On Ising model in magnetic field on the lattice
Raghav G. Jha · 2025
Read On Ising model in magnetic field on the lattice by Raghav G. Jha (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-18744.txt.
Introduction
Raghav Jha, a solo author, wrote a physics paper in two thousand twenty-five about the Ising model in a magnetic field. The model places tiny magnets on a square grid, and it is a key test case in physics. No exact solution is known when a steady field is applied.
Jha first offers a closed form built from a one-line chain and a known exact result. Checks against precise computer data show small but real gaps in that form. He then gives a refined form based on a known theorem about the model's zeros. The refined form matches the data far more closely and gives the right key values.
Catalog record
| CREATOR | Raghav G. Jha |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 15 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-18744
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On Ising model in magnetic field on the lattice Raghav G. Jha Abstract Abstract: We conjecture an approximate result for the free energy in the thermodynamic limit of the classical square lattice Ising model in a uniform (real) magnetic field. The zero-field result is well known due to Onsager for more than eighty years, but no such result exists for a nonzero magnetic field on a regular lattice. We verify our conjecture using numerical tensor renormalization group (TRG) methods and find good agreement with a maximum deviation of ∼ 2 % \sim 2\% from the numerical results for the free energy across all β \beta and real magnetic field, h h . I Introduction The solution of the two-dimensional Ising model (without magnetic field) was accomplished by Onsager in his famous work PhysRev.65.117 . This marked a significant milestone, as it was the first instance in which a spin model admitting a phase transition was solved exactly. Onsager computed the free energy on a square lattice in the absence of a magnetic field using complicated algebraic methods. Subsequently, Kac and Ward PhysRev.88.1332 solved it using combinatorial methods by expressing the partition function as a determinant that could be evaluated. The two-dimensional model is solvable only in the absence of a magnetic field, and no exact closed-form solution exists in the presence of a non-zero arbitrary real magnetic field on a regular lattice. In this paper, we revisit this famous unsolved problem in physics. We will focus exclusively on the square lattice. However, it should be noted that several notable results for the Ising model in the magnetic field have been obtained outside the lattice, directly in the continuum, and in the scaling limit ( T\to T_{c} and h\to 0 ) Zamolodchikov . In the absence of exact solutions for the two-dimensional model in an external field on the lattice, several variants of the Ising model that can be exactly solved have been explored. A famous work was done by Berlin and Kac PhysRev.86.821 where they studied a model similar to the Ising model, known as the ‘spherical model’ based on ferromagnetic nearest-neighbor interactions. They solved the model in the presence of an external magnetic field in three dimensions. In the same paper, they conjectured that the solution of the three-dimensional cubic Ising model can be written as an extended form of the solution for the two-dimensional case in zero field. However, this conjecture turned out to be incorrect. 1 1 1 The conjecture was well-motivated noting the similarity between the functional form of free energy for the one- and two-dimensional Ising models: see Ch. 5 of Statistical Mechanics: A Set of Lectures by R.P. Feynman for further details. Soon after, Fisher Fisher1960 studied a superexchange antiferromagnetic lattice model in which the partition function in the presence of an external field was computed. This model is closely related to the Ising model, but is solvable in uniform external field. In this paper, we conjecture an expression for the free energy of the Ising model on a square lattice in real magnetic field. We then check our expression using numerical computations based on the real-space tensor renormalization group method. We start with two-dimensional solution due to Onsager in terms of hypergeometric functions PhysRevE.83.051101 ; viswanathan2014hypergeometric and conjecture an expression for the solution in the presence of an external real magnetic fields for the two-dimensional Ising model. The maximum discrepancy between our conjecture and the numerical results (and known series expansions Itzykson:1989sy ) is \sim 2\% while for a wide range of h and \beta we explored, the agreement is observed to even higher accuracy (shown in Fig. 3 ). The result contained in this paper is neither exact nor has a proof but merely respresents a step in the direction of solution to this long-standing problem. II d=2 Ising model on square lattice The two-dimensional Ising model has been extensively studied on different regular lattices and on random graphs. The first exact solution was given for the square lattice, and then was extended to other lattices. In this section, we will first consider the standard case of a model without a magnetic field and nearest-neighbor interaction with \mathbb{Z}_{2} symmetry. II.1 Exact solution for h=0 The solution for the two-dimensional Ising model was first given by Onsager. for zero field. We will not present the full solution here but will simply quote the results from PhysRev.65.117 . This model has phase transition at a critical value (square lattice), \beta_{c}\approx 0.44069 which is obtained by solving the critical curve \sinh^{2}(2\beta_{c})=1 corresponding to the self-dual point. Exact results for both free energy and its derivatives with respect to \beta are well-known for the square lattice Ising model (see, for example, Eq. (109C) of PhysRev.65.117 ). The free energy 2 2 2 Another equivalent expression for the free energy is, \displaystyle f(\beta) \displaystyle=-\frac{1}{\beta}\Bigg(\ln 2 +\frac{1}{8\pi^{2}} \int_{0}^{2\pi}\int_{0}^{2\pi}\ln\Big[2\cosh^{2}(2\beta) \displaystyle-\sinh(2\beta)\cos(\phi_{1})-\sinh(2\beta)\cos(\phi_{2})\Big]d\phi_{1}d\phi_{2}\Bigg) is given by: \displaystyle f(\beta) \displaystyle=-\frac{1}{\beta}\Bigg(\ln(2\cosh(2\beta))+\frac{1}{2\pi^{2}}\int_{0}^{\pi}\int_{0}^{\pi}\ln \displaystyle\left(1-\frac{2\sinh(2\beta)\cos(\phi_{1}) \cos(\phi_{2})}{\cosh^{2}(2\beta)}\right)d\phi_{1}d\phi_{2}\Bigg), (1) while the internal energy is expressed in terms of an elliptic integral (Eq. (118A) of PhysRev.65.117 ) as: U=-\coth(2\beta)\Bigg(1+\frac{2}{\pi}\Big(2\tanh^{2}(2\beta)-1\Big){\sf K}(4\kappa)\Bigg), (2) where \kappa is defined as, \kappa=\sinh(2\beta)/2\cosh^{2}(2\beta) and the elliptic function {\sf K} is given by 3 3 3 Note that in Mathematica 13 , EllipticK is instead defined as: {\sf K}(z)=\int_{0}^{\frac{\pi}{2}}\frac{1}{\sqrt{1-z\sin^{2}\phi}}d\phi : {\sf K}(z)=\int_{0}^{\frac{\pi}{2}}\frac{1}{\sqrt{1-z^{2}\sin^{2}\phi}}d\phi. (3) Though, the expression derived by Onsager in Eq. ( II.1 ) is reasonably compact, one desires an alternate expression without any integrals. In fact, such a closed expression was written down only in the last decade PhysRevE.83.051101 ; viswanathan2014hypergeometric . The free energy per site of two-dimensional Ising model was given in terms of generalized hypergeometric function : f=-\frac{1}{\beta}\Bigg(\ln(2\cosh(2\beta))-\kappa^{2}\penalty\ {}_{4}F_{3}{\left[\genfrac{.}{.}{0.0pt}{}{1,1,\frac{3}{2},\frac{3}{2}}{2,2,2};16\kappa^{2}\right]}\Bigg), (4) where the generalized hypergeometric function is defined as: {}_{p}F_{q}{\left[\genfrac{.}{.}{0.0pt}{}{\textbf{a}}{\textbf{b}};z\right]}=\sum_{n=0}^{\infty} \frac{(a_{1})_{n}\cdots(a_{p})_{n}}{(b_{1})_{n}\cdots(b_{q})_{n}n!}z^{n}, (5) where a and b contains p and q elements respectively. Here, we have used (a)_{n}=\Gamma(a+n)/\Gamma(a)= a(a+1)\cdots(a+n-1) which denotes the rising factorial (also known as ‘Pochhammer symbol’). As required for convergence, we have 16\kappa^{2}\leq 1 for all \beta\geq 0 with phase transition at 16\kappa^{2}=1 with \kappa=\sinh(2\beta)/2\cosh^{2}{2\beta} . In fact, it was pointed out in viswanathan2014hypergeometric that Onsager and Glasser were very close 4 4 4 Note that Onsager’s solution in terms of double integral could have been expressed without integrals by noting the following result: \displaystyle\frac{-1}{2\pi^{2}z^{2}}\int_{0}^{\pi}\int_{0}^{\pi} \ln\Big(1-4z\cos\phi_{1}\cos\phi_{2}\Big)d\phi_{1}d\phi_{2}={}_{4}F_{3}{\left[\genfrac{.}{.}{0.0pt}{}{1,1,\frac{3}{2},\frac{3}{2}}{2,2,2};16z^{2}\right]}. We did not find this integral in Gradshteyn and Ryzhik. The reader can consult Ref. SE2023 for the proof. We thank David H for the response. in the 1970s in writing down a closed expression (unpublished) Glasser:2023wvm which was: \displaystyle f \displaystyle=-\frac{1}{\beta}\Bigg(\ln(2\cosh(2\beta))+\frac{1}{2}-\frac{1}{\pi}{\sf K}(\kappa) \displaystyle+\kappa^{2}\penalty\ {}_{4}F_{3}{\left[\genfrac{.}{.}{0.0pt}{}{1,1,\frac{3}{2},\frac{3}{2}}{2,2,2};16\kappa^{2}\right]}\Bigg), (6) where {\sf K} is the Elliptic K function. It is easy to check that this is not correct and one given by Eq. ( 4 ) is the correct expression. In terms of generalized hypergeometric function U(\beta ) is given by: \displaystyle U(\beta) \displaystyle= \frac{-1}{\cosh^{5}(2\beta)}\Bigg[2\sinh(10\beta)+4\sinh(2\beta)+ \displaystyle 6\sinh(6\beta)-\Bigg\{\Big[28\sinh(2\beta)-4\sinh(6\beta)\Big] \displaystyle{}_{3}F_{2}{\left[\genfrac{.}{.}{0.0pt}{}{1,\frac{3}{2},\frac{3}{2}}{2,2};16\kappa^{2}\right]}\Bigg\}\Bigg]. (7) It is easy to check that Eq. ( 2 ) and Eq. ( II.1 ) are equivalent. Before we proceed to discuss the conjecture for non-zero real magnetic field, we show that the numerical method we use to check our conjecture is a robust one. In this regard, we will show that we exactly reproduce the free energy results obtained from Eq. ( 4 ) for zero field case to several