sanctus Posted September 6, 2006 Report Posted September 6, 2006 This is a quite complicated Ising model under compression with a force 2[math]\lambda[/math], 2 dimensional, with horizontal, vertical and diagonal ising interaction. After long calculations I found that this is the density of th free energy of the system. [math]f(T,\lambda)\equiv\frac{F}{N}=\frac{-1}{\beta}lim_{N\to \infty}\ln Z_N = 2\lambda a_0 +\frac{1}{2\beta}\ln\left(\frac{\pi^2 m}{\varphi_2 \beta^2}\right)+2\varphi_0-\frac{\lambda^2}{\varphi_2}-\frac{j_1^2+j_3^2}{2\varphi_2}-\frac{1}{\beta}\ln \eta_1[/math] Where:[math]\eta_{1}={e^{\beta\,\left({\it \rho_3}-{\it \rho_4} \right)}}\cosh \left(2\,\beta\,\left( {\it \rho_1}-{\it \rho_2}\right)\right)+{e^{\beta\,\left( {\it \rho_3}+{\it\rho_4} \right)}}\cosh\left(2\,\beta\,\left({\it\rho_1}+{\it\rho_2}\right)\right)+\sqrt{\gamma}[/math] And[math] \gamma\equiv{e^{\beta\,\left({\it2\rho_3}-{\it2\rho_4}\right)}}\left(\cosh\left(2\,\beta\,\left({\it\rho_1}-{\it \rho_2}\right)\right)\right)^{2}+4\,{e^{-\beta\,{\it2\rho_3}}}[/math][math]-2\,{e^{\beta\,\left({\it\rho_3}-{\it\rho_4}\right)}}{e^{\beta\,\left({\it\rho_3}+{\it\rho_4}\right)}}\cosh\left(2\,\beta\,\left({\it\rho_1}-{\it\rho_2}\right) \right)\cdot\cosh\left(2\,\beta\,\left({\it\rho_1}+{\it\rho_2}\right) \right)+{e^{\beta\,\left({\it2\rho_3}+{\it 2\rho_4}\right) }}\left(\cosh\left(2\,\beta\,\left({\it\rho_1}+{\it\rho_2}\right) \right)\right)^{2}[/math] And all the [math]\rho_i[/math] and a0 are (adjustable they contain the Isning interaction) constants.Is there a way to compute the ground state? I don't actually ask for the calculation, but just a way you might think it works (this model is analytically solvable so there has to be a way). Just straight forward calculation gets to huge even for Maple.... And I also need to calculate specific heat and friends... Quote
Qfwfq Posted September 7, 2006 Report Posted September 7, 2006 Well I'm no expert on the topic and I don't think I can help much, but I think there seems to be a problem or two with the LaTeX. The , seems to upset it. Perhaps Alex should check it up. Quote
Erasmus00 Posted September 7, 2006 Report Posted September 7, 2006 To find the ground state energy, take the temperature to 0, and then minimize the free energy(with respect to your parameters). F = E-TS, so with T=0, minimizing F minimizes E. Hopefully the complicated free energy gets less complicated as T heads to 0. As to specific heat and the like, you have the free energy, so its relatively straight forward. The derivatives will probably be awful given the complicated free energy, but I'd guess maple/mathematica could handle it. [math]S=-\frac{\partial F}{\partial T} [/math] And [math]C_v = T\frac{\partial S}{\partial T} [/math] Essentially, you have the free energy, so you are as good as finished with the problem. -Will Quote
sanctus Posted September 10, 2006 Author Report Posted September 10, 2006 Thanks for your reply Erasmus00.Well i thought as well that it would be just putting T to zero, but then 1/beta tends to zero as well and the second term diverges to -infinity. I agree with your last sentence, or at least I did agree until last week, it was a real struggle to get this free energy (and I know it is right, verified on the original paper where as usual there are no details) and then I thought now you just derive it for minimizing and as well to get specific heat, but... Well eventuallly I managed to get the specific heat for different fixed lambdas....I guess that will already be enough to be able to hand it in and get a sufficient mark. You know I'm quite interested in finishing it quickly because so I can eventually start my diploma work (or do you say master-thesis in english?)... Quote
Erasmus00 Posted September 10, 2006 Report Posted September 10, 2006 Well i thought as well that it would be just putting T to zero, but then 1/beta tends to zero as well and the second term diverges to -infinity. Doesn't the second term go to 0? [math] \frac{1}{2\beta}ln(\frac{1}{\beta^2}) = -\frac{1}{\beta}ln(\beta). [/math] So as beta heads to infinity, the log goes to infinity slower then the over beta goes to 0. -Will Quote
sanctus Posted September 12, 2006 Author Report Posted September 12, 2006 Yes you are right I saw it as well a couple of hours after I posted it, but actually for T=0 the last term tends to -infinity so there is for sure a minimum. Quote
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.