Jordan-Wigner and Bosonization Jordan and Wigner's 1927 paper introcuced the canonical articommctatin relations fr fermions. ‘Seeking an explicit represertation ofthe operators, they estaslched a mepaing between ferrin ane ssin-1/2 operators, Much ates it was realized that thisraoping tums certain spin chain amiltonians into systers of free ferrion, cxovicing 3 rather elerertary exact solution n this lecture, well exalore some of things we can ce with this transformation, as well as an alternative way cf looking at 10 systems calles bosonization, From Fermions to Spins (and Back) Its not too hard to fac ¢ matrix representation ofthe Losoric creation anc anniilation operators satistying Tre ction of al on the occupation number eigenstate |n) is cf ccurse a'\n) = ViFTn +1) @) Inti esis we witea general stateas |) = 5% Ynin) Then we hove avn lin +1). @ all) = Sovak This tells us thatthe creation operator can be represented bya matrix Ak such that (Ab naw = Viet @ « 00 0 os 0 10 0 0 0 Ah=|0 v2 0 0 - > 6) 0 0 VB Ove (nr With Ar being given by the hermit con.uoate, heck tat (1) sted by th ereesatn ad that Al Ay — the number cert fwe have twobosors,a,a!, ard, bl, they ect inceendently on their cecucation numer states, forexample alia) & |ran} = (a rin)) & [rs) = y/7ag + Tlrta + 1) & |r} ©) ‘Atwo boson state is written |t!}2 = ransom Ma} & [Me)n using comecnents none labelled by a pai of intecer indices. The reoresertation of at (bi) acts erly onthe frst (seccnc) index. We have ('Y)nans = Van cr more abstractly CO V)nom = VRB nannt ( Aen bl=1@al. ®) 1\What about fermions? Here the occupancy cf @ state can only be zere or cre, andi nat hardto ‘show thatthe furdamertal anticemmutater faa!) = @ ras he cess reresertaton 0 0) ( a: (10) infect, because ofthe symmetry between the emotyendfiled state, we cul ust 2 wl te pod al= 6 : ay \Wevilin fact stick ith te later choice, For wo fermions, we have orcler. I we ty alZales, pZaeal, aa) thenal and b! commute rather then anticomrute. At this point let's switch to Pauli matrix rctation AL = 0° Ap = 07. Shoe diffrent Paull maces anicorute, we also Fave {o%,0°} (13) ‘Thus one (somewhat asynmetrical) solution to cur orotlen is to take al=otat, b=0°@o! (a4) Ittuns cut that his rick generalizes, Forte fermions fr example at (15) u (18) ¢ a) ard soon. For NY fermicns we can waite jtpines ojtine oe. = aa=F a BE Rot VIZ Be (1s) ‘Narrally we cinguish the cifferert spin operators ty an index rather than the place they occur in fhe tenscr procuct, anc write (3 a ( “) of (a9) For applications, we are mostly interested in the inverse rarsferration, exeressing spins in terms of ferrrions. Nctice that os sirply related to the nurber operator 95 = 2aja,-1=2n,;- (20) Alternatively -exp(im)}, en) inwhicn case : a} =(-0e=p (inn) a 2) Trat (—1))? atthe front isa bit annoying, If we just define it away all the reculted algebraic felations ae stil reserved, leaving a} =e (im) al. 3) anSolving the XY Model ‘The XY rredel is a particular case ofthe anisotroic spin chain (see Problem Set 7 in which ony the :Tanc y comporents ofthe spins are coupes (Fence the rame) H=-JY sishs + sists] (24) aS LS Ion porot 5) > y [ojop toyofu): (25) In Lecture 4 we saw that the XY ter canbe interpreted as cescribng the heping of ragnons. oF sin ips, Now apoying the mapping (23), we see that it can ecually be interpreted in terms of ferrrion hopping (26) ‘The last equality follows fem the fact tat, in order to create @ particle at site jthe site rust frst be erly, sc thatthe exponential is just 1. In this way we see that the XY Harrilcnian Cescrtes free ferrions (27) Note that itis essentthat we aren one dimension Te ordan-wigner racing coud of course be made in ry ciersion butt epencs onthe saticulr choice of te labeling ofthe soins. nwo cirersiog fr exaroa ts ret ossicle to choose tat labeling so that ell hoopin terms 0 for nearestneighbours ae expressed sirplyintemsof the ferrion. Rate thre woul be rare tal excorertial ecto eft cet Diagenalizing a finite chain descrited by (27) is of course straightforward if we choose pericdic veaneayycorciiens fr hea ay However hysial eres reecemae the spr, we shoullpse prec today cevciiona cn these certs ieee Haw ces ths aft the feign ste? Weave aha, = SNgkos “ONT 3) where N, = 52), The secon ecuaity follows fom the observation that after moving aprtce tothesite N, thee are certainly N,— I particles cn the ther sites. ‘Thus the fermicrs bey (entF)periocic bcurdary cencitions when Nj is occ (ever), AS usual, we can selve the Hamiltonian with the aid ef the plane wave states ea) (30) bere + integer Na odd. fe {f X (integer + $) Na even. 1) In either case, the Harriltonian (27) can be written Joos (kt) (32) H=)ieinim Since Sijy = 2NV, — N te filings ceterrines by te total ragnetization, In particular Sy correscords to hal flirg. \Welve actally met this before albeit in a different cuse, Recall that in Lecture 1 we saw thatthe impenetrable Bose gas in one dimension can be understooc in ters of rrinterecing fermions, Spir-1/2 cegrees of freedom are usta latice version of hardcore ferions, anc the Lercan-Wigrer a7string seves totum an antisyretricferrinic wavefunction into a bosonic one, To ferrally recover the 3ose Gas without lattice we would have to consider the limit of very lw filing Spin Correlations Solving forthe eigenstates ofa mocel is one thing, but as we've seen before its cuite another to be able to compute correlation functions, Sut let's start with something simple, and conser the correlation furction cf the 2corrponents ef the spins (O\(2nj ~ Y(2ne = 1))0) (33) = 4(O}nre|0} — 4¢0|nj 10) +1. (34) CU-8 (Olo50%0) Evidently this is elated to the density correlations in the Feri gas ts Fourier transform, definec by 1 CU-)= Ee ce (35) s 4 =H MARNua) axe (36) where Ne = (Oin{rc|0) isthe cccusaticn nurrter ofthe croune state, We fine Bh iq) < 2h ce= i, © 37) ‘ (B [a > 2ke. od where the Fermi morrentum ke = S, What about the transverse sin conelations en ® (lafael0y? (38) Inthe hardcore boson picture the transverse correlations ae ust the one body density matrix, sc the Fourier transform would tellus atout the occupation rurers ofthe since particle lane wave states inthe mary body ground state. \Wirting C*~(j — K) in terms of the fermions oives 4 (lalexp (imi mayo) 7 >k cHG-B-y ot ( 7 ie . (39) 1a} exp (in) mu) 9,0) i) wes (+ PY ait Ea + (u+ goo) pal Hy a (47) 8 comoarision with the usual oscilstor Haviltonian we can reac off the Bogoliuboy spectrum of exctations pe 2p? oe E(E vm) wor, (8) where = 52 js the spees ef sound Wie are interested inthe ong wavelength mit where Oy ~ cp). The avrcurts to ignoring the (quantum pressureterr orGunna peste mp foe (8:6? , “Cw 9 2m ‘8mpy 3? (49) + Hn © faal aio? + 20?! & [| K0.9? + lO) here p = Gxo/n. Tis cusdatc Hamiltonians krown a the TomanagaeLutinger cc snc K= a (50) 's the Luttinger parameter. ‘As usual we caninieuce the oscillator variables ola) VEE e-* (ay +e) 1) 42) Yet (ae he). co With the choice B — 52 E+ map ©) the cuacratic Haritenin becomes H,=S_ Ohhh, 63) F Boson density matrix Nowlet's tur back to cur riginal preter the computation ofthe transverse soin correlation function (39), or ecuivalently the boson cersity ratix fr herccore sors. The si aperexirationis to take 9 \/n(x) ~ Vie, (54) ‘Thejustificatio fr this is that i than the contribution from ths part. 2 correlation function the corections will decay more quickly Evaluating the density matrix is row a strightforwarc calculation Old" x} (@')10) ~ po(Ole"47e%") (0) (65) ~ poexp (30 ta) ~ 2/0) (58) cos hla’ soem (-pe fat 2K) oy ~ exp (zr le— #\) (58) How do we krow the value ofthe Lutinger parameter K7In general, we cor! However, thanks 10 orcan=Wigney, we know that there is a really a free fermion system urdereat al this, which allows us to deduce since pp = kr/m that K = 1. This clves us finally aA E-2 Ole" a v(2")/0) ~ (9) ‘tough the bosorization rretkod is cnt abe to obtain results about the asymptoticbeavicur of corelation furctions thas @ generality far teycn the Jordanwwigrer azproach. Fr example wen Je # Din the Heisenberg cel, the Jcrdan=Wigrer trarsforrration maps the spin chain te @ ortem cf interecting ferrions, Nevertheless, the Torronage~Luttinger Hariltenian is stllthe correct long distance descrition with the esyotetic behaviour cf all coreation functions depencing only on K. prevous essen and Coretcen Last updeted on Nov 16,2021 Pubes wth Manchemy~ te fe gee soua neste lena ergo ea uw