Reply to Comment on “Quantum Phase Transition of Randomly-Diluted Heisenberg Antiferromagnet on a Square Lattice” Synge Todo1∗, Hajime Takayama1, and Naoki Kawashima2 1Institute for Solid State Physics, University of Tokyo, Kashiwa 277-8581, Japan 2Department of Physics, Tokyo Metropolitan University, Tokyo 192-0397, Japan (February 1, 2008) 1 0 0 2 In Ref. [1], we have studied the quantum phase tran- sition of the two-dimensionaldiluted Heisenberg antifer- 0.05 n romagnet by using the quantum Monte Carlo method. a J Thetwomainconclusionsinthearticlearei)thecritical 0.03 1 concentration of magnetic sites is equal to the classical 32m / cl 3 percolation threshold (pcl = 0.592746), and ii) the value 0.02 slope = 0.53 of the critical exponents, such as β and Ψ, significantly ] n dependsonthespinsizeS. In[2],Sandvikhaspresented n theresultofhisquantumMonteCarlosimulationforthe 0.01 0.02 0.03 0.05 0.07 0.1 0.2 - same system only in the case of S = 1, in which the 1 / L s 2 FIG. 1. Log-log plot of cluster magnetization (Ref. [2]). di temperature dependence of the static structure factor is The solid line is obtained by least-squares fittingfor L≥28. studied quite precisely. Based on his data, he claimed . t a that the data in Ref. [1] should be seriously affected by m the finiteness of temperature in our simulation. In addi- icantly differs from those for S ≥1. Therefore, we think - tion, he calculatedthe cluster magnetizationmcl(L) just our second conclusionon the non-universalbehavior can d atp=pcl,andclaimedthatitcouldremainfiniteevenin not be excluded. n thethermodynamiclimitinthesamewayastheclassical It was also suggested [2], based on the calculation of o (S =∞)case,althoughquitelargefinite-sizecorrections the magnetization of the largest connected cluster for c [ to its asymptotic behavior should be considered in the each sample, that all the critical exponents, including quantum case. As we will discuss below, however, his Ψandβ,areequaltotheclassicalvalues. Thedatawere 2 data do not conflict with our data nor with our conclu- plotted against x ≡ 1/LD/2, and appeared to suggest a v 6 sions. In particular, the data presented in Fig. 1(b) in finite value in the thermodynamic limit. From our point 2 [2] can be explained from our view point assuming non- of view, however, his data can be interpreted as follows. 2 magnetic ground states. Thequantitym2cl(L)definedinRef.[2]shouldbeasymp- 1 As for his first claim on the effect of the temperature, totically scaled as 0 we agree that there may be a small underestimation for 01 thezero-temperaturevaluesofthestaticstructurefactor. m2cl(L)∼Ss(L,0,pcl)/L2D−d, (1) / However,evenifwereplaceourdatapointinFig.3in[1] t where S (L,T,p) is the static structure factor (Eq. (4) a for L = 48 by what we can expect from Fig. 1(a) in [2], s m in Ref. [1])and LD is the averagenumber ofspins in the it wouldresult in only a minor changein the estimate of largestconnectedclustersinasystemoflinearsizeL. Ac- - Ψ. Therefore, the error caused by this underestimation, d ifany,doesnotseemtobelargeenoughtoinvalidateour cordingtoourresults[1],ontheotherhand,Ss(L,0,pcl) n diverges as L2D−d−α with α = 0.52 for S = 1, and con- o conclusions. sequently m2(L) should be scaled as L−0.52.2 In Fig. 1, c Inaddition,inRef.[1],wehavepresentedtheresultof cl v: another finite-size scaling analysis (Fig. 4), in which the wbyeSshanowdvaiklo[2g]-.loOgnpelosteeosftthhaetctlhuestdeartmafaogrnLeti≥za2t8iocnledaartlya Xi data at higher temperatures are also used together with scalesasL−0.53,whichisfullyconsistentwithourresults the one at the lowest temperature in the simulation. In r mentioned above. this analysis, we have explicitly taken into account the a TheauthorsaregratefultoA.W.Sandvikforallowing finite-temperatureeffectonthestaticstructurefactorby us to use his numerical data. introducing the dynamical exponent z. For S = 1, the value of the exponent Ψ coincides with that obtained by theformeranalysis. Onthe otherhand,forS = 1,itbe- ∗ Present address: Theoretical Physics, ETH Zu¨rich, 8093 2 Zu¨rich, Switzerland. E-mail: [email protected] comesslightlylarger(about9%)thantheotherestimate [1] K. Kato, S. Todo, K. Harada, N. Kawashima, S. based on Fig. 3 in [1]. This small difference may be due Miyashita, and H. Takayama, Phys. Rev. Lett. 84, 4204 tothefinite-temperatureeffectpointedoutin[2],though (2000). we have not commented on it in [1]. However, even the [2] A. W.Sandvik,preprint (cond-mat/0010433). estimate based on the finite-temperature analysis signif- 1