Exploring the variational method for thermodynamic models

This work explores the possibilities of the Gibbs–Bogoliubov–Feynman variational method, aiming at finding room for designing various drawing schemes. For example, mean-field approximation can be viewed as a result of using site-independent drawing in the variational method. In subsequent sections, progressively complex drawing procedures are presented, starting from site-independent drawing in the $k$-space. In the next, each site in the real-space is again drawn independently, which is followed by an adjustable linear transformation $T$. Both approaches are presented on the discrete Ginzburg–Landau model. Subsequently, a percolation-based procedure for the Ising model is developed. It shows a general way of handling multi-stage drawing schemes. Critical inverse temperatures are obtained in two and three dimensions with a few percent discrepancy from exact values. Finally, it is shown that results in the style of the real-space renormalization group can be achieved by suitable fractal-like drawing. This facilitates a new straight-forward approach to establishing the renormalization transformation, but primarily provides a new view on the method. While the first two approaches are capable of capturing long-range correlations, they are not able to reproduce the critical behavior accurately. The main findings of the paper are developing the method of handling intricate drawing procedures and identifying the need of fractality in these schemes to grasp the critical behavior.