
5
coupling functions as follows
∂zwψ(x, z) = βwψ[ϕ(x, z), wψ(x, z), vψ(x, z), λχ(x, z)],(12)
∂zvψ(x, z) = βvψ[ϕ(x, z), wψ(x, z), vψ(x, z), λχ(x, z)],(13)
∂zλχ(x, z) = βλχ[ϕ(x, z), wψ(x, z), vψ(x, z), λχ(x, z)].(14)
Physical meaning of these first-order differential equations is clear. The real question is how to find these RG
β−functions in a nonperturbative way, to be clarified below.
ϕ(x, z) is an energy-dependent chiral symmetry breaking order-parameter field, whose dynamics is determined by
minimization of the bulk effective action in the large−Nlimit, given by
Sbulk =NZzf
0
dz ZdDxnπϕ(x, z)∂zϕ(x, z)−βϕ[ϕ(x, z), wψ(x, z), vψ(x, z), λχ(x, z)]−1
2λχ(x, z)π2
ϕ(x, z)
+Veff [ϕ(x, z), wψ(x, z), vψ(x, z), λχ(x, z)]o.(15)
πϕ(x, z) is the canonical conjugate field to ϕ(x, z). In this respect this effective bulk action is written in the Hamiltonian
formulation. The essential ingredient is an effective potential Veff [ϕ(x, z), wψ(x, z), vψ(x, z), λχ(x, z)], generated by
the RG transformation for Dirac fermions and given by
Veff [ϕ(x, z), wψ(x, z), vψ(x, z), λχ(x, z)]
=−1
Nln ZΛ(z)
Dψσ(x) exp h−ZdDxn¯
ψσ(x)wψ(x, z)γτ∂τ−ivψ(x, z)γi∂iψσ(x)−iλχ(x, z)ϕ(x, z)¯
ψσ(x)ψσ(x)oi.
(16)
Here, RΛ(z)Dψσ(x) means that the fermion path integral is performed at a given energy scale Λ(z), which will be
clarified in the next section. Accordingly, the RG β−function of the chiral condensate is given by the functional
derivative of this effective potential with respect to the order-parameter field [1] as follows
βϕ[ϕ(x, z), wψ(x, z), vψ(x, z), λχ(x, z)] = −δϕVef f [ϕ(x, z), wψ(x, z), vψ(x, z), λχ(x, z)] = i
NDλχ(x, z)¯
ψσ(x)ψσ(x)E.
(17)
Here, h...iis an ensemble average with respect to the effective action functional
SΛ(z)=ZdDxn¯
ψσ(x)wψ(x, z)γτ∂τ−ivψ(x, z)γi∂iψσ(x)−iλχ(x, z)ϕ(x, z)¯
ψσ(x)ψσ(x)o.(18)
Three other RG β−functions are given by functional derivatives of this effective potential with respect to the
corresponding coupling function in the following way
βwψ[ϕ(x, z), wψ(x, z), vψ(x, z), λχ(x, z)] = −δwψVef f [ϕ(x, z), wψ(x, z), vψ(x, z), λχ(x, z)] = −1
ND¯
ψσ(x)γτ∂τψσ(x)E,
(19)
βvψ[ϕ(x, z), wψ(x, z), vψ(x, z), λχ(x, z)] = −δvψVef f [ϕ(x, z), wψ(x, z), vψ(x, z), λχ(x, z)] = i
ND¯
ψσ(x)γx∂xψσ(x)E,
(20)
βλχ[ϕ(x, z), wψ(x, z), vψ(x, z), λχ(x, z)] = −δλχVef f [ϕ(x, z), wψ(x, z), vψ(x, z), λχ(x, z)] = i
NDϕ(x, z)¯
ψσ(x)ψσ(x)E.
(21)
This procedure is completely consistent with what we learnt in the RG transformation [1].
It is straightforward to find the Hamiltonian equation of motion in the large−Nlimit, given by
πϕ(x, z) = λχ(x, z)∂zϕ(x, z)−βϕ[ϕ(x, z), wψ(x, z), vψ(x, z), λχ(x, z)](22)
and
∂zπϕ(x, z) = −βϕ[ϕ(x, z), wψ(x, z), vψ(x, z), λχ(x, z)] −πϕ(x, z)δϕβϕ[ϕ(x, z), wψ(x, z), vψ(x, z), λχ(x, z)],(23)