# Assumptions of Landau theory

1. Aug 20, 2012

### matematikuvol

Assume that we can expand the Helmholtz potential about $$T=T_c$$, $$M=0$$ in a standard Taylor series form of functions of the variables,
$$A(T,M)=\sum^{\infty}_{j=0}L_j(T)M^j=L_0(T)+L_2(T)M^2+L_4(T)M^4+...$$

Why $$A(T,M)$$ must be even function of $$M$$?

Coefficients can be expanded about $$T=T_c$$

$$L_j(T)=\sum^{\infty}_{k=0}l_{jk}(T-T_c)^k=l_{j0}+l_{j1}(T-T_c)+...$$

How I could no that coefficients are analytic functions od $$T$$.

2. Aug 20, 2012

### tejas777

Consider a system of spins, where M is the coarse-grained magnetization of a certain “block.” In that block, if I flip every spin then there is no change in the energy in the system. Flipping of all the spins would result in M → -M. That’s why A(T,M) is even in M. I didn't quite understand your second question about analyticity

Last edited: Aug 20, 2012
3. Aug 20, 2012

### Mute

In the absence of an external magnetic field, you can change the sign of every spin and the total energy stays the same (as you can see by looking at the Ising model Hamiltonian), as tejas777 said. Flipping the sign of every spin flips the sign of the magnetization, so if the energy doesn't change, then changing the sign of M cannot matter, so the free energy must only depend on even powers of M.

As far as Landau theory is concerned, t's an assumption, the same with how you assume that the free energy is analytic in M and can be expanded in integer powers of M (otherwise terms like |M| could be present, which are not even powers but still obey the symmetry that the free energy doesn't change if you take M -> -M).

You'll find out eventually that it's actually a bad assumption.

4. Aug 22, 2012

### matematikuvol

I think that this is very hard problem, and not precise theory.

Helmholtz potential is convex function of magnetisation.

$$A(T,M)=\sum^{\infty}_{j=0}L_j(T)M^j$$

that must put some demands on coefficients in series, and if I say

$$L_j(T)=l_{j0}+l_{j1}(T-T_c)+...$$

I have a problem with different behavior below $$T_c$$ and for $$T>T_c$$.

So is there coefficients $$l_j$$ positive or negative?

5. Aug 22, 2012

### Blackforest

The Mac Laurin Taylor development that you have writen is the one depending on the M2 variable. This is the reason why you get only even coefficients. The M variable of the Landau theory is more often written as $\Psi$ and represent the bosonic field amplitude. The square of it, in extenso $\Psi$2, is the bosonic density.