The low temperature thermodynamic properties of simple metals can be

qualitatively understood in terms of the simple Sommerfeld model, which

treats the conduction electrons as noninteracting fermions in a box. Typical
results are a specic heat linear in temperature and a constant spin
susceptibility in qualitative agreement with experiments. Landaus Fermi
liquid theory rests on the assumption of quasi-particles which are in a one-to-
one correspondence to noninteracting fermions. This leads to a linear specic
heat and a constant spin susceptibility but involves renormalized quantities
like the eective mass and the quasi-particle interaction parameters, which
are dicult to calculate microscopically. The consistency of the approach was
shown using perturbation theory to innite order and more recently by
renormalization group techniques.

The problem of interacting fermions simplies in one dimension. In a

pioneering paper Tomonaga treated the case of a two-body interaction which
is long ranged in real space. He showed that the low energy excitations of the
noninteracting as well as the interacting system can be described in terms of
noninteracting bosons. The important idea to solve the case of interacting
fermions was the observation that a long range interaction in real space is
short ranged in momentum space and therefore only particles and holes in
the vicinity of the Fermi points are involved in the interacting ground state
and states with a low excitation energy.

3-Dimensional Fermi liquid theory is mostly related to a picture of quasi-

particles when we adiabatically switch on interactions, to obtain particle-hole
excitations. These quasi-particles are directly related to the original fermions.
Of course they also obey the Fermi-Dirac Statistics. Based on the free Fermi
gas picture, the interaction term: (i) it renormalizes the free Hamiltonians of
the quasi-particles such as the eective mass, and the thermodynamic
properties; (ii) it introduces new collective modes. The existence of quasi-
particles results in a nite jump of the momentum distribution function n(k) at
the Fermi surface, corresponding to a nite residue of the quasi-particle pole
in the electrons Green function. 1-Dimensional Fermi liquids are very special
because they keep a Fermi surface (by denition of the points where the
momentum distribution or its derivatives have singularities) enclosing the
same k-space volume as that of free fermions, in agreement with Luttingers
theorem.

1-Dimensional electrons spontaneously open a gap at the Fermi surface when

they are coupled adiabatically to phonons with wave vector 2kF. The mean-
eld theory tells us that there is a charge or spin density wave instability at
some nite temperature for repulsive interactions implying that there can be
no Fermi liquid in 1-Dimension. There are no fermionic quasi-particles, and
their elementary excitations are rather bosonic collective charge and spin
uctuations dispersing with dierent velocities. An incoming electron decays
into such charge and spin excitations which then spatially separate with time
(charge-spin separation). The correlations between these excitations are
anomalous and show up as interaction-dependent nonuniversal power-laws in
many physical quantities where those of ordinary metals are characterized by
universal (interaction independent) powers. In this paper I had a brief review
on how the 1-Dimensional Fermi system comes out, and how spin-charge
separation emerges in 1-Dimensional Fermi system. By using the particle-
hole excitation operators, which is the density operators for Fermions, (the
so-called bosons), we can diagonalize this Hamiltonian. Therefore we can
obtain the grounp velocity for both system of CDW and SDW, and nally nd
their dierence on velocities, and the two waves seperate.

Another interesting property of a Fermi liquid in two and three dimensions is

that the one-particle momentum distribution function n(~ k), obtained by
Fourier transforming the one-particle equal-time correlation function, has a
nite discontinuity at the Fermi surface as shown in Fig. 1 (a). This
discontinuityis called the quasiparticle renormalization factor z~ k; it is also
equal to the residue of the pole in the one-particle propagator. For non-
interacting fermions, z~ k = 1; but for interacting fermions, 0 < z~ k < 1
because a quasiparticle is a superposition of many states, only some of which
are one-particle excitations. To compute z~ k, we consider the one-particle
Greens function G(~ x,t) dened as the expectation value of the time-
ordered product of the fermion operator (~ x,t) in the ground state |0i,
namely, G(~ x,t) = h0| T(~ x,t)(~ 0,0) |0i

In contrast to a Fermi liquid, interacting fermion systems in one dimension

behave quite dierently [2, 1, 3]; we will assume again that the ground state
breaks no symmetry. Such systems are called Luttinger liquids and they have
the following general properties. First of all, there are no single particle or
quasiparticle excitations. Thus all the low-energy excitations can be thought
of as particle-hole excitations; further, all of these take the form of sound
waves with a linear dispersion relation. (As we will see below, there are also
excitations of another kind possible which correspond to adding a small
number of particles NR and NL to the right and left Fermi points. However,
these correspond to only two oscillator degrees of freedom, and therefore do
not contribute to thermodynamic properties like the specic heat). Secondly,
there is no discontinuity in the momentum distribution function at the Fermi
momentum, as indicated in Fig. 1 (b). Rather, there is a cusp there whose
form is determined by a certain exponent. Finally, this exponent depends on
the strength of the interactions in a non-universal manner, and it also
governs the power-law fall-os of the correlation functions at large space-
time distances [4].

So what results has bosonization given us here? We started with electrons

with spin and charge moving together via a hopping term, but with a strong
on-site Coulomb repulsion. This term could not be treated perturbatively.
However, when we rewrote the theory in terms of bosons, with one boson for
the spin and one for the spin, we found that the theory decoupled in
terms of new spin and charge bosons. For a generic lling, the charge boson
was just a massless free boson excitation, whereas the spin boson
Hamiltonian had a cosine term, which was relevant when U was negative, but
irrelevant for positive U. But at