2 − 2 nonzero. The subgroup of Y forp pure spatial translations group - the discrete Poincar´egroup in this section. is Λ(1,ma¯) , as we expected. The physical laws stay the same for two observers if Let{ us see} the properties of the characteristic lattice for and only if the coordinate transformation between them an odd g. At g = 3, the two primitive vectors are (0, 1)T is an element of the symmetry group of the spacetime. T Let us use to denote the overall symmetry group of the √5 3 P and , . We plot the characteristic lattice of the spacetime where our three hypotheses stand. According 2 2! 7 t t

O x O x

FIG. 1. The characteristic lattice of the translational group FIG. 2. The characteristic lattice of Y generated by g = 4. generated by g = 3. The red rectangular represents the unit The red rectangular represents the unit cell. cell of the lattice.

tively. translational symmetry for g = 3 in Fig. 1. Note that The characteristic lattice for an even g has a different the first component of the vector YN1N2 denotes the time shape. Fig. 2 plots the characteristic lattice of the trans- t which is the label of the vertical axis, while the second lational group Y generated by g = 4. For an even g, the component denotes the space x which is the label of the characteristic lattice is always a rectangular lattice. And horizontal axis. As g is an odd number, the characteristic its period in the temporal direction is g2 4a/2c. − lattice is always a centered rectangular lattice (the red The lack of the continuous time translational symme- p rectangular in Fig. 1 represents the unit cell). It includes try and then the energy conservation law sound strange, vectors that lie in the direction of t-axis. In other words, since it is generally believed that an isolated system Y includes pure temporal translations. should have conserved energy. But one should not for- The period of the characteristic lattice in temporal get that our three hypotheses stand in a spacetime where and spatial directions are incommensurate. Since the the continuous space translational symmetry has already unit of time is a/c according to our choice, the period been spontaneously broken into a discrete one. is in P of the lattice in the temporal direction is T = √5a/c for fact the symmetry group of an effective theory that de- g = 3. The pure temporal translation of an integer times scribes a system living in such an inhomogeneous space- of √5a/c is a symmetry transformation, but those of the time. Just as the electron cannot conserve its momentum other periods are not. In other words, the physical laws when moving within the periodic potential of crystals, stay the same for two observers who are at rest relative but the whole crystal as an isolated system keeps its mo- to each other and located at the same spot if and only if mentum invariant. We should understand the lack of their clocks differ by an integer times of √5a/c. Or equiv- energy conservation in the similar way as we understand alently, for a specific observer, the physical laws change the lack of momentum conservation of an electron. The periodically with time and the period is T = √5a/c. For system that the effective theory describes is not a real a general odd generator g, the period of the characteris- isolated system. tic lattice in the temporal direction is T = g2 4a/c. Following the breaking of continuous space translational− symmetry, the time translational symmetry mustp also be VI. INVARIANCE OF THE CHARACTERISTIC broken into a discrete one. This is not a surprise, since LATTICE UNDER THE DISCRETE LORENTZ the first hypothesis is a weaker version of the principle of TRANSFORMATION, TIME DILATION AND relativity, according to which the time cannot be sepa- LENGTH CONTRACTION rated from the space. According to Noether’s theorem2, the breaking of the continuous space (time) translational We already know from above that the discrete space- symmetry indicates that the momentum (energy) is not time translational symmetry can be represented by the conserved. In a spacetime where our three hypotheses characteristic lattice. In this section, we show that the stands, both the momentum and the energy are not con- characteristic lattice YN1N2 keeps invariant under a 26 { } served quantities. But according to the Bloch theorem discrete Lorentz transformation Lvj , even if each site is or the Floquet theorem27, there exist quasi-momentum transformed into another one on the same lattice. This or quasi-energy which are conserved as the system has is expected since the spacetime translational symmetry a discrete space or time translational symmetry, respec- is independent of observers. 8

t (a/c) t' (a/c) Fig. 3, respectively. Every lattice site can be expressed as A' −→ −→ x' (a) n1 OB +n2 OA with n1,n2 being arbitrary integers. 2 Notice that there are infinite ways of choosing primitive vectors. For the observer K′, the time and space axes B' ′ ′ A 1 are oriented in different directions, denoted by t and x , ′ ′ C respectively. It is not an accident that t and x cross not only the origin but also some other points (A′ and B′) O B −→ −→ -3 -2 -1 1 2 3 4 x (a) on the lattice. OA′ (the green vector) and OB′ (the blue vector) can be seen as a new pair of primitive vectors of the lattice, that is every lattice site can also be uniquely -1 −→ −→ ′ ′ ′ ′ expressed as n1 OB +n2 OA with two new integers  −→ −→ −→ ′ ′ ′ ′ n1 and n2. For example, we have OB=2 OB OA and −→ −→ −→ − OA= 3 OB′ +2 OA′. And in the reference frame K′, − −→ −→ ′ ′ √ FIG. 3. Invariance of the characteristic lattice under the the length of OB and OA is a and 3a/c, respectively, ′ ′ −→ −→ Lorentz transformation. t and x are the time and space ′ as same as the length of OB and OA in the K reference axes of a frame K moving at the velocity √3/2c relative to frame, respectively. Therefore, in the eye of K′, the char- K. The black and red numbers denote the spacetime coordi- ′ acteristic lattice is exactly the same rectangular lattice nates of an event in the frames K and K , respectively. The −→ −→ as that in the eye of K. ′ ′ green (OA ) and blue vectors (OB ) are the two primitive ′ vectors of the lattice in K . The dotted lines guides the de- Now let us discuss the time dilation and the length con- −→ −→ −→ ′ ′ ′ traction. Imagine a clock staying at rest relative to K , composition of OB into OA and OB . The pink line is the i.e., moving on the t′ axis relative to K. The world line world line of a particle that is static in K. Finally, the dashed −→ ′ ′ lines represent the light cone of O. of this clock during one period is OA with tA′ = √3a/c being the period of the time translational symmetry. But for another clock staying at rest relative to K, the event ′ Let us consider an observer K. The physical laws for A happens at the time tA′ = 2√3a/c. This reflects the him are not the same everywhere in the continuous space- fact that the clock at rest runs twice as fast as the mov- time, but change periodically. In next section, we will ing clock. In spite of the time dilation, a time interval express the physical laws in terms of the Lagrangian of a of integer periods keeps an interval of integer periods in field theory. We can then understand the physical laws any equivalent reference frames. changing as that the coupling parameters in the theory varies in the spacetime. Since the coupling parameters On the other hand, the length contraction seems to change periodically, the observer K can find a group of contradict the discrete translational symmetry at the first spacetime points which form a lattice (the characteristic sight. Let us choose two static points in the reference lattice), and on this lattice the coupling parameters stay frame K, say the points x = 0 and x = a. Because a the same. In other words, the spacetime has a discrete is the proposed period of space translational symmetry, translational symmetry represented by this lattice. Now the coupling parameters of physical laws always keep the ′ same at these two points. But in the reference frame a second observer K is moving at the velocity vj relative ′ to K. What does this spacetime lattice look like in the K , the distance between the two points contracts to eye of K′? The answer is: exactly the same! Under the a/2, which seems to cause a paradox since a is set to ′ the shortest distance for a symmetric space translation. transformation from K to K , a site YN1N2 on the lattice is transformed into There is in fact no paradox. The world lines of the two points are shown in Fig. 3, which are the t axis and the Y ′ ′ = L Y , (14) pink line, respectively. One must remember that the cou- N1N2 vj N1N2 pling parameters also change with time. Their changes which is another site on the same lattice (the proof is are synchronized in the reference frame K, but are not given in App. B where the relation between the integers ′ ′ ′ in K . The coupling parameters at the spacetime points N1, N2, N1 and N2 is presented). O and B are the same, but they are different from the Fig. 3 explains why the characteristic lattice keeps in- parameters at the point C. When the distance is mea- variant under the Lorentz transformation. We choose a sured in K′, the concerned points are O and C which are √ spacetime with g = 4 as an example and set vj = 3/2 simultaneous in K′. The spatial distance between O and which is the lowest positive equivalence velocity in this C is a/2 in K′, but the coupling parameters at these two spacetime. The characteristic lattice of g = 4 is a rectan- points are different. In previous literatures, researchers gular lattice which can be seen as created by two prim- were used to take it for granted that the Lorentz con- T −→ −→ itive vectors (0, 1)T and √3, 0 , i.e., OB and OA in traction forbids a spacetime lattice to have any Lorentz
9

t −→ −→ −→ −→ ′ ′ C' rant I). One can verify that OA, OC, OA and OC are links. In fact, at x = na for each integer n, there exists at most one site that is the end point of the link from O. D A' −→ For example, OC is the link pointing to x = a. And the arrows pointing to all the other sites above C at x = a cannot be the links because they can be decomposed into A C B' −→ −→ OC and an arrow parallel to the t-axis. Similarly, OA′ is B the unique link pointing to x =3a. And there is no link O x pointing to x =2a or x =4a. −→ −→ −→ −→ −→ −→ Among OA, OC, OA′ and OC′, OA and OC are ob- −→ −→ viously links. OA′ and OC′ are links because they can −→ −→ be obtained by Lorentz transforming OA and OC, re- −→ FIG. 4. The characteristic lattice of g = 4 as a causal set. spectively. Recall that OA′ is the primitive vector of the lattice for the observer K′ (see Fig. 3). Under the Lorentz transformation from K to K′, the primitive vec- symmetry in the traditional meaning. The above argu- −→ −→ −→ −→ tors transform as OA OA′ and OB OB′, and then ment clarifies that the Lorentz contraction can coexist −→ −→ −→ → −→ →−→ −→ with the discrete Lorentz symmetry on a proper space- OC=OA + OB transforms into OC′=OA′ + OB′. The relation ”link” is invariant under the discrete Lorentz time lattice. −→ −→ transformations. Therefore, OC′ and OA′ must also be links. VII. CAUSALITY Invariance of the relation ”link” can be proved by con- −→ tradiction. Suppose that a relation OA is a link in the The characteristic lattice is constituted of spacetime −→ frame K but not a link in K′. But OA is still a causal points (events) arranged periodically. It is therefore in- ′ teresting to discuss the causality between different events. relation in K , because the causal relation is invariant under arbitrary Lorentz transformation. We then sup- According to the causal set theory, the causal structure −→ −→ −→ of a spacetime lattice can be used to determine the ge- pose that OA can be decomposed into OX + XA with −→ −→ −→ ometry of the background manifold into which the lattice OX and XA being the causal relations in K′. But OX is embedded18. −→ 19 and XA must also be causal relations in K. Therefore, We use the symbol to denote the causal relation . −→ O and A are two events.≺ O A if and only if A is in the OA is also not a link in K, which causes a paradox. ≺ future of O and A = O. In Fig. 4, the relation O A We have infinite discrete Lorentz transformations Lv 6 ≺ j is represented by an arrow pointing from O to A. In our with j = 1, 2, 3, that keep the lattice invariant. Lvj units, the light cone of O is the 45◦ and 135◦ lines (the ··· −→ −→ for arbitrary j acting on OA and OC produces a new pair dashed lines). Therefore, an arrow is a relation if and of links. There are then infinite possible links starting only if it is along the positive-t direction and the angle from O with their end points being just above the 45◦ between it and the t-axis is less than 45◦. As is well line. known in special relativity, the causal relation is invari- ant under the Lorentz transformation. Once if O A stands, it stands in any reference frames. There are≺ infi- VIII. FIELD THEORY THAT HAS THE nite relations. One then defines the elementary relations DISCRETE POINCARE´ SYMMETRY - the links. The relation O A is a link if there exists no intervening event X so that≺ O X A. The causal structure of the spacetime lattice≺ is totally≺ determined In this section, we discuss how to construct a field the- by the links. The links are represented by the red arrows ory with the discrete Poincar´esymmetry. Such a theory −→ is expected to be the effective theory describing a sys- in Fig. 4. The blue arrow (OD) is a relation but not a −→ −→ tem in which the space translational symmetry is spon- link, because it can be decomposed into OA + AD which taneously broken into a discrete one. means O A D. We will first discuss the continuous field theory in the The characteristic≺ ≺ lattice has periodicity. All the sites subsection A. The continuous field theory is defined in a on the lattice are equivalent to each other. Therefore, we continuous spacetime, in which the time derivative and only need to find out the links starting from a specific the space derivative of the field are present in the La- point, e.g., the origin O. And the lattice has a mirror grangian density. In the subsection B, we turn to the symmetry with respect to the t axis. We then focus on lattice field theory, which is defined on the characteris- the links between the t-axis and the 45◦ line (in the quad- tic lattice of the discrete Poincar´egroup. In the lattice 10

field theory, the coupling between different lattice sites where y′ =Λy for arbitrary Λ in . But M(y) = M(Λy) take the place of the derivative operators. Both field if Λ is not in the group . P 6 theories are invariant under the discrete Poincar´etrans- Let us see how to constructP the function M(y). The formations. detailed derivation is given in App. C. Here we only give the results. M(y) is a periodic function in the 1+1- dimensional spacetime and has the same periodicity as A. Field theory the characteristic lattice of . We define two recipro- P (1) 2πg cal primitive vectors which are kµ = − , 2π Let us recall how to write down a relativistic field g2 4 ! theory that has the continuous Poincar´esymmetry. To − (2) 4π p guarantee the Poincar´esymmetry, the Lagrangian den- and kµ = , 0 . M must be expressed as the sity (y) must be a scalar (rank-0 tensor). It is made up g2 4 ! L Fourier transformation− of constants, scalar fields, or the contraction of higher- p rank tensors. For example, the Klein-Gordon Lagrangian ∞ (1) (2) µ 2 i(n1kµ +n2kµ )y density for spinless particles is expressed as M(y)= Mn1n2 e , (18) n ,n =−∞ 1 X2 1 µ 1 2 2 = ∂ φ∂µφ m φ , (15) µ L 2 − 2 where n1 and n2 are integers, y is the coordinate vec- 0 1 tor (y = t and y = x), and Mn1n2 is the coefficient where φ is a real scalar field, m is a constant, and ∂µφ and of the Fourier transformation. The Einstein summation ∂µφ are the covariant and contravariant vectors (rank-1 convention has been used in Eq. (18). Furthermore, the tensors), respectively. Here µ = 0, 1 denotes the tem- coefficients Mn1n2 should satisfy poral and spacial components, respectively. The metric M = M ′ ′ (19) signature is chosen to (+, ). The contraction of a covari- n1n2 n1n2 ant and a contravariant vectors− leaves a scalar. There- for arbitrary integer pairs (n ,n ) and (n′ ,n′ ) that have fore, the Lagrangian density (15) is a scalar which keeps 1 2 1 2 the relation invariant under arbitrary Poincar´etransformation. In ′ other words, the Lagrangian density in two coordinate n1 = zj+1n1 zj n2 ′ ′ ′ ′ − . (20) systems K and K satisfies (y)= (y ). n = zjn1 zj−1n2  2 − Let us see how to modifyL Eq. (15)L to obtain a new La- Here z is an integer sequence generated by g (see App. B grangian density that loses the continuous Poincar´esym- j for a detailed discussion about z ). For j 0, the first metry but keeps only the discrete Poincar´esymmetry . j two elements of the sequence are z = 0 and≥z = 1, and In other words, we want to construct a Lagrangian den-P 0 1 the left ones are generated according to sity that keeps invariant under a Poincar´etransformation

Λ if and only if Λ is in . There is only a single tunable zj+1 = gzj zj−1. (21) parameter in the LagrangianP density (15), which is the − 2 mass m. Alternatively, one can treat M = m as the tun- And z−j = zj is an odd function of j. According to the − able parameter. To break the continuous Poincar´esym- properties of zj, all the integer pairs (n1,n2) which are metry, we replace the constant M by a function M(y). related to each other by Eq. (20) make up an equivalence

The Lagrangian density in the coordinate system K be- class. All the coefficients Mn1n2 with (n1,n2) being in the ′ ′ comes same class must be the same. But if (n1,n2) and (n1,n2) are in different classes, M and M ′ ′ are not neces- n1n2 n1n2 1 µ 1 2 (y)= ∂ φ(y)∂µφ(y) M(y)φ (y). (16) sarily the same. For example, as g = 3, (0, 0) is by itself L 2 − 2 a class, and (1, 1), (2, 1), (1, 2), (5, 2), (2, 5), are in the same class, but (1, 3) is not in this class. Therefore,··· we To obtain the Lagrangian density in a different coordi- have M , = M , = M , = M , = M , = , but nate system K′, we need to replace y and φ by y′ and φ′, 1 1 2 1 1 2 5 2 2 5 ··· ′ ′ 1 µ ′ ′ ′ ′ M0,0, M1,1 and M1,3 may be different from each other. respectively. We obtain (y ) = ∂ φ (y )∂µφ (y ) L 2 − Eq. (18), (19) and (20) provide a method of construct- 1 M(y′)φ′2(y′). Note that φ and ∂µφ∂ φ are scalars 2 µ ing any function M(y) that is invariant under . In which keep invariant under an arbitrary Poincar´etrans- P ′ ′ µ order that M(y) is a real function, we need further re- formation, that is φ(y) = φ (y ) and ∂ φ(y)∂µφ(y) = ∗ quire M = M−n ,−n . A special example of M(y) ∂µφ′(y′)∂ φ′(y′). Therefore, ′(y′) = (y) if and only n1,n2 1 2 µ can be obtained by setting M = M¯ to a constant, if M(y) = M(y′). The LagrangianL densityL (16) keeps n1n2 that is M in different classes are all the same. The invariant under a Poincar´etransformation if and only if n1n2 corresponding M(y) is the function M(y) keeps invariant under this transforma- tion. In order that the Lagrangian density (16) has the g2 4 symmetry , we must choose a function M that satisfies M(y)= M¯ − δ x Y 1 δ t Y 0 , P 2 − N1N2 − N1N2 p N ,N X1 2 M(y)= M(y′) (17)
(22)
11 where Y 0 and Y 1 denote the temporal and the Y and Y change into Y ′ ′ and Y ′ ′ , respec- N1N2 N1N2 P1P2 Q1Q2 P1P2 Q1Q2 spatial components of the vector Y (see Eq. (12)), tively, i.e. Y ′ ′ = ΛY and Y ′ ′ = ΛY . The N1N2 P1P2 P1P2 Q1Q2 Q1Q2 respectively. M(y) is a Dirac-δ function centered at action S keeps invariant under the transformation Λ if the sites of the characteristic lattice. In the deriva- and only if inx tion of Eq. (22) we used the relation n e = 2π δ (2πN x). hP P ,Q Q = hP ′P ′ ,Q′ Q′ . (24) N − 1 2 1 2 1 2 1 2 By using the above approach, we can changeP any rela- tivisticP field theory (e.g., the theory of vector fields and For S being invariant under the group , we need the P spinor fields) into a theory that has the discrete Poincar´e coupling function h to be invariant under . As proved P symmetry . We start from a theory that keeps invari- in App. C, such a coupling function depends only upon P ant under arbitrary continuous Poincar´etransformations. the difference between the sites YP1P2 and YQ1Q2 , i.e. We then replace the constants (e.g., the coupling or the hP P ,Q Q = h(P Q , P Q ). (25) mass) in the theory into the functions like M(y). Since 1 2 1 2 1 − 1 2 − 2 M(y) keeps invariant only under , so is the new La- P Furthermore, if we use the notation YN N = YP P grangian density. 1 2 1 2 − YQ Q , or equivalently, N = P Q and N = P Q , The charge conjugation (C), the parity (P), and the 1 2 1 1 − 1 2 2 − 2 time-reversal (T) symmetries are frequently considered in the function h(N1,N2) must satisfy the study of field theories. The charge conjugation con- ′ ′ cerns the transformation in the internal space, therefore, h(N1,N2)= h(N1,N2) (26) it is independent of whether the theory has a continu- ′ ′ ous or a discrete Poincar´esymmetry. The parity and the for arbitrary integer pairs (N1,N2) and (N1,N2) that time-reversal symmetries concern the coordinate trans- have the relation formation, just like the Poincar´esymmetry. If we need ′ N1 = zj+1N1 + zjN2 the Lagrangian density (16) to have the PT symmetry, ′ . (27) N2 = zjN1 zj−1N2 we need to impose a further constraint on the function  − − M(y), that is M(y) = M( y). According to Eq. (18), Here the integer sequence z appears again. Similarly, all − j this can be realized by demanding Mn1,n2 = M−n1,−n2 . the integer pairs (N1,N2) which are related to each other The function (22) has this property. Because it is cen- by Eq. (27) make up an equivalence class. And Eq. (26) tered on the characteristic lattice, and the characteristic in fact states that h(N1,N2) in each class has a unique lattice for arbitrary g has the PT symmetry (see Fig. 1 value. One should notice the difference between Eq. (27) and 2). and (20). As (N1,N2) denotes a site on the characteristic lattice, (n1,n2) denotes a site on the reciprocal lattice. Eq. (25), (26) and (27) give the conditions of the cou- B. Lattice field theory pling function h. Equipped with a coupling function sat- isfying these conditions, the lattice field theory must have the symmetry . Sometimes we hope to simulate the field theory by P using computers and then need to discretize the space- time. It is impossible to discretize a spacetime without breaking the continuous Poincar´esymmetry. But one can IX. CONSERVATION OF QUASI-ENERGY AND maintain the discrete Poincar´esymmetry of a theory QUASI-MOMENTUM when discretizing the spacetime into theP characteristic lattice of . Because the characteristic lattice keeps in- We will not quantize the field theories in this paper, variant underP the transformations in . Next we discuss which is left for future study. But it is interesting to how to build a lattice field theory28 thatP has the symme- discuss some possible features of the quantized theory. try . Eq. (16) describes a physical system in a periodic poten- P Let us write down a general noninteracting lattice tial which varies both with time and space. The con- model whose action is sequence of such a potential is well known to condensed matter community. The electrons moving in a crystal feel S = hP P ,Q Q φP P φQ Q , (23) a typical periodic potential varying with space. While 1 2 1 2 × 1 2 × 1 2 P1P2Q1Q2 the electrons in an irradiated material are often treated X as moving in the time-periodic electromagnetic potential. where φP1 P2 denotes the value of the field φ at the lattice The momentum and energy are not conserved in the site YP1P2 , and hP1P2,Q1Q2 denotes the coupling between presence of periodic potentials. They are not good quan- the sites YP1P2 and YQ1Q2 with P1, P2,Q1 and Q2 being tum numbers any more. But according to the Floquet all integers. The characteristic lattice by itself is invari- theorem and the Bloch theorem, the quasi-energy and the ant under an arbitrary transformation Λ . But each quasi-momentum are good quantum numbers instead, lattice site changes into another one after the∈P transforma- which are defined as the energy or the momentum mod- tion. Suppose that, under a transformation Λ, the sites ulo 2π~/T or 2π~/a, respectively, where T and a are the 12 periods of the potential in the temporal and spatial di- then the discrete Lorentz group L contains only a sin- rections, respectively. Our analysis in Sec. V has already gle element Λ(1, 0) (the identity element). On the other established the relation between T and a. hand, the general version of our theory assumes g > 2, Due to the spatial periodic potential, the single- and then there exist infinite number of Lorentz transfor- particle spectrum should include a series of Bloch mations with nonzero velocities in the symmetry group. bands. The energy levels are distinguished by the quasi- Our theory involves more symmetries than the models momentum and the band label. If we further consider in previous studies. According to our theory, the La- the potential being temporally periodic, the solutions of grangian of a model can only take some specific form, the Schr¨odinger equation must be further distinguished which the previous models do not have. by the quasi-energy and the Floquet band label. The Our theory is built on three hypotheses. Recall that quasi-energy, the quasi-momentum, the Bloch band and the first hypothesis is similar to but weaker than the the Floquet band together determine a solution. principle of relativity, while the second one states an in- Recent study showed that an isolated generic Floquet variant speed c which represents the speed limit of the system will be heated up until it reaches the infinite- propagation of information or matter in crystals. These temperature state29. But we should not forget that our two hypotheses are not deduced from any known prin- three hypotheses stand in a system where the continu- ciples. Whether they are true should be examined by ous translational symmetry has been spontaneously bro- experiments. We would like to point out some results ken. The Lagrangian (16) is the effective theory for the coming out from these hypotheses that could be checked symmetry-breaking state, describing something like the by experiments. One is the breaking of the continuous electrons moving in a crystal. Therefore, the system is in time translational symmetry. In a spacetime with an odd fact an open system, to which the argument in Ref. [29] generator g, the time translational symmetry has a pe- does not apply. riod T = g2 4a/c, while in that with an even g the − 2 period is Tp= g 4a/2c. If the spacetime has a dis- crete time translational− symmetry, the local observables p X. CONCLUSIONS AND OUTLOOK should change periodically with time, just as they vary periodically within the space of the crystal lattice. Let us In summary, we propose a theory about the Lorentz estimate the magnitude of the temporal period T which and Poincar´esymmetries in a spacetime with discrete depends on a/c. The lattice constant of a crystal is typ- space translational symmetry based on three hypotheses. ically at the nanoscale. c is distinguished from the light In solid-state physics, the spacetime occupied by crystals speed in vacuum but is expected to be at the same mag- is expected to have these properties. We describe the nitude as it. We choose a = 1nm and c = 3 108m/s, whole symmetries of the spacetime which include the dis- and find a/c =3.3 10−18s. The temporal period× is only crete Lorentz, space translational and time translational a few attoseconds (too× short), which maybe explains why symmetries, and show how to construct a Lagrangian or such a periodicity has not been observed up to now. An action under these symmetries. It is worth emphasizing alternate way is to examine the absorption spectrum of that, our hypotheses and results are expected to stand crystals. According to the Floquet theorem (see Ref. [27] in the case of the continuous space translational symme- for a recent review), the energy of a time-periodic system try being spontaneously broken but not being broken by is not conserved, and should be replaced by the quasi- periodic external fields. energy which has a period of ~2π/T . A resonance hap- It is worth mentioning the difference between our hy- pens between the quasi-energy levels whose difference is potheses and those hiding behind the effective models of an integer times of ~2π/T , which may cause a peak at crystals (such as the Hubbard model30) that were fre- the frequency 1/T in the absorption spectrum of crys- quently used up to now. In these effective models, the tals. Note that 1/T 1018Hz is in the frequency range ∼ kinetic energy is either expressed as p2/2m with p and of X-rays. m denoting the momentum and the mass, respectively, It is worth mentioning that the time-periodic oscilla- or expressed as the hopping energy between neighbor tion of observables has been predicted in the theory of sites on the crystal lattice. In the language of quantum ”time crystals”8. But whether there exists a ”time crys- field theory, the kinetic part of the Hamiltonian looks tal” is still under debate. It was argued that a time- ~2 2 † like ψ(x) ∂x/2m ψ(x) in a field theory or ci ci+1 periodic oscillation of observables cannot happen in an − 12 in a lattice field theory where ψ(x) and ci denote the equilibrium state described by the Gibbs ensemble . field operators of particles
(e.g. electrons) in the crystal. On the other hand, our theory indicates that the La- The discrete space translational symmetry is considered, grangian of a model for crystals should be time-periodic, but the Lorentz symmetry is not under consideration in in which case the idea of describing the equilibrium states these models whose Lagrangians or Hamiltonians always by Gibbs ensemble should be reexamined since it does not change as transformed from one reference frame to the put space and time on an equal footing. other moving at a different velocity. The complete loss Finally, we would like to mention the open problems of Lorentz symmetry can be viewed as a special case of that are expected to be solved in future. These include our theory as the generator of the spacetime is g = 2 and the construction of the discrete Lorentz symmetry in 13

2+1-dimensional and 3+1-dimensional spacetimes, and properties of a group. It is a cyclic group. The set of the quantization of a field theory that has the discrete observers Kj j =0, 1, 2, are all equivalent to Poincar´esymmetry. ± ± ···   each other in describing the physical laws.

Next we prove that L is a subset of , i.e., any element ACKNOWLEDGEMENT Λj(L , 0) of L is also in . Let us denoteV the velocity of v(g) V Kj relative to K0 as vj . By definition, we have v0(g)=0 This work is supported by NSFC under Grant 4 and v (g) = v(g) = 1 , and the Lorentz matrix No. 11304280. 1 − g2 r j relating K0 to Kj is L = L . Similarly, the veloc- vj (g) v1(g) ity of Kj relative to Ki is denoted as vj−i which satisfies Appendix A: The discrete Lorentz group j−i L − = L , and the velocity of Ki relative to K is vj i(g) v1(g) 0 i i vi satisfying L = L . Note that L is the ith We have proved in Sec. III B that the velocity of a vi(g) v1(g) v1(g) Lorentz transformation in the symmetry group must be power of L . We then have Lj = Li Lj−i or v1(g) v1(g) v1(g) v1(g) 4 L = L L . By using the expression of L in v = 1 . In other words, the symmetry group of vj (g) vi(g) vj−i(g) v ± − m2 Eq. (A2), we obtain pure Lorentzr transformations is a subset of vi(g)+ vj−i(g) vj (g)= . (A3) 4 1+ vi(g)vj−i(g) = Λ(Lv, 0) v = 1 , m =2, 3, 4 . V ± − m2 ··· ( r ) Eq. (A3) is the velocity-addition formula which is as same

(A1) as that in special relativity, because both are derived

In this section, we prove that such a group must be a from the Lorentz transformation. It is easy to verify cyclic group generated by some integer g, as defined in that vi is an odd function of i, i.e., v−i = vi, and Sec. IV. Recall that a group is closed with respect to vj vi − multiplication and the Lorentz matrix is expressed as vj±i = ± according to Eq. (A3). 1 vivj ± 1 v 2 Now we define mj (g)= for each velocity − 2 √1 v2 √1 v2 1 vj (g) Lv = . (A2) −  −v 1−  vj (g). One can easily see m = 2 and m (g) = g from − p0 1  √1 v2 √1 v2  4 − − v0 = 0 and v1(g)= 1 , respectively. By definition,   − g2 The multiplication between two Lorentz transformations r mj (g)= m−j (g) is an even function of j. To prove that is Λ(L , 0)Λ(L ′ , 0) = Λ(L L ′ , 0) where L L ′ is the v v v v v v Λj(L , 0) = Λ(L , 0) is an element of , we only product of two matrices. v(g) vj (g) need to prove that m (g) is an integer. ThisV is done by The simplest group that is a subset of is the triv- j finding an iterative formula for m (g). Expressing m ial group containing only the identity transformationV at j j±i by using v and then by v and v , we obtain v = 0 or m = 2. It is a special cyclic group. On the j±i i j other hand, according to our first hypothesis, the sym- mi+j + mj−i = mimj . (A4) metry group should contain at least one Lorentz trans- formation with v = 0. Let us suppose that except for Choosing i = 1, we have the identity element6 the symmetry group contains an 4 mj+1 = gmj mj−1. (A5) element Λ(L , 0) with v(g) = 1 for some in- − v(g) − g2 r Since we already know m0 and m1, Eq. (A5) can be used teger g > 2. Note that supposing v(g) > 0 does not to calculate m iteratively. For example, we find m = lose the generality since Λ(L , 0) and Λ−1(L , 0) = j 2 v(g) v(g) g2 2, m = g3 3g, . The numbers m ,m ,m , Λ(L , 0) must be the elements of a group simulta- 3 0 1 2 −v(g) make− up an infinite− sequence.··· And because m = 2 and··· neously. Once if Λ(L , 0) is an element, according 0 v(g) m = g are both integers, m for arbitrary j in the se- j j 1 j to the property of a group, Λ (Lv(g), 0) = Λ(Lv(g), 0) quence must be an integer according to Eq. (A5). There- j must be an element of the group for arbitrary integer fore, Λ (Lv(g), 0) for arbitrary j and g is an element of j j = 0, 1, 2, . The physical meaning of Λ (Lv(g), 0) , and L is a subset of . ± ± ··· V V is clear. We choose an observer K0, and call who is Up to now, we proved that the cyclic group L gener- moving at the velocity v(g) relative to K0 the observer ated by an integer g is a subset of . Next we prove V K1. Similarly, the observer Kj+1 is moving at the ve- that L is the only possible group that is a subset of locity v(g) relative to Kj . Therefore, the coordinate . We will construct a proof by contradiction. We as- j V transformation from K0 to Kj is Λ (Lv(g), 0). The set sume that there exists a group which is included in G V j but not a cyclic group. By definition, must include L = Λ (Lv(g), 0) j =0, 1, 2, satisfies all the G ± ± ··· at least two elements Λ(Lv(g), 0) and Λ(Lv(g′), 0) where  

14 g,g′ > 2 are both integers and not in the same sequence and ′ mj generated by an integer. Especially, g and g are sgn(j) 2 not in the sequence generated by each other. Without mj /2 − mj 4 ′ L = 2 − . loss of generality, we suppose 2 0, and also v = v(g), oth- First,P we prove that is a group, i.e., is closed under ′ 6 erwise, we have v(g ) = 2v(g)/(1 + v(g)2) and then multiplication, and P has the above-mentionedP proper- ′ ′ P g = m2(g) which contradicts g = mj (g) for arbitrary ties. Second, we prove that a group that has these prop- 6 j. Since Λ(Lv′′ , 0) is in , it must be also in , there- erties must contain by proving that each element in G V after, g′′ = 2/√1 v′′2 is an integer satisfying g′′ = g can be expressed asP a product of the elements in L andP and g′′ 2, since6 g′′ = 2 indicates v′′ = 0 Let us list some important properties of the integer se- ′ ′ and then v(g) = v(g ) or g = g which contradicts our quence mj which will be used in the proof. The iterative assumption. In consequence, we constructed an integer formula (A5) can be reexpressed as g′′ which is different from both g and g′ and is less than the max of them. In the same way, we can use g′′ and g g2 4 ′ mj − − mj the smaller one of g and g (g in this case) to construct +1 − 2 ′′′ ′′ p a new integer g that is different from g or g and less 2 2 (B4) ′′ g + g 4 g g 4 than the max of them. We can do this because g (g ) is = − mj − − mj−1 , ′′ 2 − 2 not in the sequence mj generated by g (g), otherwise, p p ! ′ we can deduce that g is also in the sequence mj gen- erated by g′′ (g) which contradicts the assumption that from which we derive an expression of mj : ′ g and g cannot be in the same sequence. The process j j of constructing new integers can be repeated for infinite g g2 4 g + g2 4 m = + . (B5) number of times. Every time we choose the smallest two j − − − p2 ! p2 ! in the integers that we already obtained to construct a ′′ ′′′ new one. The sequence of integers (g,g ,g , ) that we For convenience of presentation, we define a new sequence obtain are all different to each other and all··· less than g′ and larger than 2. But this is impossible, because there m2 4 ′ j − only exist finite number of integers between 2 and g . zj = sgn(j) . (B6) Our assumption must be false. The only possible groups q g2 4 included in are cyclic groups. This finishes the proof. − V It is straightforward to prove thatp zj can be expressed as

j j 2 2 2 g + g 4 g g 4 Appendix B: The discrete Poincar´egroup g 4zj = − − − . − p2 ! − p2 ! p (B7) Our hypotheses infer that the overall symmetry group The iterative formula of zj is as same as that of mj , being of the spacetime should have next properties: its subgroup for pure Lorentz transformations is L = zj+1 = gzj zj−1. (B8) − Λ Lvj (g), 0 , and its subgroup for pure spatial trans- lations is A = Λ(1,ma¯) with m an integer anda ¯ = The first two elements of zj are z0 = 0 and z1 = 1 which (0, 1) T . In this
section,{ we} prove that the minimum group are both integers, thereafter, zj must be also an integer that has these properties is sequence just like mj ! zj for arbitrary j is an integer, and zj = z−j is an odd function of j. The useful formulas − involving zj and mj are = Λ L , YN N (g) j, N ,N =0, 1, 2, , P vj (g) 1 2 1 2 ± ± ··· g mj   zj+1 = zj +
(B1) 2 2 where . (B9) g mj  zj− = zj 1 2 − 2 1 2 0 g 4 And a generalized iterative formula for zj is Y = N + N 2 − , (B2) N1N2 1 1 2  1  p g   zi j = zi zj zizj, (B10)  2  + +1 +1 +1 −   15

N1 N2 which can also be expressed in a matrix form as Therefore, Λ(1, YN1,N2 ) = Λ(1, Y1,0) Λ(1, Y0,1) can be expressed as a product of the elements in L and A zi+1 zi zj+1 zj zi+j+1 zi+j = . for arbitrary N1 and N2. zi zi−1 zj zj−1 zi+j zi+j−1 For the element Λ(Lv , Y ′ ′ ) in , we can factorize  − −   − −   − −  j N1N2 (B11) it into P

Especially, by taking i = j we obtain Λ(Lv , YN ′ N ′ )=Λ(Lv , 0)Λ(1, YN N ), (B16) − j 1 2 j 1 2 z z z z j−1 j j+1 j ′ ′ ′ ′ − − =1. (B12) where YN N = Lvj YN1N2 or YN1N2 = Lv−j YN N . zj zj+1 zj zj−1 ′ ′ 1 2 1 2    − −  (N1,N2) and (N1,N2) satisfy the relation (B14). Ac- cording to Eq. (B12), this relation is invertible and its inverse is 1. is a group P ′ ′ N1 = zj−1N1 zj N2 − ′ − ′ . (B17) According to Eq. (3), the product of arbitrary two el- N2 = zjN + zj+1N  1 2 ements in is ′ ′ P For arbitrary (N1,N2), we can find integers N1 and N2

Λ Lvj , YP1P2 Λ (Lvi , YN1N2 ) that satisfy Eq. (B16). This means that each element in can be expressed as the product of an element in =Λ Lvj Lvi ,
YP1P2 + Lvj YN1N2 (B13) P L and Λ(1, YN1N2 ). But the latter has been proved to =Λ Lvi+j , YP1P2 + Lvj YN1N2 ,
be a product of the elements in L and A. Therefore,
each element in can be expressed as a product of the where i,j,N1,N2, P1 and P2 are all integers. To prove elements in L andP A. that is closed with respect to multiplication, we need P to prove that Λ Lv , YP P + Lv YN N is in . This i+j 1 2 j 1 2 P is equivalent to prove that Lv YN N = YN ′ N ′ is a vector j 1 2
1 2 Appendix C: Theories that has the discrete Poincar´e in the characteristic lattice for arbitrary Lvj and YN1N2 in symmetry P the lattice. By using the expression of Lvj (see Eq. (B3)) and Eq. (B9), we obtain 1. The field theory ′ N1 = zj+1N1 + zjN2 ′ . (B14) N2 = zjN1 zj−1N2 In this subsection, we explain how to construct the  − − function M(y) which satisfies ′ ′ Since zj for arbitrary j is an integer, N1 and N2 must be integers. Therefore, Y ′ ′ is a vector in the char- N1N2 M(y)= M(Λy) (C1) acteristic lattice, and then is closed with respect to P multiplication. for arbitrary Λ . In other words, M(y) is invari- In Sec. V, we already showed that L and A are the sub- ant under .∈ Each P element of can be factorized groups of for pure Lorentz transformations and pure into Λ L P , Y = Λ(1, Y P)Λ L , 0 where P vj (g) N1N2 N1N2 vj (g) spatial translations, respectively. We conclude that is Λ (1, Y ) and Λ L , 0 are also the elements of . P N1N2
vj (g)
a group that satisfies the conditions of the overall sym- Therefore, M(y) is invariant under if and only if M(Py) metry group.
P is invariant under the transformations Λ Lvj (g), 0 and

Λ (1, YN1N2 ), i.e., M(y) is invariant under the discrete Lorentz group L and the discrete translational group
Y. 2. The symmetry group cannot be smaller than P We notice that Λ (1, YN1N2 ) y = y+YN1N2 . Because M

is invariant under Y, we obtain M(y) = M(y + YN1N2 ) In this subsection, we prove that each element in for arbitrary N and N . This means that M is a peri- P 1 2 can be expressed as a product of the elements in L and odic function in the 1+1-dimensional spacetime, and has A. the same periodicity as the characteristic lattice YN N . { 1 2 } Recall that the characteristic lattice YN1N2 has Such a periodic function can be expressed as a Fourier T { } two primitive vectors: Y1,0 = (0, 1) and Y0,1 = transformation. The characteristic lattice has two prim- T 1 1 itive vectors: Y1,0 and Y0,1. For convenience of presenta- 2 (1) T g 4, g . By using the expression of Lv−1 in tion, in this subsection we rename them as Y = (0, 1) 2 − 2 T terms p of g (see Eq. (B3)), we express the second primitive 1 1 and Y (2) = g2 4, g . Each vector in the char- vector as Y0,1 = Lv−1 Y1,0. We then obtain 2 − 2  p  (1) acteristic lattice can be expressed as YN1N2 = N1Y + Λ(1, Y0,1)=Λ(Lv− , 0)Λ(1, Y1,0)Λ(Lv , 0). (B15) (2) 1 1 N2Y . The reciprocal lattice has also two primitive 2πg Λ(1, Y1,0) denotes the minimum spatial translation which vectors which are found to be k(1) = − , 2π 2 is an element of A, and Λ(Lv±1 , 0) are the elements of L. g 4 ! − p 16

(2) 4π Eq. (B11) and (B12). The integer pairs (n1,n2) that are and k = , 0 . The inner product between related to each other by Eq. (C5) form an equivalence g2 4 ! − class. All the coefficients M with (n ,n ) being in the primitive vectors of the characteristic lattice and n1n2 1 2 p (a) (b) the same class must be the same. the reciprocal lattice satisfies k Y = 2πδa,b where · a,b = 1, 2 and δa,b is the Kronecker delta function. For (1) (2) the momentum vector k = n1k + n2k with n1 and 2. The lattice field theory ik· y+Y n being integers, we have eik·y = e ( N1N2 ). eik·y 2 { } at different (n1,n2) form a basis of the periodic func- In this subsection, we explain how to construct the cou- tions on the characteristic lattice. Therefore, M must be pling function h in a lattice field theory that is invariant expressed as under . h must satisfy P i n k(1) +n k(2) ·y M(y)= M e ( 1 2 ) , (C2) h = h ′ ′ ′ ′ (C7) n1n2 P1P2,Q1Q2 P1P2,Q1Q2 n1,n2 with YP ′ P ′ =ΛYP P and YQ′ Q′ =ΛYQ Q for arbitrary X 1 2 1 2 1 2 1 2 Λ . ∈P where Mn n is the coefficient of the Fourier transforma- Again, each element of can be factorized into 1 2 P tion. Λ Lvj (g), YN1N2 = Λ(1, YN1N2 )Λ Lvj (g), 0 . The cou- M(y) should also be invariant under L, which imposes pling function h is invariant under if and only if it
P
a constraint on the coefficients Mn1,n2 . Substituting is invariant under the transformations Λ (1, YN1N2 ) and

Eq. (C2) into the condition M(y) = M Λ Lvj , 0 y , Λ Lvj (g), 0 . Since h is invariant under Λ (1, YN1N2 ), we we obtain have hP ,P ,Q ,Q = hP +N ,P +N ,Q +N ,Q +N for arbi-

1 2
1 2 1 1 2 2 1 1 2 2 trary integers N1 and N2. This means that hP1P2,Q1Q2 (1) (2) i(n1k +n2k )·y Mn1n2 e depends only upon the difference between (P1, P2) and n1,n2 (Q1,Q2). We can then reexpress the coupling function X (C3) (1) (2) as i(n1k Lvj +n2k Lvj )·y = Mn1n2 e , n ,n 1 2 hP P ,Q Q = h(P Q , P Q ). (C8) X 1 2 1 2 1 − 1 2 − 2 where we used the properties of the inner product and Let us use the notation YN N = YP P YQ Q , or T 1 2 1 2 − 1 2 Lv = Lvj . We notice that equivalently, N = P Q and N = P Q . The cou- j 1 1 − 1 2 2 − 2 pling function hP1P2,Q1Q2 should be invariant under the (1) (2) ′ (1) ′ (2) Lorentz transformation Λ Lv , 0 , under which we have n1k Lvj + n2k Lvj = n1k + n2k , (C4) j YP ′ P ′ = Λ Lv , 0 YP P and YQ′ Q′ = Λ Lv , 0 YQ Q . 1 2 j 1 2
1 2 j 1 2 ′ ′ ′ ′ ′ ′ where We then find YN N = YP P YQ Q = Lvj YN1N2 . The 1
2 1 2 −′ 1′ 2
integer pairs (N1,N2) and (N1,N2) have the next rela- ′ n1 = zj+1n1 zj n2 tion: ′ − . (C5) n2 = zjn1 zj−1n2 ′  − N1 = zj+1N1 + zjN2 ′ . (C9) N = zjN1 zj−1N2 Therefore, Eq. (C3) stands if and only if the coefficients  2 − − M satisfy n1n2 Substituting Eq. (C8) into Eq. (C7), we obtain

′ ′ Mn n = Mn′ n′ (C6) 1 2 1 2 h(N1,N2)= h(N1,N2). (C10)

′ ′ for the integer pairs (n1,n2) and (n1,n2) that are related The coupling function must satisfy Eq. (C10) for being to each other by Eq. (C5). The relation (C5) is in fact invariant under . Again, the integer pairs that are re- an equivalence relation which is reflexive, symmetric and lated to each otherP by Eq. (C9) form an equivalence class. transitive. The reflexivity, symmetry and transitivity can Eq. (C10) says that h(N1,N2) with (N1,N2) being in the be easily proved by using the properties of zj given in same class must be the same.

∗ [email protected] 4 C. S. Wu, E. Ambler, R. W. Hayward, D. D. Hoppes, and 1 M. P. Marder, Condensed Matter Physics (John Wiley & R. P. Hudson, Physical Review 105, 1413 (1957). Sons, Inc., Hoboken, New Jersey, 2010). 5 M. Z. Hasan and C. L. Kane, Reviews of Modern Physics 2 W. Greiner and J. Reinhardt, Field Quantization 82, 3045 (2010). (Springer-Verlag, Berlin, Heidelberg, 1996). 6 A. Altland and B. D. Simons, Condensed Matter Field The- 3 T. D. Lee and C. N. Yang, Physical Review 104, 254 ory (Cambridge University Press, New York, 2010). (1956). 7 W.-K. Tung, Group Theory in Physics (World Scientific 17

Publishing Co Pte Ltd., Philadelphia, 1985). Review Letters 59, 521 (1987). 8 F. Wilczek, Physical Review Letters 109, 160401 (2012). 19 R. D. Sorkin, arXiv:0309009 (2003). 9 A. Shapere and F. Wilczek, Physical Review Letters 109, 20 H. Yamamoto, Nuclear Physics B (Proc. Suppl.) 6, 154 160402 (2012). (1989). 10 T. Li, Z.-X. Gong, Z.-Q. Yin, H. T. Quan, X. Yin, 21 E. R. Livine and D. Oriti, Journal of High Energy Physics P. Zhang, L. M. Duan, and X. Zhang, Physical Review 06, 050 (2004). Letters 109, 163001 (2012). 22 M. A. Rieffel, Mem. Amer. Math. Soc. 168, 1 (2004). 11 P. Bruno, Physical Review Letters 111, 070402 (2013). 23 M. A. Rieffel, Mem. Amer. Math. Soc. 168, 67 (2004). 12 H. Watanabe and M. Oshikawa, Physical Review Letters 24 R. Resnick, Introduction to Special Relativity (John Wiley 114, 251603 (2015). & Sons, Inc., New York, 1968). 13 K. Sacha, Physical Review A 91, 033617 (2015). 25 E. Wigner, Annals of Mathematics 40, 149 (1939). 14 D. V. Else, B. Bauer, and C. Nayak, Physical Review Let- 26 N. W. Ashcroft and N. D. Mermin, Solid State Physics ters 117, 090402 (2016). (Harcourt, Inc., Philadelphia, 1976). 15 V. Khemani, A. Lazarides, R. Moessner, and S. L. Sondhi, 27 A. Eckardt and E. Anisimovas, New Journal of Physics 17, Physical Review Letters 116, 250401 (2016). 093039 (2015). 16 J. Zhang, P. W. Hess, A. Kyprianidis, P. Becker, A. Lee, 28 J. Smit, Introduction to Quantum Fields on a Lattice J. Smith, G. Pagano, I.-D. Potirniche, A. C. Potter, (Cambridge University Press, Cambridge, 2002). A. Vishwanath, et al., Nature 543, 217 (2017). 29 A. Lazarides, A. Das, and R. Moessner, Physical Review 17 S. Choi, J. Choi, R. Landig, G. Kucsko, H. Zhou, J. Isoya, E 90, 012110 (2014). F. Jelezko, S. Onoda, H. Sumiya, V. Khemani, et al., Na- 30 J. Hubbard, Proceedings of the Royal Society of London ture 543, 221 (2017). 276, 238 (1963). 18 L. Bombelli, J. Lee, D. Meyer, and D. S. Rafael, Physical