Detection of Symmetry Protected Topological Phases with MERA

• Main Authors: Hsueh-Wen Chang, 張學文
• Other Authors: Ying-Jer Kao
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/97595838266399644005

碩士 === 國立臺灣大學 === 物理研究所 === 101 === Traditionally, we understand phases of matter by spontaneous symmetry breaking and therefore we have order parameters to characterize different phases. However, there are states of matter which fall beyond this type of characterization, we call these phases topological phases. The first experimental discovery of topological phase was at 1980s, however, we still do not have a clear understanding of topological phase nor do we know exactly what the topological orders are for characterizing these phases (in contrast with normal orders). Recently, using the ideas based on local unitary transformations, it is shown that an topological order is associated with the pattern of long-range entanglement in gapped quantum systems with finite topological entanglement entropy. On the other hand, several nonlocal string-order parameters are proposed for characterizing symmetry-protected topological states for different kinds of symmetries in 1D base on infinite-size matrix product state (iMPS) calculation . iMPS is a powerful tool to represent ground state of 1D gapped system. However, accuracy decrease near critical point. Multiscale entanglement renormalization ansatz (MERA) on the other hand, is more accurate near critical point. MERA is a numerical method that use the concept of entanglement renormalization, with the help of inserting disentanglers into the system so that short-range entanglement can be removed before renormalization. This prevents the accumulation of degrees of freedom during the renormalization so that this method can be addressed to large size of 1D or 2D system even at quantum critical points. Therefore, if one can use MERA to calculate the nonlocal string-order parameters proposed, not only will it be a confirmation, but also better results could be get. However, calculation of these parameters with MERA turned out to be highly nontrivial because the operator is nonlocal and it breaks the causal cone property of MERA, which makes the calculation exponentially hard. Nevertheless, we found that we can reduce the calculation by using symmetric MERA. We need Z_2 and inversion symmetry. The former is already proposed but there''s no method to do inversion symmetry in any kind of tensor network until now. By inspecting the geometry of MERA, we successfully find the method to incorporate inversion symmetry in MERA with a brick-and-rope representation. These enable people to calculate the inversion string-order parameter with MERA. We calculate the proposed parameters and confirmed that the Haldane phase is protected by both inversion and time reversal symmetry. This thesis is structured as follows: In Chap. 1, we give a brief introduction to topological phase, including historical review and its relation with quantum entanglement and symmetry. Followed by the review of S = 1 Haldane chain. In Chap. 2, we review different kinds of tensor network methods and focus onMERA, with a detailed explanation and interpretation of the tensors in it. In Chap. 3, we discuss how to implement a Z2 symmetric MERA and introduce our algorithm on spatial inversion symmetric MERA based on a brick-and-rope representation. In Chap. 4, we present the results of stringorder parameters using both time-reversal and inversion symmetric MERA. We also observe the RG flow in the 2 phase goes to 2 different fixed point. We conclude our work in Chap. 5.