Sie sind auf Seite 1von 2

Problem Set 5

Phys 7240
Due: Nov 13

Tricritical Ising Model

The tricritical Ising model is defined by its partition function

X K σj σk −∆ σj2 +h σj
Z= e hjki j j . (0.1)
σi =−1,0,1

hjki refers to the nearest neighbors and σ takes values ±1 and 0.

The tricritical Ising model, in 2 dimensions or above, has a line of first order phase
transitions at T < Ttc as a function of ∆ (at h = 0), and second order phase transitions at
T > Ttc , again as ∆ is varied. Exactly at T = Ttc , at ∆ = ∆c and h = 0, it has a tricritical
point. For more details, see J. Cardy’s book, section 4.1.

1 Tricritical Ising model in one dimension

Compute the partition function above in 1D by transfer matrix technique, at h = 0.
Introduce k = e−K , δ = e−∆ , and show that at in the limit of zero temperature the free
energy is not differentiable at k/δ = 1 (corresponding to a first-order phase transition).
Check that at nonzero temperature the discontinuity disappears, confirming that there are
no phase transitions at finite temperature in one dimension. The discontinuity corresponds
to the transition between spin up or spin down state and spin zero state.

2 Mean Field Theory

John Cardy, “Scaling and Renormalization in Statistical Physics”, problem 2.5.

Again, take h = 0. Do a substitution σi = M + (σi − M ) in the σj σk term and derive

mean field theory for the tricritical Ising model, by neglecting terms (σj − M )(σk − M )
(however, do not neglect (σj − M )2 ). Show that the mean field free energy has the form
F h i
= KdM 2 − ln 1 + 2 cosh (2KdM ) e−∆ , (2.1)
where d is the dimensionality of space and N is the number of sites in the lattice.
Expand F in Taylor series in M , as in
≈ tM 2 + uM 4 + u6 M 6 (2.2)
Show that in the vicinity of the point where both u and t change sign from positive to
negative, u6 is positive. If u > 0, then the model has second order phase transition when
t changes sign. If, however, u < 0, then the model has first order phase transitions as t
goes through some critical value tc (u). At u = 0, the model has a tricritical point.
By minimizing F , calculate how M depends on t (at u = 0) and on u (at t = 0) in the
vicinity of a tricritical point.

3 Renormalization Group analysis

John Cardy, “Scaling and Renormalization in Statistical Physics”, problem 5.1.

Consider the partition function Z = DM e−F , with F given by


2−d d 2 −d d 2 −d d 4 −d
F = Ka d r (∇M ) + ta d r M + ua d r M + u6 a dd r M 6 . (3.1)

A tricritical point is where t = 0 and u = 0. Find the upper critical dimension dc for
the tricritical Ising model (hint: examine the RG dimension of u6 ). Calculate the critical
exponents below three dimensions in  = dc − d expansion. Don’t forget that there are
two thermal scaling variables at the tricritical point (t and u) with two separate scaling