-
Machine learning topological phases in real space
Authors:
N. L. Holanda,
M. A. R. Griffith
Abstract:
We develop a supervised machine learning algorithm that is able to learn topological phases of finite condensed matter systems from bulk data in real lattice space. The algorithm employs diagonalization in real space together with any supervised learning algorithm to learn topological phases through an eigenvector ensembling procedure. We combine our algorithm with decision trees and random forest…
▽ More
We develop a supervised machine learning algorithm that is able to learn topological phases of finite condensed matter systems from bulk data in real lattice space. The algorithm employs diagonalization in real space together with any supervised learning algorithm to learn topological phases through an eigenvector ensembling procedure. We combine our algorithm with decision trees and random forests to successfully recover topological phase diagrams of Su-Schrieffer-Heeger (SSH) models from bulk lattice data in real space and show how the Shannon information entropy of ensembles of lattice eigenvectors can be used to retrieve a signal detailing how topological information is distributed in the bulk. We further use insights obtained from these information entropy signatures to engineer global topological features from real space lattice data that still carry most of the topological information in the lattice, while greatly diminishing the size of feature space, thus effectively amounting to a topological lattice compression. Finally, we explore the theoretical possibility of interpreting the information entropy topological signatures in terms of emergent information entropy wave functions, which lead us to Heisenberg and Hirschman uncertainty relations for topological phase transitions. The discovery of Shannon information entropy signals associated with topological phase transitions from the analysis of data from several thousand SSH systems illustrates how model explainability in machine learning can advance the research of exotic quantum materials with properties that may power future technological applications such as qubit engineering for quantum computing.
△ Less
Submitted 27 July, 2020; v1 submitted 7 January, 2019;
originally announced January 2019.
-
From worldline to quantum superconformal mechanics with/without oscillatorial terms: $D(2,1;α)$ and $sl(2|1)$ models
Authors:
I. E. Cunha,
N. L. Holanda,
F. Toppan
Abstract:
In this paper we quantize superconformal $σ$-models defined by worldline supermultiplets. Two types of superconformal mechanics, with and without a DFF term, are considered. Without a DFF term (Calogero potential only) the supersymmetry is unbroken. The models with a DFF term correspond to deformed (if the Calogero potential is present) or undeformed oscillators. For these (un)deformed oscillators…
▽ More
In this paper we quantize superconformal $σ$-models defined by worldline supermultiplets. Two types of superconformal mechanics, with and without a DFF term, are considered. Without a DFF term (Calogero potential only) the supersymmetry is unbroken. The models with a DFF term correspond to deformed (if the Calogero potential is present) or undeformed oscillators. For these (un)deformed oscillators the classical invariant superconformal algebra acts as a spectrum-generating algebra of the quantum theory. Besides the $osp(1|2)$ examples, we explicitly quantize the superconformally-invariant worldine $σ$-models defined by the ${\cal N}=4$ $(1,4,3)$ supermultiplet (with $D(2,1;α)$ invariance, for $α\neq 0,-1$) and by the ${\cal N}=2$ $(2,2,0)$ supermultiplet (with two-dimensional target and $sl(2|1)$ invariance). The parameter $α$ is the scaling dimension of the $(1,4,3)$ supermultiplet and, in the DFF case, has a direct interpretation as a vacuum energy. In the DFF case, for the $sl(2|1)$ models, the scaling dimension $λ$ is quantized (either $λ=\frac{1}{2}+{\mathbb Z}$ or $λ={\mathbb Z}$). The ordinary two-dimensional oscillator is recovered, after imposing a superselection restriction, from the $λ=-\frac{1}{2}$ model. In particular a single bosonic vacuum is selected. The spectrum of the unrestricted two-dimensional theory is decomposed into an infinite set of lowest weight representations of $sl(2|1)$. Extra fermionic raising operators, not belonging to the original $sl(2|1)$ superalgebra, allow (for $λ=\frac{1}{2}+{\mathbb Z}$) to construct the whole spectrum from the two degenerate (one bosonic and one fermionic) vacua.
△ Less
Submitted 23 August, 2017; v1 submitted 23 October, 2016;
originally announced October 2016.
-
A world-line framework for 1D Topological Conformal sigma-models
Authors:
L. Baulieu,
N. L. Holanda,
F. Toppan
Abstract:
We use world-line methods for pseudo-supersymmetry to construct $sl(2|1)$-invariant actions for the $(2,2,0)$ chiral and ($1,2,1)$ real supermultiplets of the twisted $D$-module representations of the $sl(2|1)$ superalgebra. The derived one-dimensional topological conformal $σ$-models are invariant under nilpotent operators. The actions are constructed for both parabolic and hyperbolic/trigonometr…
▽ More
We use world-line methods for pseudo-supersymmetry to construct $sl(2|1)$-invariant actions for the $(2,2,0)$ chiral and ($1,2,1)$ real supermultiplets of the twisted $D$-module representations of the $sl(2|1)$ superalgebra. The derived one-dimensional topological conformal $σ$-models are invariant under nilpotent operators. The actions are constructed for both parabolic and hyperbolic/trigonometric realizations (with extra potential terms in the latter case). The scaling dimension $λ$ of the supermultiplets defines a coupling constant, $2λ+1$, the free theories being recovered at $λ=-\frac{1}{2}$. We also present, generalizing previous works, the $D$-module representations of one-dimensional superconformal algebras induced by ${\cal N}=(p,q)$ pseudo-supersymmetry acting on $(k,n,n-k)$ supermultiplets. Besides $sl(2|1)$, we obtain the superalgebras $A(1,1)$, $D(2,1;α)$, $D(3,1)$, $D(4,1)$, $A(2,1)$ from $(p,q)= (1,1), (2,2), (3,3), (4,4), (5,1)$, at given $k,n$ and critical values of $λ$.
△ Less
Submitted 24 July, 2015; v1 submitted 17 July, 2015;
originally announced July 2015.
-
Four types of (super)conformal mechanics: D-module reps and invariant actions
Authors:
N. L. Holanda,
F. Toppan
Abstract:
(Super)conformal mechanics in one dimension is induced by parabolic or hyperbolic/trigonometric transformations, either homogeneous (for a scaling dimension $λ$) or inhomogeneous (at $λ=0$, with $ρ$ an inhomogeneity parameter). Four types of (super)conformal actions are thus obtained. With the exclusion of the homogeneous parabolic case, dimensional constants are present. Both the inhomogeneity an…
▽ More
(Super)conformal mechanics in one dimension is induced by parabolic or hyperbolic/trigonometric transformations, either homogeneous (for a scaling dimension $λ$) or inhomogeneous (at $λ=0$, with $ρ$ an inhomogeneity parameter). Four types of (super)conformal actions are thus obtained. With the exclusion of the homogeneous parabolic case, dimensional constants are present. Both the inhomogeneity and the insertion of $λ$ generalize the construction of Papadopoulos [CQG 30 (2013) 075018; arXiv:1210.1719]. Inhomogeneous $D$-module reps are presented for the $d=1$ superconformal algebras $osp(1|2)$, $sl(2|1)$, $B(1,1)$ and $A(1,1)$. For centerless superVirasoro algebras $D$-module reps are presented (in the homogeneous case for ${\cal N}=1,2,3,4$; in the inhomogeneous case for ${\cal N}=1,2,3$). The four types of $d=1$ superconformal actions are derived for ${\cal N}=1,2,4$ systems. When ${\cal N}=4$, the homogeneously-induced actions are $D(2,1;α)$-invariant ($α$ is critically linked to $λ$); the inhomogeneously-induced actions are $A(1,1)$-invariant.
△ Less
Submitted 31 May, 2014; v1 submitted 28 February, 2014;
originally announced February 2014.
