Botet and Jullien [Phys. Rev. B 27, 613 (1983)] have conducted a finite-chain scaling analysis on the spin-1 antiferromagnetic Heisenberg-Ising chain. They claim that the ground-state properties are completely different from the case of spin
21. Using finite-size scaling techniques, they conclude that between the
XY gapless phase and the doublet-ground-state N\'eel phase, a new phase appears in an extended anisotropy range. This phase, which encompasses the isotropic Heisenberg model, is characterized by a nonmagnetic singlet ground state and a nonzero excitation-energy gap. We perform a similar finite-chain scaling analysis on the spin-
21 antiferromagnetic Heisenberg-Ising chain and obtain strikingly similar results. Since the spin-
41 system is known rigorously not to exhibit the type of behavior proposed by Botet and Jullien, their analysis is open to doubt. The possibility arises that a strong singularity at the Heisenberg point for spin 1 (similar to the essential singularity for spin
21) may be giving rise to misleading results. The
T=0 phase behavior of the spin-1 Heisenberg-Ising antiferromagnetic chain, therefore, remains in question and specific additional numerical studies are proposed to clarify the matter.
work' which also agrees very well with field theory pre- dictions once logarithmic corrections are taken into account. Our statement summarizing previous numerical work pertinentto critical behavior at the Bethe-Ansatz- solvable point was not correct.Critical exponents at this point, along with the possibility of logarithmic correc- tions, had been studied previously by Blote, Bonner, and co-workers, ' and, more recently, by Alcaraz and Martins.We are grateful to J. C. Bonner for bringing this to our attention.We have also discovered that, due to a coding error, the expansions in the dimerization parameter X for an- isotropy h, W1 were incorrect.The estimated critical points are only affected significantly for A& 1: The point C in Fig. 1 should be moved from k =0 to
There has recently been considerable interest in the numerical calculation of critical exponents for quantum spin chains with general spin s, stimulated by a remarkable prediction by Haldane and conformally invariant field theoretic calculations. In several instances numerical calculations have differed considerably from theoretical predictions, with misleading results. These discrepancies have generally been attributed to the presence of marginal operators resulting in logarithmic corrections which slow numerical convergence. Such phenomena seem to be ubiquitous in the area of quantum spin chains. However, until recently, no calculation was available for estimating quantitatively the expected shifts in exponent values obtained by numerical means. The purpose of this work is to call attention to such a calculation1 for the integrable Takhtajan–Babujian family of quantum spin chains, and make specific comparisons with a variety of spin chain exponents obtained numerically.2,3 Agreement with the analytic formulas is extremely good in all cases. This is an indication that finite-size scaling is a powerful tool for extracting critical exponents even in models with logarithmic corrections, provided that these corrections are taken into account in lowest order using the method of Ref. 1.