betaFeedback

WorksPhysics and Astronomy

Impedance responses and size-dependent resonances in topolectrical circuits via the method of images

Abstract

Resonances in an electric circuit occur when capacitive and inductive components are present together. Such resonances appear in admittance measurements depending on the circuit's parameters and the driving AC frequency. In this study, we analyze the impedance characteristics of nontrivial topolectrical circuits such as one- and two-dimensional Su-Schrieffer-Heeger circuits and reveal that size-dependent anomalous impedance resonances inevitably arise in finite LCLC circuits. Through the method of images, we study how resonance modes in a multi-dimensional circuit array can be nontrivially modified by the reflection and interference of current from the structure and boundaries of the lattice. We derive analytic expressions for the impedance across two corner nodes of various lattice networks with homogeneous and heterogeneous circuit elements. We also derive the irregular dependency of the impedance resonance on the lattice size, and provide integral and dimensionally-reduced expressions for the impedance in three dimensions and above.

Introduction

Electric circuit networks are extremely versatile platforms for simulating a variety of condensed matter phenomena through their tight-binding representations [1, 2, 3]. For instance, a uniform tiling of LCLC oscillators composed of capacitor and inductor pairs gives rise to an AC signal propagating along an ideal transmission line, which simulates the lattice dynamics of a one-dimensional (1D) solid-state medium. Circuits that mimic condensed matter lattices, known as topolectrical (TE) circuits, have been extremely successful in demonstrating a wide range of topological and critical condensed matter phenomena [4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 20] [19]. In these circuits, topologically protected zero modes can be measured as impedance resonances at the resonant frequency. Beyond the simulation of linear condensed matter systems, the simulation of complex networks processes such as search algorithms has also been proposed through the use of non-linear circuit elements [21, 22, 23, 24, 25, 26].

While tight-binding lattice models are usually faithful in representing their respective solid state systems [27, 28, 29, 30, 31, 32, 33, 34], their intrinsic discreteness sometimes leads to additional anomalous contributions to the impedance with no analog in the continuum limit. In this work, we present an analytic formulation for computing the impedance across various finite circuit arrays. We also investigate the anomalous impedance behavior that emerges when components with different phase lags are simultaneously present in a finite electric circuit, which we call a heterogeneous circuit. We focus more specifically on heterogeneous circuits in which the nodes are coupled by a mixture of capacitors and inductors.

The electric potential distribution in electric circuit networks satisfies Kirchhoff’s laws and can be described by the circuit Laplacian. The solutions of the circuit Laplacian contain all available information about the potential distribution. Therefore, any desired computation can be performed by solving the circuit Laplacian with appropriate boundary conditions. Although the two-point impedance in homogeneous circuits has been widely studied, most results pertaining to the condensed matter context are valid only for infinite circuit networks [35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88]. Two issues that present challenges in forming full analogies with condensed matter are the finite circuit boundaries (which differ from the usual open boundary conditions) and the homogeneity of the lattice array; while computations can certainly be performed numerically, a comprehensive analytical expression for the impedance in heterogeneous finite circuits has yet to be obtained. One way to implement boundary conditions in electrostatic theory is through the method of images, in which the boundaries are replaced by image charges located opposite the original charges [89, 90, 91, 92, 93]. As a demonstration of this approach, we apply the method of images to periodic tilings of finite electric circuit networks to compute the two-point impedance of the finite circuit networks.

While the impedance generally scales in a logarithmic manner with the circuit size in homogeneous circuits [39, 46, 64, 77, 86, 94, 95, 96], the same is not true in heterogeneous circuits, where the impedance can deviate very strongly from logarithmic scaling [97] at certain circuit sizes. The circuit size NN thus becomes a functional parameter alongside LCLC and the driving AC frequency ω\omega. These independent parameters collectively affect the impedance resonances. This is in contrast to a waveguide or transmission line, in which the system size does not affect the behavior of the system in ideal cases. Moreover, the discreteness of NN results in fractal-like resonances when NN is varied at fixed LL and CC values.

In this work, we study the size-dependent impedance resonances both numerically and analytically. We derive analytic expressions for the size-dependent impedance between two opposite corner nodes by utilizing the method of images. To reveal the origin of this anomalous impedance behavior, we examine circuits with a single node per unit cell, as well as those with nontrivial unit cells containing more than one node. As paradigmatic examples of the latter, we present detailed calculations for 1D and two-dimensional (2D) circuit lattices with Su-Schrieffer-Heeger (SSH) type dimerizations. As for homogeneous circuits with a single-type node per unit cell, we start from a 1D circuit and build higher-dimensional circuits by linking every node along the new direction with the same type of component (i.e., resistor, inductor, or capacitor). In the heterogeneous circuits section, we follow the same process but introduce at least two different types of components with different phase shifts. Finally, we discuss the emergent fractal-like structures that arise from the violated logarithmic scaling in LCLC circuits.

Formalism for the two-point impedance

We first review the generic derivation of the expression for the two-point impedance in terms of the eigenvalues and eigenvectors of the circuit Laplacian [43, 44, 46, 75, 94, 98, 99]. A RLC circuit can be represented as a graph in which the vertices of the graph represent the voltage nodes and the edges represent the couplings between the nodes due to RR, LL, and CC components between the nodes. Under driving at a single AC frequency ω\omega, the circuit can be mathematically represented by its Laplacian matrix JJ, which relates the currents injected into the nodes with the voltages at each node via

I=JV(1)I = JV \tag*{(1)}

where II is a vector of the currents injected into each node and VV the corresponding vector of the node voltages. The Laplacian matrix for a circuit can be obtained simply by writing Kirchhoff’s current law at each voltage node. For example, consider a simple circuit consisting of two voltage nodes connected by a single capacitor with capacitance CC. Applying Kirchhoff’s current law at the two nodes gives I1=iωC(V1−V2)I_1 = i\omega C(V_1 - V_2) and I2=iωC(V2−V1)I_2 = i\omega C(V_2 - V_1) where IaI_a and VaV_a are the injected current and voltage at node aa, respectively. Using Eq. (1) and these relations between the node current and voltages, the Laplacian matrix JJ of this simple circuit is then given by

J=iωC(1−1−11).J = i\omega C \begin{pmatrix} 1 & -1 \\ -1 & 1 \end{pmatrix}.

By definition, the impedance between two nodes ii and jj is the factor of proportionality between the voltage difference Vi−VjV_i - V_j that develops between the two nodes when a current of magnitude II is injected into node ii and extracted at node jj:

Zij=Vi−VjI.(2)Z_{ij} = \frac{V_i - V_j}{I}. \tag*{(2)}

To determine the impedance between two nodes, the voltages ViV_i and VjV_j have to be determined. To achieve this, we employ the circuit Green’s function GG, which is defined as the pseudo-inverse of the circuit Laplacian G=J−1G = J^{-1}. Using the Green’s function, Eq. (1) can be rewritten as V=GIV = GI. The electrical voltage at node ii can thus be expressed as

Vi=∑j=1NGijIj,(3)V_i = \sum_{j=1}^{N} G_{ij}I_j, \tag*{(3)}

where GijG_{ij} is the (i,j)(i,j) element of the pseudoinverse matrix GG. With Ii=II_i = I and Ij=−II_j = -I in Eq. (3), Eq. (2) can be rewritten as

Zij=∑k=i,jGikIk−GjkIkI,(4)Z_{ij} = \sum_{k=i,j} \frac{G_{ik}I_k - G_{jk}I_k}{I}, \tag*{(4)}

which gives

Zij=Gii+Gjj−Gij−Gji.(5)Z_{ij} = G_{ii} + G_{jj} - G_{ij} - G_{ji}. \tag*{(5)}

By resolving in the space of eigenstates, JJ can be written in terms of its right eigenvectors ∣ψk⟩|\psi_k\rangle, left eigenvectors ⟨ψk∣\langle\psi_k|, and eigenvalues λk\lambda_k as J=∑k∣ψk⟩λk⟨ψk∣J = \sum_k |\psi_k\rangle\lambda_k\langle\psi_k|. In general, ∣ψk⟩≠⟨ψk∣†|\psi_k\rangle\ne\langle\psi_k|^\dagger, because JJ is not Hermitian. However, ∣ψk⟩≠⟨ψk∣†|\psi_k\rangle\ne\langle\psi_k|^\dagger holds in the following special cases: (i) in a purely LCLC circuit, where JJ is anti-Hermitian and the λk\lambda_k’s are imaginary, and (ii) in a purely resistive circuit, where JJ is Hermitian and the λk\lambda_k are real. For a general RLCRLC circuit, the λk\lambda_k are complex, and the general relation G=∑k∣ψk⟩(λk)−1⟨ψk∣G = \sum_k |\psi_k\rangle(\lambda_k)^{-1}\langle\psi_k| holds. Note that because GG is defined as the pseudoinverse of JJ, any zero eigenvalues are excluded from the sum if they exist. Writing ∣ψk⟩|\psi_k\rangle as a column vector of the node voltages ∣ψk⟩={ψk,1,ψk,2,…,ψkN}|\psi_k\rangle= \{\psi_{k,1},\psi_{k,2},\ldots,\psi_{kN}\} where NN is the total number of nodes, Eq. (5) can rewritten as

Zij=∑k,λk≠0∣ψki−ψkj∣2λk,(6)Z_{ij} = \sum_{k,\lambda_k\ne0} \frac{|\psi_{ki} - \psi_{kj}|^2}{\lambda_k}, \tag*{(6)}

where λk\lambda_k is the corresponding eigenvalue of the Laplacian, and the ∣…∣|\ldots| norm is the biorthogonal norm. Any arbitrary two-point impedance can then be numerically calculated using Eq. (2), Eq. (5), or Eq. (6).

Method of images for analytic impedance formulas across bounded circuit lattices

In this section, we review and derive the general analytical formalism for calculating the impedance between two edge or corner nodes in finite discrete circuit lattices based on the method of images and discuss their impedance behaviors. The method of images is needed to put finite lattice impedance computations, which are not easily represented by exact analytic formulas due to the lack of translation symmetry, on equal footing with periodic lattices. We shall illustrate the approach with exemplary circuit arrays with nontrivial unit cells, such that the tiling of the periodic images is not trivial.

In the context of electrostatic theory, the electric potential distribution inside an area of interest in the vicinity of a grounded conducting plate can be obtained by placing an image charge reflected across the conducting plate [89, 90, 91, 92, 93, 100, 101, 102, 103]. The image charge causes the surface of the plate to become equipotential (which is considered zero for a grounded conductor) and thus satisfy the boundary conditions of a grounded conducting plate.

Inspired by this approach, we apply an analogous idea by placing image circuits symmetrically about our desired circuit to replicate the boundary conditions satisfied by a finite circuit under open boundary conditions [93, 104]. The translation and inversion symmetries obeyed by the original and image circuits in our method force the voltage potentials at the boundary nodes of the original and the image circuits to be the same(see Figure 2(b). The equal potentials at the boundary nodes result in zero current flow across the boundaries between the physical and image circuits, and therefore replicates the boundary condition of zero current flow across the open physical boundaries of the finite bounded circuit. The potential distribution of the original circuit is then the same as that of the finite circuit with physical boundaries. This is the key insight that underlies our derivation of analytical expressions for the potential profiles of circuits with open boundaries via the method of images.

Figure 2. The matrix representation of the periodic circuit Laplacian of the 2D SSH circuit for N=2N=2 and its spatial voltage distribution for N=10N=10. (a) While the admittances between the node links are represented by the red, green, blue, and black squares in the off-diagonal elements, the darker red squares in the diagonal elements represent the total node conductance. (b) Spatial voltage distribution matrix of the 2D SSH circuit for N=10N=10 presented as a density plot. The red dashed lines separating the entire matrix into four blocks represent the boundaries between the physical and image circuits. The colors of the squares represent the magnitude of the voltage potential. Identical colors on both sides of the red dashed lines imply that there are zero potential differences between the boundary nodes. The red and white circles with black frames represent the nodes where the current is injected and extracted, respectively.

Example: 1D SSH circuits

To illustrate how the method of images yields exact analytical expressions, we first consider sufficiently non-trivial circuit arrays in which each unit cell contains more than one type of node, such as the 1D and 2D topological Su-Schrieffer-Heeger (SSH) circuits. In such lattices, the transitions from the topologically trivial to non-trivial phases are accompanied by strong impedance resonances at the resonant frequency [4, 5, 6, 8, 18, 105, 106, 107]. However, our focus here is on their edge-to-edge (1D SSH) and corner-to-corner (2D SSH) impedance profiles rather than the topological impedance characteristics.

The 1D SSH circuit consists of unit cells in which each unit cell contains two distinct nodes labeled as s1s_1 and s2s_2 where the nodes are connected to each other by intra-cell capacitors of capacitance C1C_1 and inter-cell capacitors of capacitance C2C_2 [see Figure 1(a)]. The corresponding 2×22 \times2 circuit Laplacian of a periodic 1D SSH circuit in momentum space is therefore written as

Figure 1. Lattice structures of illustrative SSH circuits and their periodic image lattices for finite-lattice impedance computations. (a) The 1D SSH circuit comprises two capacitors (C1C_1 and C2C_2) and two nodes (s1s_1 and s2s_2) in a unit cell. (b) The superlattice comprising the physical and image 1D SSH circuits is denoted by the blue dashed box, so constructed such that the impedance through a finite single block is recast into a problem with periodically placed injected and extracted currents. The green and purple dashed boxes denote the physical circuit and its unit cell, respectively. (c) The physical 2D SSH circuit consisting of two kinds of capacitors with capacitances C1C_1 (bold) and C2C_2 (thin) connecting the nodes along the horizontal direction and two kinds of inductors with inductances L1L_1 (bold) and L2L_2 (thin) linking the nodes along the vertical direction. The four distinct nodes in a unit cell (purple dashed square) s1s_1, s2s_2, s3s_3, and s4s_4 are represented as magenta, yellow, orange, and cyan circles, respectively. To measure the corner-to-corner impedance, current is injected at node s1s_1 of unit cell (N+1,N+1)(N+1,N+1) and extracted at node s4s_4 of unit cell (2N,2N)(2N,2N) [also see the matrix representation in Eq. (21)]. (d) The infinite periodic lattice tiling of the (2N)2(2N)^2-unit cell superlattice consisting the image and physical circuits (blue dashed square), which is constructed such that the impedance across the finite lattice can be expressed in terms of translation-invariant momentum contributions. The nodes at which current is injected and extracted are denoted with green and black outlines, respectively.

L1DSSH(k1)=iω(C1+C2−C1−C2e−ik1−C1−C2eik1C1+C2)(7)\mathcal{L}^{\mathrm{SSH}}_{\mathrm{1D}}(k_1) = i\omega\begin{pmatrix} C_1 + C_2 & -C_1-C_2e^{-ik_1} \\ -C_1-C_2e^{ik_1} & C_1+C_2 \end{pmatrix} \tag*{(7)}

where ω\omega is the driving AC frequency and k1k_1 is the crystal momentum along the xx direction. We label the nodes in the SSH chain as r~=(r,μ)\tilde{\mathbf{r}}=(\mathbf{r},\mu) where μ∈(s1,s2)\mu\in(s_1,s_2) denotes the node within each unit cell, r=n1a1\mathbf{r}=n_1\mathbf{a}_1 denotes the location of the unit cell with a1\mathbf{a}_1 being the unit vector along the length of the chain and n1n_1 the coordinate of the unit cell, and the tilde on r~\tilde{\mathbf{r}} denotes a composite index consisting of both the location of the unit cell and the sub-lattice site.

To investigate the behavior of the impedance as the circuit size increases, we consider the impedance between node s1s_1 in the leftmost unit cell and node s2s_2 in the rightmost unit cell in the 1D SSH chain circuit. Note that the circuit size is increased unit cell-by-unit cell so as to preserve lattice uniformity. Therefore, NN refers to the number of unit cells throughout this study. Henceforth, we shall call the actual circuit with open boundaries (the green dashed rectangles in Figure 1(b) and 1(d)) the physical lattice or the physical circuit, the image(s) of the actual circuit the image lattice(s) or the image circuit(s), and the block containing physical and image circuits (the blue dashed rectangles in Figure 1(b) and 1(d)) the superlattice or the supercircuit. To obtain the voltages in a finite SSH chain containing NN unit cells that result from current injection at the left-most node and current extraction at the right-most node, we place an image chain of the same length to the left of physical chain circuit, but importantly reflected about the chain boundary so that no current flows across it due to reflection symmetry. We then inject current at the right-most node of the image circuit and the left-most node of the physical circuit, and extract the injected current at the left-most node of the image circuit and the right-most node of the physical circuit, as shown in Figure 1(b). The impedance between any two lattice points can then be obtained using by Eq. (2) once the voltage distribution is known.

To determine the voltage distribution explicitly, we utilize Ohm’s law, which states that the current distribution J=σE\mathcal{J}=\sigma\mathbf{E} where the electric field is given by E=−∇V\mathbf{E}=-\nabla V. Therefore, the current distribution can be written as J=−z−1∇V\mathcal{J}=-z^{-1}\nabla V, where zz, the uniform impedance between each node, is the inverse of the electrical conductivity σ\sigma. Owing to Kirchhoff’s current law, the current density is also written as ∇⋅J=I(δ(r′,ν)∈r~in(r,μ,r′,ν)−δ(r′,ν)∈r~out(r,μ,r′,ν))\nabla\cdot\mathcal{J}=I\left(\delta_{(\mathbf{r}',\nu)\in\tilde{\mathbf{r}}_{\mathrm{in}}}(\mathbf{r},\mu,\mathbf{r}',\nu)-\delta_{(\mathbf{r}',\nu)\in\tilde{\mathbf{r}}_{\mathrm{out}}}(\mathbf{r},\mu,\mathbf{r}',\nu)\right) where δ\delta denotes the Kronecker delta and II represents the current magnitude. After performing the relevant substitutions, we arrive at a Poisson-type equation. Here, because the Green’s function satisfies ∇2G(r,μ,r′,ν)=−δ(r,μ,r′,ν)\nabla^2G(\mathbf{r},\mu,\mathbf{r}',\nu)=-\delta(\mathbf{r},\mu,\mathbf{r}',\nu) by definition [58, 69, 108, 109, 110, 111] (note that we have rewritten the circuit Green’s function GG of Sect. II in the quasi-continuum picture to facilitate the discussion), the voltage at sub-lattice node μ\mu of the unit cell at location r\mathbf{r} is found as

V(r,μ)=I(∑(r′,ν)∈r~inG(r,μ,r′,ν)−∑(r′,ν)∈r~outG(r,μ,r′,ν))(8)V(\mathbf{r},\mu)=I\left(\sum_{(\mathbf{r}',\nu)\in\tilde{\mathbf{r}}_{\mathrm{in}}}G(\mathbf{r},\mu,\mathbf{r}',\nu)-\sum_{(\mathbf{r}',\nu)\in\tilde{\mathbf{r}}_{\mathrm{out}}}G(\mathbf{r},\mu,\mathbf{r}',\nu)\right) \tag*{(8)}

where

r~in={(Na1,s2),((N+1)a1,s1)},\tilde{\mathbf{r}}_{\mathrm{in}}=\{(N\mathbf{a}_1,s_2),((N+1)\mathbf{a}_1,s_1)\},
r~out={(1a1,s1),(2Na1,s2)}(9)\tilde{\mathbf{r}}_{\mathrm{out}}=\{(\mathbf{1}\mathbf{a}_1,s_1),(2N\mathbf{a}_1,s_2)\} \tag*{(9)}

denote the nodes where the currents are injected (r~in\tilde{\mathbf{r}}_{\mathrm{in}}) and extracted (r~out\tilde{\mathbf{r}}_{\mathrm{out}}). The spatial Green’s function G(r,μ,r′,ν)G(\mathbf{r},\mu,\mathbf{r}',\nu) in Eq. (8) can be determined by applying the discrete Fourier transform given by

G(r,μ,r′,ν)=1(2N)D∑kG(k)[μ,ν]eik⋅(r−r′)(10)G(\mathbf{r},\mu,\mathbf{r}',\nu)=\frac{1}{(2N)^D}\sum_{\mathbf{k}}G(\mathbf{k})_{[\mu,\nu]}e^{i\mathbf{k}\cdot(\mathbf{r}-\mathbf{r}')} \tag*{(10)}

where the momentum space index k=k1a1\mathbf{k}=k_1\mathbf{a}_1 in which k1=n1π/Nk_1=n_1\pi/N and n1n_1 varies over a 2N2N period and G(k)[μ,ν]G(\mathbf{k})_{[\mu,\nu]} is the matrix element of the pseudoinverse of the momentum-space circuit Laplacian [for this example, it is L1DSSH(k1)\mathcal{L}^{\mathrm{SSH}}_{\mathrm{1D}}(k_1) given in Eq. (7)]. DD represents the circuit dimension. We then tile the 2N2N unit cells comprising the physical and image circuits to form an infinite-sized lattice with a period of 2N2N. Therefore, the Green’s function and the circuit Laplacian are constructed for the superlattice with a period of 2N2N such that the symmetric current injections and extractions in this periodic infinite lattice lead to a symmetric spatial voltage distribution with the period of 2N2N; hence, the current entering the physical circuit cannot leak out through the boundaries of the physical circuit. Accordingly, the open boundary condition for the physical circuit with NN unit cells is satisfied. To realize this, we use the current distribution Eq. (9) with Eq. (8) to determine the voltage at the left-most node of the physical circuit as

V(r=N+1,μ=s1)=I(G((N+1),s1,(N),s2)+G((N+1),s1,(N+1),s1)−G((N+1),s1,(1),s1)−G((N+1),s1,(2N),s2))(11)V(\mathbf{r}=N+1,\mu=s_1)=I\big(G((N+1),s_1,(N),s_2) +G((N+1),s_1,(N+1),s_1) -G((N+1),s_1,(1),s_1) -G((N+1),s_1,(2N),s_2)\big) \tag*{(11)}

and that at the rightmost edge node as

V(r=2N,μ=s2)=I(G((2N),s2,(N),s2)+G((2N),s2,(N+1),s1)−G((2N),s2,(1),s1)−G((2N),s2,(2N),s2)).(12)V(\mathbf{r}=2N,\mu=s_2)=I\big(G((2N),s_2,(N),s_2) +G((2N),s_2,(N+1),s_1) -G((2N),s_2,(1),s_1) -G((2N),s_2,(2N),s_2)\big). \tag*{(12)}

To find the voltages explicitly, we insert the momentum-space Green’s function given in Eq. (10) into Eqs. (11) and (12) so that the impedance between the two edge nodes can be calculated as Z1DSSH=V((N+1),s1)−V((2N),s2)Z_{1D}^{\mathrm{SSH}} = V((N+1),s_1)-V((2N),s_2). At this point, we utilize the inversion and translation symmetries (i.e., J(r)=J(−r)J(\mathbf{r})=J(-\mathbf{r}) and J(r)=JT(r)J(\mathbf{r})=J^{\mathsf{T}}(\mathbf{r}) (where (T)(\mathsf{T}) denotes the transpose operation), respectively) that our circuit possesses [47, 96, 110, 111, 112]. Because of the infinite periodic tiling, these symmetries imply that the voltage at the nodes where current is injected has the same magnitude as that at the nodes where current is extracted but with the opposite sign, i.e., V(r~∈r~in)=−V(r~∈r~out)V(\tilde{\mathbf{r}}\in\tilde{\mathbf{r}}_{\mathrm{in}})=-V(\tilde{\mathbf{r}}\in\tilde{\mathbf{r}}_{\mathrm{out}}). Therefore, by using these symmetries and performing the relevant substitutions, the voltages at the edge nodes are found to be

V((N+1),s1)=−V((2N),s2)=I4N∑n1=12N(ein1π−1)(C2(1−e−in1π)+C1(1−e−in1πein1π/N))iωC1C2(1−cos⁡(n1π/N))(13)V((N+1),s_1)=-V((2N),s_2)=\frac{I}{4N}\sum_{n_1=1}^{2N}\frac{(e^{in_1\pi}-1)\left(C_2(1-e^{-in_1\pi})+C_1(1-e^{-in_1\pi}e^{in_1\pi/N})\right)}{i\omega C_1C_2\left(1-\cos(n_1\pi/N)\right)} \tag*{(13)}

where we introduced r=n1a1\mathbf{r}=n_1\mathbf{a}_1 and k=k1a1\mathbf{k}=k_1\mathbf{a}_1 where k1=n1π/Nk_1=n_1\pi/N. Here, because ein1π−1=0e^{in_1\pi}-1=0 when the integer n1n_1 is even, the summation is performed only for odd n1n_1s. By considering Z1DSSH=2V((N+1),s1)/I=−2V((2N),s2)/IZ_{1D}^{\mathrm{SSH}}=2V((N+1),s_1)/I=-2V((2N),s_2)/I, and by means of trigonometric conversions [e.g., 1+ein1π/N=1+cos⁡(n1π/N)+isin⁡(n1π/N)1+e^{in_1\pi/N}=1+\cos(n_1\pi/N)+i\sin(n_1\pi/N) where the sine function can be neglected because of its zero contribution to the real part of the impedance], the edge-to-edge impedance as a function of the circuit size NN is given by

Z1DSSH(N)=1N∑k1∗2C2/C1+(1+cos⁡k1)−iωC2(1−cos⁡k1)(14)Z_{1D}^{\mathrm{SSH}}(N)=\frac{1}{N}\sum_{k_1}^{*}\frac{2C_2/C_1+(1+\cos k_1)}{-i\omega C_2(1-\cos k_1)} \tag*{(14)}

As before, k1=n1π/Nk_1=n_1\pi/N where n1n_1 is varied over the superlattice, i.e., n1∈{1,2,…,2N}n_1\in\{1,2,\ldots,2N\}; however, the summation is restricted over n1∈oddn_1\in\mathrm{odd} because of the above-mentioned summation rule, which the asterisk (∗*) on the summation operator indicates.

Using the same procedure, the edge-to-edge impedance in a circuit where the C1C_1 capacitors are replaced by capacitors with capacitance CC and the C2C_2 capacitors by inductors with inductance LL is obtained as

Z1DSSHL(N)=1N∑k1∗2−ω2CL(1+cos⁡k1)−iωC(1−cos⁡k1)(15)Z_{1D}^{\mathrm{SSH_L}}(N)=\frac{1}{N}\sum_{k_1}^{*}\frac{2-\omega^2CL(1+\cos k_1)}{-i\omega C(1-\cos k_1)} \tag*{(15)}

These formulas provide the impedance between nodes s1s_1 and s2s_2 in the leftmost and rightmost unit cells, respectively. Note that our sum includes a total of NN points, which corresponds to the NN unit cells.

Example: 2D SSH circuit

We now proceed with a higher-dimensional circuit in which all the principal directions are non-trivial. For example, a 2D SSH circuit can be constructed by extending the 1D SSH circuit along the new yy direction. We first consider a capacitive 1D SSH circuit with intracell coupling C1C_1 and intercell coupling C2C_2. We then extend the circuit along the yy direction by using two inductances L1L_1 and L2L_2 to connect the nodes along the vertical direction, as shown in Figure 1(c). The resultant unit cell [dashed purple square in Figure 1(c)] has four distinct nodes denoted as s1s_1 to s4s_4 in which nodes s1s_1 and s4s_4 are located at opposite corners of the unit cell. Since preserving the uniformity of the unit cells requires the circuit size to be increased in multiples of the unit cells, the corner-to-corner impedance is measured between node s1s_1 in the first unit cell and node s4s_4 in the unit cell at the opposite corner of the 2D SSH circuit. The circuit Laplacian for a periodic 2D SSH circuit in momentum space is written as

L2DSSH(k1,k2)=iω(ΣΓΔ0Γ∗Σ0ΔΔ∗0ΣΓ0Δ∗Γ∗Σ)(16)\mathcal{L}_{2D}^{\mathrm{SSH}}(k_1,k_2)=i\omega \begin{pmatrix} \Sigma& \Gamma& \Delta& 0\\ \Gamma^{*} & \Sigma& 0 & \Delta\\ \Delta^{*} & 0 & \Sigma& \Gamma\\ 0 & \Delta^{*} & \Gamma^{*} & \Sigma \end{pmatrix} \tag*{(16)}

where Γ=−C1−C2e−ik1\Gamma=-C_1-C_2e^{-ik_1}, Δ=1ω2L1+1ω2L2e−ik2\Delta=\frac{1}{\omega^2L_1}+\frac{1}{\omega^2L_2}e^{-ik_2}, and Σ=C1+C2−1ω2L1−1ω2L2\Sigma=C_1+C_2-\frac{1}{\omega^2L_1}-\frac{1}{\omega^2L_2}. Similar to the 1D SSH circuit, we evaluate the voltage distribution for the impedance measurement by introducing image circuits around the physical circuit [Figure 1(d)], injecting current at r~in\tilde{\mathbf{r}}_{\mathrm{in}}, and extracting current at r~out\tilde{\mathbf{r}}_{\mathrm{out}} where

r~in∈{((N+1,N+1),s1),((N,N+1),s2),((N+1,N),s3),((N,N),s4)}r~out∈{((1,1),s1),((2N,1),s2),((1,2N),s3),((2N,2N),s4)}.(17)\begin{aligned} \tilde{\mathbf{r}}_{\mathrm{in}} &\in\{((N+1,N+1),s_1),((N,N+1),s_2),((N+1,N),s_3),((N,N),s_4)\}\\ \tilde{\mathbf{r}}_{\mathrm{out}} &\in\{((1,1),s_1),((2N,1),s_2),((1,2N),s_3),((2N,2N),s_4)\}. \tag*{(17)} \end{aligned}

Here, the spatial positions of the nodes at which current is injected and extracted are written in the form of ((n1a1,n2a2),sα)((n_1\mathbf{a}_1,n_2\mathbf{a}_2),s_\alpha) where α=(1,2,3,4)\alpha=(1,2,3,4) and n2n_2 and a2\mathbf{a}_2 are the coordinate and unit vector along the yy direction, respectively. The voltage V(r=(N+1,N+1),μ=s1)V(\mathbf{r}=(N+1,N+1),\mu=s_1) at the lower left corner of the physical circuit [see Figure 1(d)] is thus given by

V(r=(N+1,N+1),μ=s1)=I(G((N+1,N+1),s1,(N+1,N+1),s1)+G((N+1,N+1),s1,(N,N+1),s2)+G((N+1,N+1),s1,(N+1,N),s3)+G((N+1,N+1),s1,(N,N),s4)−G((N+1,N+1),s1,(1,1),s1)−G((N+1,N+1),s1,(2N,1),s2)−G((N+1,N+1),s1,(1,2N),s3)−G((N+1,N+1),s1,(2N,2N),s4)).(18)V\left(\mathbf{r}=(N+1,N+1),\mu=s_1\right)= I\Bigl(G((N+1,N+1),s_1,(N+1,N+1),s_1)+G((N+1,N+1),s_1,(N,N+1),s_2) +G((N+1,N+1),s_1,(N+1,N),s_3)+G((N+1,N+1),s_1,(N,N),s_4)-G((N+1,N+1),s_1,(1,1),s_1) -G((N+1,N+1),s_1,(2N,1),s_2)-G((N+1,N+1),s_1,(1,2N),s_3)-G((N+1,N+1),s_1,(2N,2N),s_4)\Bigr). \tag*{(18)}

Similarly, the voltage at the upper right corner of the physical circuit is given by

V(r=(2N,2N),μ=s4)=I(G((2N,2N),s4,(N+1,N+1),s1)+G((2N,2N),s4,(N,N+1),s2)+G((2N,2N),s4,(N+1,N),s3)+G((2N,2N),s4,(N,N),s4)−G((2N,2N),s4,(1,1),s1)−G((2N,2N),s4,(2N,1),s2)−G((2N,2N),s4,(1,2N),s3)−G((2N,2N),s4,(2N,2N),s4)).(19)V\left(\mathbf{r}=(2N,2N),\mu=s_4\right)= I\Bigl(G((2N,2N),s_4,(N+1,N+1),s_1)+G((2N,2N),s_4,(N,N+1),s_2)+G((2N,2N),s_4,(N+1,N),s_3) +G((2N,2N),s_4,(N,N),s_4)-G((2N,2N),s_4,(1,1),s_1)-G((2N,2N),s_4,(2N,1),s_2) -G((2N,2N),s_4,(1,2N),s_3)-G((2N,2N),s_4,(2N,2N),s_4)\Bigr). \tag*{(19)}

We now employ the discrete Fourier transform given in Eq. (10) to evaluate Eqs. (18) and (19) explicitly. Because of the aforementioned circuit symmetries, the potential difference between the nodes at two opposite corners when the nodes are connected by a current source is V((N+1,N+1),s1)−V((2N,2N),s4)=2V((N+1,N+1),s1)=−2V((2N,2N),s4)V((N+1,N+1),s_1)-V((2N,2N),s_4)=2V((N+1,N+1),s_1)=-2V((2N,2N),s_4). By substituting Eq. (10) into Eq. (18) [Eq. (19)], the voltage at the lower left (upper right) corner node can be obtained. The two-point impedance between opposite corner nodes is given by Z2DSSH(N)=2V((N+1,N+1),s1)Z_{2D}^{\mathrm{SSH}}(N)=2V((N+1,N+1),s_1) or Z2DSSH(N)=−2V((2N,2N),s4)Z_{2D}^{\mathrm{SSH}}(N)=-2V((2N,2N),s_4), for which we provide the full impedance expression in Appendix A. Although the analytical expression looks complicated, it demonstrates the utility of the method of images technique. The resulting voltage distribution over the superlattice reflects the inversion and translation symmetries of the circuit. We now present an example of the spatial voltage distribution in the superlattice containing the physical 2D SSH circuit and its image copies.

Spatial voltage distribution of the 2D SSH circuit

To show how the symmetric current injection and extraction over a periodic superlattice gives rise to equal potentials between the boundary nodes of the image and physical circuits, we present the spatial voltage distribution for a periodic 2D SSH circuit. The spatial voltage distribution can be calculated using the numerical Laplacian formalism [Eq. (1)]

V=(J2D-SSHperiodic)−1I2D-SSH,(20)V=\left(J_{2D\text{-SSH}}^{\mathrm{periodic}}\right)^{-1}I_{2D\text{-SSH}}, \tag*{(20)}

where the voltage matrix (VV) over the periodic superlattice is obtained by performing the matrix multiplication of the inverse Laplacian matrix (J2D-SSHperiodicJ_{2D\text{-SSH}}^{\mathrm{periodic}}) and the current matrix (I2D-SSHI_{2D\text{-SSH}}). The current matrix corresponding to the 2D SSH circuit shown in Figure 1(d) is written as

I2D-SSH=I(−10⋯0000⋯0−100⋯0000⋯00⋮⋮⋱⋮⋮⋮⋮⋱⋮⋮00⋯0000⋯0000⋯0000⋯0000⋯0110⋯0000⋯0010⋯0000⋯0000⋯00⋮⋮⋱⋮⋮⋮⋱⋱⋮⋮00⋯0000⋯00−10⋯0000⋯0−1)2N×2N.(21)I_{2D\text{-SSH}}=I \begin{pmatrix} -1&0&\cdots&0&0&0&0&\cdots&0&-1\\ 0&0&\cdots&0&0&0&0&\cdots&0&0\\ \vdots&\vdots&\ddots&\vdots&\vdots&\vdots&\vdots&\ddots&\vdots&\vdots\\ 0&0&\cdots&0&0&0&0&\cdots&0&0\\ 0&0&\cdots&0&0&0&0&\cdots&0&0\\ 0&0&\cdots&0&1&1&0&\cdots&0&0\\ 0&0&\cdots&0&0&1&0&\cdots&0&0\\ 0&0&\cdots&0&0&0&0&\cdots&0&0\\ \vdots&\vdots&\ddots&\vdots&\vdots&\vdots&\ddots&\ddots&\vdots&\vdots\\ 0&0&\cdots&0&0&0&0&\cdots&0&0\\ -1&0&\cdots&0&0&0&0&\cdots&0&-1 \end{pmatrix}_{2N\times2N}. \tag*{(21)}

The matrix elements framed by the red square in the upper right block corresponds to the current matrix of the physical circuit. Because we consider an infinite lattice tiling with a superlattice with a total period of 2N2N unit cells along each direction, we employ the periodic circuit Laplacian (J2D-SSHperiodicJ_{2D\text{-SSH}}^{\mathrm{periodic}}). In Figure 2(a), we display an example of the periodic circuit Laplacian of the 2D SSH circuit for N=2N=2. Therefore, the matrix multiplication of the periodic Laplacian and the current matrix yields the spatial voltage distribution of the 2D SSH circuit. An example for the voltage distribution when N=10N=10 is given in Figure 2(b) where the voltage matrix is presented as a density plot. As can be seen from Figure 2(b), the boundary nodes of the physical circuit have the same voltages as those of the boundary nodes of the image circuits. Due to the equal voltage potentials between the boundary nodes, the current injected at node N+1N+1 cannot flow into the image circuits and is instead contained within the physical circuit. Therefore, the symmetrical current engineering as an analogy of the method of images leads to a perfectly symmetrical voltage distribution that therefore satisfies the boundary conditions. This makes it possible to obtain analytical expressions for finite-size circuits by applying the method of images to an infinite periodic lattice.

Simplified analytic impedance formulas for RLC circuits with a single node per unit cell

Here, we derive a generalized analytical expression for the impedance between two nodes at the opposite edges in 1D and at opposite corners in higher dimensions in homogeneous and heterogeneous finite circuits that have a single node per unit cell. The nodes are connected by components with an admittance of zαz_\alpha along the α\alphath direction where α∈(1,2,⋯ ,D)\alpha\in(1,2,\cdots,D) where DD is the dimension of the circuit. We define a homogeneous circuit as one in which the zαz_\alpha’s along all the directions have the same phase, i.e., they are either all capacitors or all inductors, and a heterogeneous circuit as one in which the zαz_\alpha’s along the different directions have different phases. To derive a generic expression, let us first consider a 2D infinite lattice made of image copies of a physical finite circuit [refer to Figure 3(b)] where the current is injected at nodes rin\mathbf{r}_{\mathrm{in}} and extracted from nodes rout\mathbf{r}_{\mathrm{out}}. Using the definition of the Green’s function GJ=−δGJ=-\delta [Eq. (1)] and the translation invariance of the circuit [which implies that G(r,r′)=G(r−r′)G(\mathbf{r},\mathbf{r}')=G(\mathbf{r}-\mathbf{r}')], the voltage at any lattice point r\mathbf{r} is found to be

Circuits lattices of different dimensionalities and their method of images implementations

Figure 3. Circuits lattices of different dimensionalities and their method of images implementations. The impedance is measured (a) edge-to-edge in the 1D circuit and (b)-(d) corner to corner in the 2D and higher-dimensional circuits. Red and gray blocks represent the physical and image circuits constructing a periodic infinite lattice. Each superlattice consisting of a physical circuit and image circuits is marked by the magenta dashed lines in each illustration. The cyan and orange dots indicate the spatial positions of the nodes at which the current is injected and extracted, respectively. We label the corner nodes in the 2D square and 3D cube circuits as NaN_{\mathrm{a}} such that NaN_{\mathrm{a}} is the diagonally opposite corner node to node 1 in every circuit. (b) The bigger red dots on the red plate denote the corner nodes NcN_{\mathrm{c}} and NbN_{\mathrm{b}} along the xx and yy directions, respectively. (c) The lower (upper) gray circle focuses on the node clusters where the current is injected (extracted). The red circles in the gray circles indicate the nodes belonging to the physical circuit i.e., nodes 1 and NaN_{\mathrm{a}}. (d) An illustrative representation of the 4D hypercube circuit. The input and output currents are not drawn to avoid excessive clutter. These circuits are homogeneous when the links are all resistive with admittances of 1/Ri1/R_i, all capacitive with admittances of zi=iωCiz_i=i\omega C_i, or all inductive with admittances zi=1/(iωLi)z_i=1/(i\omega L_i) [here i=(1,2,…,D)i=(1,2,\ldots,D)], but are heterogeneous when at least two distinct admittances with opposite phases such as z1=iωCz_1=i\omega C and z2=1/(iωL)z_2=1/(i\omega L) are present.

V(r)=I(∑r′∈rinG(r−r′)−∑r′∈routG(r−r′))(22)V(\mathbf{r})=I\left(\sum_{\mathbf{r}'\in\mathbf{r}_{\mathrm{in}}}G(\mathbf{r}-\mathbf{r}')-\sum_{\mathbf{r}'\in\mathbf{r}_{\mathrm{out}}}G(\mathbf{r}-\mathbf{r}')\right) \tag*{(22)}

where r=n1a1+n2a2\mathbf{r}=n_1\mathbf{a}_1+n_2\mathbf{a}_2 with integers (n1,n2)∈{−(N−1),⋯ ,N}(n_1,n_2)\in\{-(N-1),\cdots,N\}, and a1\mathbf{a}_1 and a2\mathbf{a}_2 are unit vectors corresponding to the xx and yy directions, respectively. Note that we define the limits of the superlattice as {−(N−1),⋯ ,N}\{-(N-1),\cdots,N\} by choice unlike the limits of the superlattice that we define for the 1D and 2D SSH circuits. The period of the superlattice remains unchanged at 2N2N. We now perform the discrete Fourier transformation [i.e., G(r−r′)=1(2N)D∑kG(k)eik⋅(r−r′)G(\mathbf{r}-\mathbf{r}')=\frac{1}{(2N)^D}\sum_{\mathbf{k}}G(\mathbf{k})e^{i\mathbf{k}\cdot(\mathbf{r}-\mathbf{r}')}] and recall L(k)=L−1(k)\mathcal{L}(\mathbf{k})=\mathcal{L}^{-1}(\mathbf{k}) to determine the spatial Green’s function G(r−r′)G(\mathbf{r}-\mathbf{r}') in Eq. (22) as

G(\mathbf{r}-\mathbf{r}')=\frac{1}{(2N)^2}\sum_{\mathbf{k}}\frac{e^{i\mathbf{k}\cdot(\mathbf{r}-\mathbf{r}')}{\mathcal{L}(\mathbf{k})},} \tag*{(23)}

where k=(k1a1+k2a2)\mathbf{k}=(k_1\mathbf{a}_1+k_2\mathbf{a}_2) is the momentum space index where ki=niπ/Nk_i=n_i\pi/N where i=(1,2)i=(1,2) and ni∈{−(N−1),⋯ ,N}n_i\in\{-(N-1),\cdots,N\}. Notice that, because we derive an analytical expression for the circuit that has a trivial unit cell [i.e., the circuit is made of a single-type node], the momentum-space Laplacian can take the role of momentum-space Green’s function, since G(k)G(\mathbf{k}) is no longer a matrix but is instead just the reciprocal of L(k)\mathcal{L}(\mathbf{k}). To achieve a symmetric voltage distribution over the superlattice, the current is injected and extracted at the nodes as depicted in Figure 3(b):

rin∈{(0,0),(1,0),(0,1),(1,1)}\mathbf{r}_{\mathrm{in}}\in\{(0,0),(1,0),(0,1),(1,1)\}
rout∈{(N,N),(N+1,N),(N,N+1),(N+1,N+1)},(24)\mathbf{r}_{\mathrm{out}}\in\{(N,N),(N+1,N),(N,N+1),(N+1,N+1)\}, \tag*{(24)}

where (n1,n2)(n_1,n_2) is a short-hand notation for r=(n1a1+n2a1)\mathbf{r}=(n_1\mathbf{a}_1+n_2\mathbf{a}_1). From Eqs. (24) and (22), the node voltage V(r)V(\mathbf{r}) can be found through

V(r)=I(G(r)+G(r+a1)+G(r+a2)+G(r+u)−G(r+Nu)−G(r+Nu+a1)−G(r+Nu+a2)−G(r+Nu+u)),(25)V(\mathbf{r})=I\big(G(\mathbf{r})+G(\mathbf{r}+\mathbf{a}_1)+G(\mathbf{r}+\mathbf{a}_2)+G(\mathbf{r}+\mathbf{u}) -G(\mathbf{r}+N\mathbf{u})-G(\mathbf{r}+N\mathbf{u}+\mathbf{a}_1) -G(\mathbf{r}+N\mathbf{u}+\mathbf{a}_2)-G(\mathbf{r}+N\mathbf{u}+\mathbf{u})\big), \tag*{(25)}

where u=a1+a2\mathbf{u}=\mathbf{a}_1+\mathbf{a}_2 denotes the unit vector for 2D lattices. To proceed, we insert the Green’s function defined in Eq. (23) into Eq. (25) to obtain the voltages at crosswise nodes 1 and NαN_\alpha where 1=1a1+1a2\mathbf{1}=\mathbf{1}\mathbf{a}_1+\mathbf{1}\mathbf{a}_2 and Nα=Na1+Na2\mathbf{N}_\alpha=N\mathbf{a}_1+N\mathbf{a}_2. As mentioned, the translational symmetry in the circuit implies that V(r)=−V(−r)V(\mathbf{r})=-V(-\mathbf{r}). Thus, the voltage difference between nodes 1 and NαN_\alpha is V(1)−V(Nα)=2V(1)=−2V(Nα)V(1)-V(N_\alpha)=2V(1)=-2V(N_\alpha). Therefore, we find the voltage at the corner nodes as

V(1)=−V(Nα)=I4N2∑n1∑n21L(k)×(1−eiπ(n1+n2))(1+eiπn1/N+eiπn2/N+eiπ(n1+n2)/N).(26)V(1)=-V(N_\alpha)=\frac{I}{4N^2}\sum_{n_1}\sum_{n_2}\frac{1}{\mathcal{L}(\mathbf{k})}\times \left(1-e^{i\pi(n_1+n_2)}\right)\left(1+e^{i\pi n_1/N}+e^{i\pi n_2/N}+e^{i\pi(n_1+n_2)/N}\right). \tag*{(26)}

Notice that the term (1−eiπ(n1+n2))\left(1-e^{i\pi(n_1+n_2)}\right) in the numerator of Eq. (26) is zero when (n1+n2)(n_1+n_2) is even and is 2 when (n1+n2)(n_1+n_2) is odd. Therefore, this term can be replaced by 2 provided that the summation is restricted to odd (n1+(n_1+ n2)n_2). After simplifying Eq. (26), the impedance between the corner nodes 1 and NaN_{\mathrm{a}} in a 2D square circuit, as a function of circuit size NN, can be written as

Z2D1,Na(N)=2N2∑k∗cos⁡(k1/2)cos⁡(k2/2)cos⁡((k1+k2)/2)L(k),(27)Z^{1,N_{\mathrm{a}}}_{2\mathrm{D}}(N)=\frac{2}{N^2}\sum_{\mathbf{k}}^{*}\frac{\cos(k_1/2)\cos(k_2/2)\cos((k_1+k_2)/2)}{\mathcal{L}(\mathbf{k})}, \tag*{(27)}

where the asterisk on the sum operator indicates that k=∑ikiai\mathbf{k}=\sum_i k_i\mathbf{a}_i where ki=niπ/Nk_i=n_i\pi/N and i=(1,2)i=(1,2) and a restricted summation over odd (n1+n2)(n_1+n_2). L(k)\mathcal{L}(\mathbf{k}) represents the corresponding 2D circuit Laplacian. One can then calculate the two-point impedance for both 2D homogeneous and heterogeneous circuits by simply assigning the corresponding circuit Laplacian to Eq. (27). This equation is also valid for the impedance between any pair of corner nodes as long as the summation is taken over n2∈n_2\in odd for Z2D1,NbZ^{1,N_{\mathrm{b}}}_{2\mathrm{D}} and n1∈n_1\in odd for Z2D1,NcZ^{1,N_{\mathrm{c}}}_{2\mathrm{D}}. This is because in the derivation of the expressions for Z2D1,NbZ^{1,N_{\mathrm{b}}}_{2\mathrm{D}} or Z2D1,NcZ^{1,N_{\mathrm{c}}}_{2\mathrm{D}}, one arrives at Eq. (26) with the factor

(1−ein2π)(1-e^{in_2\pi}) in the numerator for Z2D1,NbZ^{1,N_{\mathrm{b}}}_{2\mathrm{D}} and (1−ein1π)(1-e^{in_1\pi}) for Z2D1,NcZ^{1,N_{\mathrm{c}}}_{2\mathrm{D}} because, for example, the current distribution when the current is injected and extracted at nodes 1 and NcN_{\mathrm{c}} is written as rin∈{(0,0),(1,0),(0,1),(1,1)}\mathbf{r}_{\mathrm{in}}\in\{(0,0),(1,0),(0,1),(1,1)\} and rout∈{(N,0),(N+1,0),(N,1),(N+1,1)}\mathbf{r}_{\mathrm{out}}\in\{(N,0),(N+1,0),(N,1),(N+1,1)\}, respectively. (Here, Nb=Na1N_{\mathrm{b}}=N_{\mathrm{a}1} and Nc=Na2N_{\mathrm{c}}=N_{\mathrm{a}2} are the vertical and horizontal opposite corner nodes to the lower left node 1=1a1+1a21=1\mathbf{a}_1+1\mathbf{a}_2, respectively [see Figure 3(b)].) Therefore, there are contributions to the impedance only when n1∈n_1\in odd for Z2D1,NcZ^{1,N_{\mathrm{c}}}_{2\mathrm{D}} or n2∈n_2\in odd for Z2D1,NbZ^{1,N_{\mathrm{b}}}_{2\mathrm{D}}. From here, we can deduce that the Fourier component(s) of the principal direction(s) corresponding to the corner node only contribute to the impedance when its (their) summation is odd.

Inspired by the derivation for the impedance formula for the 2D square circuit (Eq. (27)) and taking into account the summation rule for the impedance between different corner nodes, we can obtain a general analytical expression for the corner-to-corner impedance of both DDdimensional homogeneous and heterogeneous circuits as

Z(N)=2ND∑k∗(∏i=1Dcos⁡(ki/2))×cos⁡(∑i=1Dki/2)(∑i=1Dλi(1−cos⁡(ki)))+zgnd/2(28)Z(N)=\frac{2}{N^{D}}\sum_{\mathbf{k}}^{*}\frac{\left(\prod_{i=1}^{D}\cos(k_i/2)\right)\times\cos\left(\sum_{i=1}^{D}k_i/2\right)}{\left(\sum_{i=1}^{D}\lambda_i(1-\cos(k_i))\right)+z_{\mathrm{gnd}}/2} \tag*{(28)}

where λi\lambda_i is the admittance of the coupling along each direction, DD represents the dimension of the circuit, and k=∑i=1Dkiai\mathbf{k}=\sum_{i=1}^{D}k_i\mathbf{a}_i where ki=niπNk_i=\frac{n_i\pi}{N} where ni∈{1,2,…,2N}n_i\in\{1,2,\ldots,2N\} and i=(1,2,…,D)i=(1,2,\ldots,D). The asterisk sign (*) on the summation operator implies that the impedance computation must be performed considering the summation rule. For example, since the diagonally opposite corner nodes can only be defined by considering all the spatial indices nin_i's, one must take the summation over (n1+n2+⋯+nD)∈odd(n_1+n_2+\cdots+n_D)\in\mathrm{odd}. To uniformly attach a grounding component of a single type to every node in the circuit, the admittance zgndz_{\mathrm{gnd}} can be assigned a non-zero admittance if desired, but it will be set to 0 if not. Note that because of the translation symmetry of the periodic infinite lattice, for simplicity, we can relabel the superlattice boundaries as ni∈{1,2,…,2N}n_i\in\{1,2,\ldots,2N\} instead of {−(N−1),…,N}\{-(N-1),\ldots,N\}. The corner-to-corner impedance between any pair of corners in a homogeneous or heterogeneous circuit can thus be calculated using Eq. (28) by simply setting the λi\lambda_i's in the denominator to the admittance values of the coupling along each principal direction. We will now apply this generalized expression to homogeneous and heterogeneous circuits in the following sections. We only consider passive circuit elements such as capacitors, inductors, and resistors, which can only store or dissipate energy pumped into the circuit [113].

Impedance results for homogeneous RLC circuits with trivial unit cells in various dimensions

In general, the corner-to-corner impedance of a homogeneous circuit constructed from passive components can be expected to increase uniformly with the circuit size because the components operate at the same phase (i.e., their admittances have the same sign). Since passive components cannot pump energy into the circuit, we intuitively expect the circuit to behave like a waveguide as the circuit size increases. To illustrate this, we consider a 1D circuit constructed from a single type of circuit element with admittance z1z_1 and calculate the impedance between the two opposite edge nodes as new unit cells are added. Employing Eq. (28) with D=1D=1, λ1=z1\lambda_1=z_1, zgnd=0z_{\mathrm{gnd}}=0, and 2cos⁡2(k1/2)=1+cos⁡(k1)2\cos^2(k_1/2)=1+\cos(k_1), the edge-to-edge impedance in 1D homogeneous circuits is obtained as

Z1Dhom(N)=1N∑k1∗1+cos⁡k1z1(1−cos⁡k1)(29)Z_{1D}^{\mathrm{hom}}(N)=\frac{1}{N}\sum_{k_1}^{*}\frac{1+\cos k_1}{z_1(1-\cos k_1)} \tag*{(29)}

where the asterisk means that k1=n1π/Nk_1=n_1\pi/N and n1∈{1,2,…,2N}n_1\in\{1,2,\ldots,2N\} and n1∈oddn_1\in\mathrm{odd}. Here, the impedance takes real values if the nodes are connected by resistors and takes imaginary values if z1z_1 corresponds to either a capacitor with an admittance of iωCi\omega C or an inductor with an admittance of 1/(iωL)1/(i\omega L). Regardless of the component represented by z1z_1, the impedance increases linearly with the circuit size, as shown in Figure 4. Note that Eqs. (14) and (15) provide the same edge-to-edge impedance as Eq. (29) when only one type of circuit element is present in the circuit and when NN is increased in steps of 2. This is because while a unit cell consists of two nodes for Eqs. (14) and (15), a unit cell comprises only a single node for Eq. (29).

Figure 4. Impedance across two opposite (edge) corner nodes of homogeneous (1D), 2D, 3D, 4D, and 5D circuits. All the circuits are constructed from only a single type of capacitor CC and we set C=1 μFC=1\,\mu\mathrm{F} for all impedance computations. (The edge-to-edge impedance in the 1D chain circuit is normalized by Z/10Z/10 for illustration.) While the impedance between two edge nodes in a 1D chain circuit scales linearly with the size, the impedance between two opposite corner nodes scales logarithmically in 2D finite circuits and rapidly approaches a finite saturation value in three dimensions or higher, as further detailed in Appendix C.

We now extend the 1D chain circuit to a 2D square circuit in which the nodes are connected by components with admittance z1z_1 along the horizontal direction and by components with admittance z2z_2 along the vertical direction. The corresponding circuit Laplacian becomes L2D(k1,k2)=z1(1−cos⁡k1)+z2(1−cos⁡k2)\mathcal{L}_{2D}(k_1,k_2)=z_1(1-\cos k_1)+z_2(1-\cos k_2). The two-point impedance for 2D circuits can be obtained by recalling Eq. (28) with k1=n1π/Nk_1=n_1\pi/N and k2=n2π/Nk_2=n_2\pi/N and z1z_1 and z2z_2 set to iωCi\omega C for capacitors, 1/(iωL)1/(i\omega L) for inductors, and 1/R1/R for resistors. Note that z1z_1 and z2z_2 must have the same phase (i.e., both are capacitors, inductors, or resistors; or capacitors + resistors; or inductors + resistors) in order to preserve the homogeneous structure of the circuit. We can intuitively expect the impedance of the 2D homogeneous circuits to scale logarithmically with the circuit size (Figure 4) because every addition of a new layer of unit cells results in a uniform increment in the overall impedance between two corner nodes with Z=∫NρdN′N′∼zlog⁡NZ = \int^{N} \rho\frac{\mathrm{d}N'}{N'} \sim z \log N. Aside from the corner-to-corner impedance, the impedance between two first nearest-neighbor nodes (NNs) along the horizontal or vertical directions also scales uniformly with the size. Figure 5 shows that the impedance between NNs is highest at the corner nodes and lower approaching the bulk nodes. This is because the current is pushed to the boundaries and accumulates there as it is reflected by the boundaries in a finite network. This trend also applies to higher dimensions such as 3D, 4D, 5D and beyond, for which interesting new phenomena can arise, and which can be feasibly implemented via circuits [7, 10, 11, 13, 15, 114].

Nearest-neighbor impedance distribution in 1D, 2D, and 3D homogeneous bounded circuit arrays

Figure 5. Nearest-neighbor impedance distribution in 1D, 2D, and 3D homogeneous bounded circuit arrays, with higher impedances near the boundaries. The impedance is calculated between a node and its horizontal nearest neighbor (NN). Every cell on the density plot is colored according to the overall impedance between that node and its NN. The total impedance at a node decreases approaching the center, and increases approaching the boundaries. All the circuits are made of only a single type of capacitor with capacitance CC which is set as C=1 μFC = 1\ \mu\mathrm{F}.

Similar to the procedure with which we constructed the 2D homogeneous circuit, the 2D square circuit can be extended to a 3D cube circuit by simply linking every node with an additional component z3z_3 along the third direction, for which the Fourier component is written as k3k_3 where k3=n3π/Nk_3 = n_3\pi/N. Figure 4 shows the two-point impedance behavior for homogeneous circuits of different dimensions as the circuit size increases. What is most remarkable is the qualitatively different behavior in higher dimensionalities [115]: for homogeneous circuits is, while 1D and 2D circuits exhibit unique scaling behaviors of linear and logarithmic impedance alternations, respectively, the circuits with the dimensionality of three and above tend to saturate at some finite constant values. For instance, the two-point impedance in the 3D cube circuit starts saturating after N∼25N \sim25, as can be seen in Figure 4.

starts saturating after N∼25N \sim25, as can be seen in Figure 4. In general, for D≥3D \ge3, the impedance saturates to a finite value more quickly as the dimensionality of the network increases. The overall scaling in homogeneous circuits for D≥3D \ge3 can be characterized by

Z(N)∼ZD−dimsat−(ZD−dimsat)DND−1,(30)Z(N) \sim Z_{\mathrm{D-dim}}^{\mathrm{sat}}-\frac{\left(Z_{\mathrm{D-dim}}^{\mathrm{sat}}\right)^D}{N^{D-1}}, \tag*{(30)}

where ZD−dimsatZ_{\mathrm{D-dim}}^{\mathrm{sat}} is the saturation value, DD is the dimension of the circuit, and NN is the circuit size in terms of unit cells. To find the ZD−dimsatZ_{\mathrm{D-dim}}^{\mathrm{sat}}s, we take the limit of Eq. (28) as N→∞N \to\infty. We present the full analytical expression for the saturation values in Appendix C. The saturation values obtained from Eq. (47) are: Z3-dimsat=1.44015 ΩZ_{3\text{-dim}}^{\mathrm{sat}} = 1.44015\ \Omega, Z4-dimsat=0.774964 ΩZ_{4\text{-dim}}^{\mathrm{sat}} = 0.774964\ \Omega, and Z5-dimsat=0.542093 ΩZ_{5\text{-dim}}^{\mathrm{sat}} = 0.542093\ \Omega, which confirm the impedance trends in the circuits for D≥3D \ge3 in Figure 4. For D≥3D \ge3, one can analytically show that in the continuum limit, the saturation impedances can be reduced to lower-dimensional integrals, thereby expressing lower-dimensional slices of the circuit in terms of lumped effective resistances.

Impedance results for heterogeneous RLC circuits with nontrivial unit cells

2D circuits

We now turn to heterogeneous circuits, where the corner-to-corner impedance exhibits a peculiar scaling behavior that differs significantly from that of homogeneous circuits. Heterogeneous circuits refer to circuits that have at least two types of components for which the admittances have different complex phases, such as capacitors and inductors. To illustrate the scaling behavior, we first consider a homogeneous 1D chain circuit consisting of N−1N-1 capacitors with the admittance of z1=iωCz_1 = i\omega C along the horizontal direction xx. We then connect N−1N-1 inductors with the admittance of z2=1/(iωL)z_2 = 1/(i\omega L) along the vertical direction yy to each node in the 1D chain circuit. This results in a two-dimensional LCLC circuit with N×NN \times N nodes. To calculate the impedance across the diagonal corner nodes, we recall Eq. (28) and assign the admittances z1z_1 and z2z_2 to the λi\lambda_i's such that z1z_1 and z2z_2 correspond to the principal directions. The corner-to-corner impedance of the 2D LCLC circuit is thus given by

Z2Dhet(N)=2N2∑k∗cos⁡(k1/2)cos⁡(k2/2)cos⁡((k1+k2)/2)iωC(1−cos⁡k1)+1iωL(1−cos⁡k2),(31)Z_{2D}^{\mathrm{het}}(N) = \frac{2}{N^2}\sum_{\mathbf{k}}^{*}\frac{\cos(k_1/2)\cos(k_2/2)\cos((k_1+k_2)/2)}{i\omega C(1-\cos k_1)+\frac{1}{i\omega L}(1-\cos k_2)}, \tag*{(31)}

where the asterisk on the summation operator indicates that k=k1a1+k2a2\mathbf{k} = k_1\mathbf{a}_1+k_2\mathbf{a}_2 where k1=n1π/Nk_1 = n_1\pi/N and k2=n2π/Nk_2 = n_2\pi/N and n1n_1 and n2n_2 are integers satisfying (n1,n2)∈{1,2,⋯ ,2N}(n_1,n_2) \in\{1,2,\cdots,2N\} where (n1+n2)(n_1+n_2) is odd. This equation can be used for any corner-to-corner impedance measurement in a 2D circuit by simply considering the following summation rules (∗*): For instance, for the impedance across two opposite corner nodes, one must take the summation over (n1+n2)∈odd(n_1+n_2)\in\text{odd}, while for the impedance between the two vertical corner nodes the summation should be taken over n2n_2 is odd. Apart from the driving AC frequency ω\omega and component parameters CC and LL, which are the independent parameters in the circuit, the circuit size NN becomes an additional independent parameter that affects the two-point impedance. It is well known that in LCLC resonator circuits, resonances occur at the resonant frequency ω=1/LC\omega=1/\sqrt{LC} regardless of the value of NN. Here, we uncover a curious phenomenon in which strong resonances occur only at particular circuit sizes. Figure 6 shows the occurrence of impedance jumps with the variation of the circuit size in several dimensions. The origin of these impedance resonances can be explained by considering the denominator of Eq. (31). An impedance resonance occurs when the values of LL, CC, and NN are such that there exist integers n1n_1 and n2n_2 at which the denominator of Eq. (31) becomes nearly zero when the terms proportional to iCiC and 1/iL1/iL cancel each other almost completely. Hence, strong resonances occur only at certain circuit sizes. Moreover, the impedance peaks stem not from the numerator but arise because of the almost-vanishing denominator at particular circuit sizes in LCLC circuits. Therefore, heterogeneous circuits with D>2D>2 can potentially exhibit more interesting circuit-size dependent impedance resonances since their Laplacians have more elements than that of lower-dimensional circuits leading to more possible combinations for the cancellation in the denominator.

Analytically and numerically calculated impedance between two opposite corner nodes in heterogeneous 2D, 3D, 4D, and 2D SSH circuits

Figure 6. Analytically and numerically calculated impedance between two opposite corner nodes in heterogeneous 2D (red), 3D (magenta), 4D (cyan), and 2D SSH (green) circuits. Sharp peaks emerge at certain circuit sizes in the heterogeneous circuits, unlike the logarithmic-like impedance scaling in homogeneous circuits. The component values are ω(C,L)=(1.7 Ω−1,2 Ω)\omega(C,L)=(1.7\ \Omega^{-1},2\ \Omega) for the 2D, ω(C1,C2,L)=(2.1 Ω−1,1 Ω−1,2.5 Ω)\omega(C_1,C_2,L)=(2.1\ \Omega^{-1},1\ \Omega^{-1},2.5\ \Omega) for the 3D, and ω(C1,C2,L1,L2)=(2.5 Ω−1,1.8 Ω−1,2.0 Ω,1.0 Ω)\omega(C_1,C_2,L_1,L_2)=(2.5\ \Omega^{-1},1.8\ \Omega^{-1},2.0\ \Omega,1.0\ \Omega) for the 4D LCLC circuits, and ω(C1,C2,L1,L2)=(2.2 Ω−1,1 Ω−1,2 Ω,1 Ω)\omega(C_1,C_2,L_1,L_2)=(2.2\ \Omega^{-1},1\ \Omega^{-1},2\ \Omega,1\ \Omega) for the 2D SSH circuit.

Heterogeneous circuits in 3D and higher dimensions

We next investigate higher-dimensional LCLC circuits starting with a 3D cube circuit array. We construct a cube circuit by extending the 2D LCLC square circuit along the zz direction using capacitors with the admittance z3z_3. The cube circuit illustrated in Figure 3(c) therefore comprises inductors with the admittance 1/(iωL)1/(i\omega L) along the yy direction and capacitors with the admittances of iωC1i\omega C_1 and iωC2i\omega C_2 linking nodes along the xx and zz directions, respectively. We calculate the impedance across the corner nodes by using the D=3D=3 analog of Eq. (28) and employing the Laplacian L3D(k1,k2,k3)=iωC1(1−cos⁡k1)+1/(iωL)(1−cos⁡k2)+iωC2(1−cos⁡k3)\mathcal{L}_{3D}(k_1,k_2,k_3)=i\omega C_1(1-\cos k_1)+1/(i\omega L)(1-\cos k_2)+i\omega C_2(1-\cos k_3) where ki=niπ/Nk_i=n_i\pi/N where ni∈{1,2,…,2N}n_i\in\{1,2,\ldots,2N\} and i=(1,2,3)i=(1,2,3). Due to the oddness of the discrete momentum (i.e., (1−(−1)k)(1-(-1)^k)), refer to Eq. (26), the summation rule is determined by the corner nodes between which the impedance is to be measured. For example, to find the impedance between the diagonally opposite corner nodes Z3D1,NaZ_{3D}^{1,N_a}, the summation is taken over (n1+n2+n3)∈odd(n_1+n_2+n_3)\in\text{odd}; while for Z3D1,NbZ_{3D}^{1,N_b} the rule is (n2+n3)∈odd(n_2+n_3)\in\text{odd}, for Z3D1,NcZ_{3D}^{1,N_c} it is (n1+n3)∈odd(n_1+n_3)\in\text{odd} and finally for Z3D1,NdZ_{3D}^{1,N_d} it is n1∈oddn_1\in\text{odd}. Therefore, only the coordinate indices of the corner node opposite to the node 1 contribute to the impedance calculation and only when their sum is odd.

Similar to the 2D case, the denominator in Eq. (28) with this Laplacian leads to an anomalously large impedance measurement when there exist integer values of (n1,n2,n3)∈{1,…,2N}(n_1,n_2,n_3)\in\{1,\ldots,2N\} where it becomes almost zero. The proposed cube circuit we have followed to build the cube circuit from the square circuit can be extended to construct hypercube circuits. For example, a 4D LCLC circuit can be constructed by introducing an additional direction ω\omega, which is perfectly feasible in 3D space due to the versatile connectivity of electrical circuits, without resorting to synthetic (non-genuine) lattice dimensions. The momentum-space Laplacian of the 4D circuit in which the nodes along the ω\omega direction are linked by inductors with the admittance of z4=1/(iωL2)z_4=1/(i\omega L_2) is

L4Dhet(k1,k2,k3,k4)=iω[C1(1−cos⁡k1)−1ω2L1(1−cos⁡k2)+C2(1−cos⁡k3)−1ω2L2(1−cos⁡k4)].(32)\mathcal{L}_{4D}^{\mathrm{het}}(k_1,k_2,k_3,k_4) = i\omega\left[C_1(1-\cos k_1)-\frac{1}{\omega^2L_1}(1-\cos k_2) +C_2(1-\cos k_3)-\frac{1}{\omega^2L_2}(1-\cos k_4)\right]. \tag*{(32)}

where ki=niπ/Nk_i=n_i\pi/N where ni∈{1,2,…,2N}n_i\in\{1,2,\ldots,2N\} and i=(1,2,3,4)i=(1,2,3,4). One can straightforwardly extend this circuit further to five dimensions (5D) or higher dimensions by invoking Eq. (28). Whereas the impedance between two opposite corner nodes in homogeneous circuits rapidly approaches a constant saturation value, it no longer exhibits a uniform trend in the heterogeneous cube and hyper-cube circuits. Instead, large impedance peaks are observed at certain values of NN. Figure 6 shows the variation of the corner-to-corner impedance of the 4D LCLC circuit with the circuit size NN. Interestingly, plotting the impedance peaks due to these size-dependent resonances against the driving frequency and circuit size gives a pattern that is reminiscent of fractals, as shown in Figure 7. These fractal-like patterns are not due to the dimensionality of the circuits but rather depend on the homogeneity of circuits and, indeed, arise in the circuit size-versus-parameter diagrams of heterogeneous circuits of any dimensions, as we shall discuss in the following section.

Impedance diagrams showing fractal-like structures for 2D SSH, 2D diagonal, 3D cube, and 2D Chern circuits

Figure 7. Fractal-like structures emerging in the plots of the corner-to-corner impedance in heterogeneous circuits versus the circuit size NN and driving frequency ω\omega [for (d), NN versus CgC_g]: (a) 2D SSH circuit, (b) 2D square LCLC circuit with diagonal cross links (obtained numerically), and (c) 3D cube circuit. The strong impedance resonances that appear as brighter branches are reminiscent of fractals. Despite the consideration of parasitic and intrinsic resistances that cause a variation of the component parameters, the structures always survive robustly. (d) Impedance diagrams for the non-trivial topolectrical 2D Chern kink circuit. The brightest impedance branches arise in the non-trivial parameter regime and correspond to the topological zero edge modes. The circuit parameters are (ωC1,ωCy,ωL,R)=(1.2 Ω−1,0.6 Ω−1,0.6 Ω,20 Ω)(\omega C_1,\omega C_y,\omega L,R)=(1.2\ \Omega^{-1},0.6\ \Omega^{-1},0.6\ \Omega,20\ \Omega).

Emergent fractal-like resonances in impedance scaling behavior

Most physical systems do not change fundamentally as their system sizes are varied, except for special critical systems exhibiting size-induced phase transitions [116, 117, 118, 119]. However, in our heterogeneous circuits, we observe sharp impedance deviations from the overall trend. Whereas special resonances can be expected to occur because of specific modulations of the circuit parameters, the observed anomalous impedance resonances that arise due to the system length are unexpected. These anomalous impedance resonances can therefore be treated as violations of the logarithmic impedance scaling between two opposite corner nodes. Equation (6) relates the two-point impedance to the eigensystem of the circuit Laplacian. The most significant difference between the eigenvalue spectrum of the Laplacians of homogeneous and heterogeneous circuits is that heterogeneous circuits typically have very small eigenvalues that can give rise to huge impedance readouts. Such a cancellation occurs in heterogeneous circuits between different components with admittance values of opposite signs, whereas no such cancellation occurs in homogeneous circuits as the admittance values have the same sign. Further, our analytical formulas show that at certain circuit sizes, there exist combinations of the integers n1,n2,…,nDn_1, n_2, \ldots, n_D denoting the Fourier components at which the denominator of the impedance expression almost cancels out and results in strong impedance resonances. Most interestingly, the length-dependent impedance resonances seem self-organized and exhibit a fractal-like pattern in the impedance-length plots. As can be seen in Figure 7, while the strongest impedance resonances (the brightest branches) occur along certain branches, the magnitude of the impedance varies between the branches. While one can obtain higher resolution diagrams (and spanning larger parameter spaces) with more computational resources, the overall trends do not change. Since the number of bulk states is defined by the total number of nodes in the circuit, the number of branches arising in these diagrams must be the same as the number of bulk states. Therefore, every row corresponding to a selected circuit size NN (i.e., the impedance peaks in a row corresponding to a fixed NN) is indeed the projection of the zero energy axis of the admittance spectrum. Consequently, because the number of admittance states increases with the circuit size NN, the number of states intersecting with the zero-axis increases, thus resulting in the formation of curly peak lines in the diagrams. In contrast, the topological impedance resonances, as seen in Figure 7(d), form straight lines because the number of topological modes is not defined by the circuit size but rather defined by the topological invariants. As a result, the emergent fractal-like diagrams can be treated as a complete picture of the resonant properties of a circuit regardless of whether it is topological or not. We next examine two exemplary circuits to further discuss how these fractal-like diagrams differ depending on the circuit arrays.

Two further examples for the emergent fractal-like resonances

Example: 2D square LC circuit with diagonal connections

The heterogeneous 2D LCLC circuit that we have studied so far (see Figure 3(b)) can be made more complicated, which in turn leads to richer fractal-like diagrams. This circuit has capacitors C1C_1 along with the vertical direction, inductors LL along the horizontal direction and a second capacitor C2C_2 connecting every node with its diagonal neighbors [see Figure 8]. Because of these cross connections, the circuit Laplacian has an additional term representing the nearest neighbor links with the admittance iωC2i\omega C_2. Therefore, the circuit Laplacian [Eq. (31)] becomes

2D LC circuit with diagonal cross connections

Figure 8. 2D LC circuit with diagonal cross connections with capacitors C2C_2. The usual 2D LCLC circuit in which the horizontal and vertical nodes are coupled with capacitors C1C_1 and inductors LL, respectively, can be recovered by simply setting C2=0C_2=0. To obtain the fractal-like diagram [Figure 7(b)], the impedance is numerically measured between two opposite corner nodes as the circuit size increases.

L2Dcross(k1,k2)=2iωC1(1−cos⁡k1)+2iωL(1−cos⁡k2)+4iωC2(1−cos⁡k1cos⁡k2)(33)\mathcal{L}^{\mathrm{cross}}_{2\mathrm{D}}(k_1,k_2)=2i\omega C_1(1-\cos k_1)+\frac{2}{i\omega L}(1-\cos k_2)+4i\omega C_2(1-\cos k_1\cos k_2) \tag*{(33)}

where ki=niπ/Nk_i=n_i\pi/N where ni∈{1,2,…,2N}n_i\in\{1,2,\ldots,2N\} and i=(1,2)i=(1,2). As we have discussed above, the size-dependent impedance resonances are induced by the attenuation of the Laplacian. Because G=J−1G=J^{-1} and V=GIV=GI, the attenuation of the Laplacian leads to voltage accumulation at the node where the current is injected. Thus, an enormous impedance read-out (Z∝VZ\propto V) occurs. Since the circuit Laplacian for the usual 2D LCLC circuit has a simpler form as given in the denominator of Eq. (31), it is expected that more complicated and fascinating branches may emerge in the fractal diagram of the 2D LCLC diagonally connected circuit. As can be seen in Figure 7(b), as the circuit size increases, many resonance branches that

emerge shape a pattern. Although the diagonal connections result in a well-organized fractal-like diagram, we emphasize that the diagonal links in this circuit lead to a current leakage from the physical circuit to the image circuits. In the usual 2D LCLC circuit [see Figure 2(b)], the symmetrical current injection and extraction ensures that the same potential exists at the neighbor boundary nodes of the physical and image circuits so that no current leakage occurs at the boundaries. However, in a circuit with diagonal links, the diagonal links connect boundary nodes with different electrical potentials, resulting in current flow through the cross links. Therefore, it is not possible to obtain a simple analytical expression via the method of images for the diagonally connected 2D LCLC circuit unless complicated current injection and extraction engineering is performed, which is not practical.

Example: 2D Chern kink topolectrical circuit

Topolectrical circuits can exhibit sophisticated physical phases such as valley-dependent edge states [120]. For example, the valley-dependent corner or edge states can be modulated by setting the onsite capacitors CgC_g connecting every node to the ground [refer to Figure 9]. The negative capacitance −Cg-C_g and resistance −R-R can be achieved by using negative impedance converters (NICs) and negative resistance converters (NRCs), respectively [18]. The circuit Laplacian at the resonant

The schematic of the 2D Chern kink circuit

Figure 9. The schematic of the 2D Chern kink circuit studied in Ref. [120]. The circuit has two distinct nodes in a unit cell labeled as node AA (red circle) and node BB (orange circle). The two nodes within a unit cell are connected with intra-cell inductors, whose admittance is equal to that of the capacitor −C1-C_1, i.e., L1=1/(ω2C1)=−C1L_1=1/(\omega^2 C_1)=-C_1. The horizontally adjacent unit cells are connected to each other via capacitors C1C_1. While the vertically adjacent AA nodes are coupled via positive resistors RR, two neighboring vertical BB nodes are connected via negative resistors −R-R realized by means of negative resistance converters (NRCs). The circuit has also capacitors CyC_y connecting A−BA-B and B−AB-A nodes along the yy direction.

frequency is written as

LTEChern(k1,k2)=(−C1+C1cos⁡k1+2Cycos⁡k2)σx+C1sin⁡k1σy+(Cg+2Rsin⁡k2)σz(34)\mathcal{L}_{\mathrm{TE}}^{\mathrm{Chern}}(k_1,k_2)=(-C_1+C_1\cos k_1+2C_y\cos k_2)\sigma_x +C_1\sin k_1\sigma_y+(C_g+2R\sin k_2)\sigma_z \tag*{(34)}

where the σα\sigma_\alpha's (α=x,y,z\alpha=x,y,z) represent the Pauli matrices and where ki=2niπ/Nk_i=2n_i\pi/N where i=(1,2)i=(1,2). As discussed in Appendix A, although a uniform grounding mechanism can be employed to isolate the topological states from the bulk states, a Chern topolectrical circuit can be designed so that it exhibits clear chiral propagating boundary states in the admittance spectrum without a uniform grounding mechanism. This is due to nontrivial center-of-mass pumping through a Laughlin-style pumping argument [121, 122, 123, 124, 125, 126, 127], which has been demonstrated in a variety of classical and quantum settings [17, 128, 129, 130, 131, 132]. Here, due to the nearest neighbor connections and the onsite capacitors represented by the factor σz\sigma_z in the Laplacian, the topological states are separated despite the nonuniform grounding. Therefore, this circuit is an ideal candidate to demonstrate impedance resonances arising in the fractal-like diagrams due to the topological states. For instance, the impedance resonances stemming from the edge states clearly appear in the fractal-like diagram as seen in Figure 7(d). The constant-magnitude impedance peaks can be treated as the evidence for zero-energy topological modes since the smallest eigenvalues always belong to these eigenstates. Aside from being the strongest resonances, the other remarkable property of these impedance resonances is that while the bulk states result in curved lines with the variation of the circuit size, the topological branches form distinctive straight lines. This is because even though the number of bulk states increases with the increase in circuit size, the number of topological states remains constant and the position of these states depends on the component parameters, which are fixed. Since we maintain the same component values except for the varying element (CgC_g), the topological impedance resonances appear at the same CgC_g value in the fractal-like diagram.

Conclusion

In conclusion, we have revealed the circuit size-dependent anomalous impedance resonances in topolectrical circuits as well as various dimensional finite LC circuits. We observed that the impedance scaling in heterogeneous circuits differs from the logarithmic-like scaling exhibited by homogeneous circuits as the circuit size NN increases. Conventionally, frequency-dependent elements such as capacitors and inductors are considered independent circuit parameters that can be set individually. However, we have demonstrated that the strong impedance resonances in our circuits are not due to variations in these parameters, but rather to specific circuit sizes. Therefore, the circuit size becomes an independent, albeit not widely known, parameter that affects the impedance behavior of the circuit. We invoked the method of images, inspired by free-space electrostatics, as a means to calculate the corner-to-corner impedance and provided a generic exact analytical expression homogeneous or heterogeneous circuits of any dimensions. This method naturally satisfies the open boundary conditions because it ensures that the potentials of the nodes at the boundaries of the physical and image circuits are equal [see Figure 2(b)]. The size-dependent impedance jumps result in fractal-like patterns in the circuit size-resonant frequency plots. The existence of these patterns is common to all heterogeneous circuits but the details within the patterns are unique to each dimensionality and circuit structure. Therefore, our work establishes a framework for further investigation of anomalous impedance behaviors in more complex circuits, as well as their experimental realizations [133].

Acknowledgments

This work is supported by the Ministry of Education (MOE) Tier-II Grant No. MOE-T2EP50121-0014 (NUS Grant No. A-8000086-01-00), and MOE Tier-I FRC grant (NUS Grant No. A-8000195-01-00). C.H.L. acknowledges support from Singapore Ministry of Education’s Tier I grant (NUS Grant No. A-8000022-00-00). H.S. would like to thank the Agency for Science, Technology and Research (A*STAR) for its support of our research through the SINGA fellowship program.

Explicit analytical expression for the impedance of the 2D SSH circuit under periodic boundary conditions

Here, we provide the full and explicit analytical expression for the corner-to-corner two-point impedance of the 2D PBC SSH circuit. The impedance can be obtained from the voltage difference between the lower left corner [r=(N+1,N+1),μ=s1][\mathbf{r}=(N+1,N+1),\mu=s_1] and upper right corner [r=(2N,2N),μ=s4][\mathbf{r}=(2N,2N),\mu=s_4] nodes given by Eqs. (18) and (19), respectively. There, to evaluate the voltages at the corner nodes, the Fourier transformed circuit Green’s

function given in Eq. (10) is substituted into Eqs. (18) and (19). Notice that when taking the inverse of the circuit Laplacian given in Eq. (16), any component that has zero eigenvalues should be omitted to avoid singularities. After performing the substitutions, we arrive at

Z2DSSH(N)=∑k1∑k2Numerator⁡(k1,k2)Denominator⁡(k1,k2)(35)Z^{\mathrm{SSH}}_{2\mathrm{D}}(N)=\sum_{k_1}\sum_{k_2}\frac{\operatorname{Numerator}(k_1,k_2)}{\operatorname{Denominator}(k_1,k_2)} \tag*{(35)}

where the numerator and denominator are explicitly given by

Numerator⁡=1+(−1)1+n1+n22iω3L12L22×[L1(4ω2L2(2C1+3C2)−4+ei(k1−k2)ω2C1L2−2ω4L22(C12+4C1C2+2C22)+e−ik1ω4C1C2L22−e−i(k1−k2)ω4C1C2L22+e−ik2(2−ω2C1L2)−eik1ω2C1L2(ω2L2(2C1+C2)−4)−ei(k1+k2)ω2C1L2(ω2L2(2C1+3C2)−3)+eik2(2+ω2L2(C1+4C2−2ω2L2(C12+2C1C2+2C22)))]−4eik1/2ω2L12(2ω2L2(C1+C2)−2−ω4C1C2L22(1−cos⁡(k1)))×((C1+C2)cos⁡(k1/2)−iC2sin⁡(k1/2))+L2(2ω2L2(2C2(1+eik2)+C1(1+eik1+eik2+ei(k1+k2)))+cos⁡(2k2)+2isin⁡(k2)(cos⁡(k2)−1)−1)],\begin{aligned} \operatorname{Numerator}={}&\frac{1+(-1)^{1+n_1+n_2}}{2i\omega^3L_1^2L_2^2}\times\Bigl[L_1\Bigl(4\omega^2L_2(2C_1+3C_2)-4+e^{i(k_1-k_2)}\omega^2C_1L_2-2\omega^4L_2^2(C_1^2+4C_1C_2+2C_2^2)\\ &\quad+e^{-ik_1}\omega^4C_1C_2L_2^2-e^{-i(k_1-k_2)}\omega^4C_1C_2L_2^2+e^{-ik_2}(2-\omega^2C_1L_2)\\ &\quad-e^{ik_1}\omega^2C_1L_2(\omega^2L_2(2C_1+C_2)-4)-e^{i(k_1+k_2)}\omega^2C_1L_2(\omega^2L_2(2C_1+3C_2)-3)\\ &\quad+e^{ik_2}\bigl(2+\omega^2L_2(C_1+4C_2-2\omega^2L_2(C_1^2+2C_1C_2+2C_2^2))\bigr)\Bigr]\\ &\quad-4e^{ik_1/2}\omega^2L_1^2(2\omega^2L_2(C_1+C_2)-2-\omega^4C_1C_2L_2^2(1-\cos(k_1)))\\ &\qquad\times\bigl((C_1+C_2)\cos(k_1/2)-iC_2\sin(k_1/2)\bigr)\\ &\quad+L_2\Bigl(2\omega^2L_2\bigl(2C_2(1+e^{ik_2})+C_1(1+e^{ik_1}+e^{ik_2}+e^{i(k_1+k_2)})\bigr)\\ &\qquad\quad+\cos(2k_2)+2i\sin(k_2)(\cos(k_2)-1)-1\Bigr)\Bigr], \end{aligned}
Denominator⁡=4N2ω4L12L22(−ω8C12C22L12L22(cos⁡(2k1)+3)+4ω4C1C2cos⁡(k1)(ω4C1C2L12L22−ω2L1L2(C1+C2)(L1+L2)+L1L2cos⁡(k2)+L12+L22+L1L2)+4cos⁡(k2)(ω4L1L2(C12+C1C2+C22)−ω2(C1+C2)(L1+L2)+1)+4ω6C1C2L1L2(C1+C2)(L1+L2)−4ω4(C1C2L12+L1L2(C12+3C1C2+C22)+C1C2L22)+4ω2(C1+C2)(L1+L2)−cos⁡(2k2)−3).(36)\begin{aligned} \operatorname{Denominator}={}&\frac{4N^2}{\omega^4L_1^2L_2^2}\Bigl(-\omega^8C_1^2C_2^2L_1^2L_2^2(\cos(2k_1)+3)\\ &\quad+4\omega^4C_1C_2\cos(k_1)\bigl(\omega^4C_1C_2L_1^2L_2^2-\omega^2L_1L_2(C_1+C_2)(L_1+L_2)+L_1L_2\cos(k_2)+L_1^2+L_2^2+L_1L_2\bigr)\\ &\quad+4\cos(k_2)\bigl(\omega^4L_1L_2(C_1^2+C_1C_2+C_2^2)-\omega^2(C_1+C_2)(L_1+L_2)+1\bigr)\\ &\quad+4\omega^6C_1C_2L_1L_2(C_1+C_2)(L_1+L_2)-4\omega^4(C_1C_2L_1^2+L_1L_2(C_1^2+3C_1C_2+C_2^2)+C_1C_2L_2^2)\\ &\quad+4\omega^2(C_1+C_2)(L_1+L_2)-\cos(2k_2)-3\Bigr). \tag*{(36)} \end{aligned}

where k1k_1 and k2k_2 are the discrete momenta, k1=n1π/Nk_1=n_1\pi/N and k2=n2π/Nk_2=n_2\pi/N where (n1,n2)∈(1,2,⋯ ,2N)(n_1,n_2)\in(1,2,\cdots,2N). Each (k1,k2)(k_1,k_2) contribution represents the impedance contribution from the length scale (2π/k1,2π/k2)(2\pi/k_1,2\pi/k_2) in units of the lattice spacing.

Aside from the usual impedance peaks stemming from the LCLC resonances, a topolectrical circuit exhibits an enormous impedance readout at the resonant frequency when the circuit is topologically non-trivial. It is well known that topological systems differ from the usual bulk systems owing to their special boundary modes. These boundary states are isolated from the bulk states and appear as mid-gap states in the admittance spectrum. Such mid-gap topological states lie on the zero-energy axis, and they are associated with very small eigenvalues. Therefore, any large impedance readout at the resonant frequency can be directly related to the topological mid-gap zero-modes when they exist [8] [19]. For example, the topological phase is defined by the ratio of two capacitors in the usual 1D SSH circuit in Ref. [4] and the circuit displays a non-trivial topological phase when C1/C2<1C_1/C_2<1 and a trivial phase when C1/C2>1C_1/C_2>1. However, even though the mid-gap states are protected by topological invariants such as a non-zero integer winding number, the unequal onsite energies lead to ill-defined invariants, hence, resulting in indistinguishable topological states [134, 135, 136]. In the TE context, the onsite energies are represented by the diagonal elements of the circuit Laplacian. Therefore, to unveil the topological boundary states, the circuit requires uniform grounding such that all the diagonal terms in the circuit Laplacian can vanish when the driving frequency is set to the resonant frequency. Otherwise, the boundary modes join the bulk modes due to the unequal potentials at different nodes and are no longer found as mid-gap states.

To uniformly ground every node in the circuit, an artificial treatment is required, which would not be possible for analytical methods. Throughout this study, since we consider an infinite periodic lattice tiled with the original physical circuits and apply the method of images to obtain the exact analytical expression for the physical circuit, it may not be possible to introduce uniform grounding into the analytical two-point impedance formula. This is because the open boundary conditions are fulfilled as a direct consequence of the method of images, which results in the Neumann boundary condition in the circuit when the circuit Laplacian has non-uniform diagonal elements. Therefore, although the non-uniform diagonal elements representing the grounding mechanism cause the disappearance of the mid-gap topological states in most cases, in some examples, the topological states can remain isolated from the bulk states and appear in fractal-like diagrams as in Figure 7d.

Application of the method of images to general geometries

The method of images can be applied to a general lattice model to derive an analytical expression. As discussed in the main text, the configuration of the current injection and extraction determines the spatial distribution of node voltages. To induce boundaries by utilizing the equipotential emerging on both sides of the symmetry axes, it is essential to inject and extract the current symmetrically. This results in a mirror-symmetric dispersion around the considered symmetry axes. (Note that such a symmetrical voltage distribution can only be achieved if inversion symmetry is present.) For instance, since the unit cell of the 2D SSH circuit (as well as the 2D square LCLC circuit) exhibits translational symmetry along the xx and yy symmetry axes, the spatial potential profiles of the physical and image circuits are mirror symmetric about the xx and yy axes when there is symmetric current injection and extraction [refer to Figure 2(b)]. On the other hand, the symmetry axes of a hexagonal lattice provide additional opportunities for creating unique boundary designs by strategically positioning current injection and extraction sites with respect to these axes. Consequently, we can achieve diverse boundary designs by configuring the current injection and extraction points relative to the symmetry axes in a honeycomb lattice.

To demonstrate how a symmetrical current injection and extraction configuration with respect to certain symmetry axes can be utilized to establish a boundary in general geometries, we examine a 2D honeycomb lattice composed of two sublattice nodes AA and BB. Our objective is to derive an analytical expression for a ribbon geometry incorporating a honeycomb lattice structure and zigzag edges. We begin by considering a geometrically rhombic honeycomb lattice with a zigzag edge design. Its momentum space circuit Laplacian reads as

L◊(k)=iωC(3−(1+e−ik1+e−ik2)−(1+eik1+eik2)3),(37)\mathcal{L}_{\lozenge}(\mathbf{k})=i\omega C\begin{pmatrix}3 & -(1+e^{-ik_1}+e^{-ik_2})\\ -(1+e^{ik_1}+e^{ik_2}) & 3\end{pmatrix}, \tag*{(37)}

where CC represents the coupling capacitance, ω\omega is the driving frequency, and k1k_1 and k2k_2 are the momentum indices. We proceed to tile the space with rhombus-shaped circuits and apply currents at specific nodes to achieve a symmetrical voltage distribution about one of the symmetry axes. Since we initially assume a periodic lattice along each direction, it is sufficient to achieve a symmetrical voltage distribution around a single axis in order to obtain the desired ribbon geometry. In Figure 10(b), the magenta dashed lines represent the symmetry axes under consideration. Our current injection and extraction configuration results in a mirror-symmetric potential distribution about the vertical symmetry axis. This implies that the voltages of the sublattice nodes within a unit cell along the vertical axis have equal magnitudes, ensuring no current flows between these nodes. Since our circuit remains periodic in both directions, the equal voltages on either side of the vertical axis satisfy the boundary condition required to achieve a ribbon geometry. We now move forward with deriving an analytical expression for the two-point impedance of the ribbon lattice. We recall Eq. (8) to determine the node voltages and express the spatial positions of the nodes where the current is injected and extracted, respectively, as

Figure 10. An exemplary implementation of the method of images to achieve finite circuits in a general geometry using a 2D honeycomb lattice. a) The rhombus-shaped finite physical circuit with N=6N = 6. A unit cell comprises two sublattice nodes AA and BB, each connected by coupling capacitance CC. b) The spatial voltage distribution response to the injected (yellow-outlined nodes) and extracted (green-outlined nodes) current configuration. The node colors represent the relative magnitude of the voltage at each node. A mirror-symmetrical potential distribution occurs about the vertical symmetry axis, ensuring the emergence of a boundary. c) The resultant distribution leads to a ribbon geometry with zigzag edges. The alternating node colors represent the voltage magnitude. d) The measured impedance between any two nodes outlined with magenta and cyan dashed circles is identical. The impedance increases logarithmically (inset) as the circuit size expands.

r~in∈{((1,1),1),((1,2N),1),((1,2N),2),((2N,1),1),((2N,1),2),((2N,2N),2)},(38)\tilde{\mathbf{r}}_{\mathrm{in}} \in\{((1,1),1),((1,2N),1),((1,2N),2),((2N,1),1),((2N,1),2),((2N,2N),2)\}, \tag*{(38)}
r~out∈{((N,N),2),((N+1,N),1),((N+1,N),2),((N,N+1),1),((N+1,N+1),1)}.(39)\tilde{\mathbf{r}}_{\mathrm{out}} \in\{((N,N),2),((N+1,N),1),((N+1,N),2),((N,N+1),1),((N+1,N+1),1)\}. \tag*{(39)}

where ((n1,n2),ν)((n_1,n_2),\nu) is shorthand for r=(n1a1+n2a2)\mathbf{r}=(n_1\mathbf{a}_1+n_2\mathbf{a}_2) where a1\mathbf{a}_1 and a2\mathbf{a}_2 are the unit vectors, n1n_1 and n2n_2 are the position indices and ν∈(1,2)\nu\in(1,2) represents the sublattice nodes A and B, respectively. We can calculate the voltage at the first node as

V(r=1,μ=1)=I(G((1,1),1,(1,1),1)+G((1,1),1,(1,2N),1)+G((1,1),1,(1,2N),2)+G((1,1),1,(2N,1),1)+G((1,1),1,(2N,1),2)+G((1,1),1,(2N,2N),2)\begin{aligned} V(\mathbf{r}=1,\mu=1) \\ &=I\bigl(G((1,1),1,(1,1),1)+G((1,1),1,(1,2N),1) \\ &+G((1,1),1,(1,2N),2)+G((1,1),1,(2N,1),1) \\ &+G((1,1),1,(2N,1),2)+G((1,1),1,(2N,2N),2) \end{aligned}
−G((1,1),1,(N,N),2)−G((1,1),1,(N+1,N),1)-G((1,1),1,(N,N),2)-G((1,1),1,(N+1,N),1)
−G((1,1),1,(N+1,N),2)−G((1,1),1,(N,N+1),1)-G((1,1),1,(N+1,N),2)-G((1,1),1,(N,N+1),1)
−G((1,1),1,(N,N+1),2)−G((1,1),1,(N+1,N+1),1)).(40)-G((1,1),1,(N,N+1),2)-G((1,1),1,(N+1,N+1),1)\bigr). \tag*{(40)}

By employing the momentum space Green’s function in Eq. (10), and L◊−1(k)=G(k)\mathcal{L}_{\lozenge}^{-1}(\mathbf{k})=G(\mathbf{k}), the voltage at node V(r=1,μ=1)V(\mathbf{r}=1,\mu=1) is obtained as

V(r=1,μ=1)=14iωCN2(3−cos⁡(k1)−cos⁡(k2)−cos⁡(k1−k2))(41)V(\mathbf{r}=1,\mu=1)=\frac{1}{4i\omega C N^2\left(3-\cos(k_1)-\cos(k_2)-\cos(k_1-k_2)\right)} \tag*{(41)}
×(5(1+cos⁡(k1)+cos⁡(k2))+2cos⁡(k1−k2)+cos⁡(k1+k2)−cos⁡(N−1N(n1+n2)π)(42)\times\left(5\left(1+\cos(k_1)+\cos(k_2)\right)+2\cos(k_1-k_2)+\cos(k_1+k_2)-\cos\left(\frac{N-1}{N}(n_1+n_2)\pi\right) \right. \tag*{(42)}
−2cos⁡(N−1N(n1−n2)π)−4cos⁡(N−1Nn1π−n2π)−cos⁡(N−1Nn1π+n2π)−cos⁡(N−1Nn2π+n1π)),(43)\left. -2\cos\left(\frac{N-1}{N}(n_1-n_2)\pi\right)-4\cos\left(\frac{N-1}{N}n_1\pi-n_2\pi\right)-\cos\left(\frac{N-1}{N}n_1\pi+n_2\pi\right)-\cos\left(\frac{N-1}{N}n_2\pi+n_1\pi\right) \right), \tag*{(43)}

where k1=n1π/Nk_1=n_1\pi/N and k2=n2π/Nk_2=n_2\pi/N. To simplify the above equation, we assume that n1∈(1,2,3,⋯ ,2N)n_1\in(1,2,3,\cdots,2N) and n2∈(1,3,5,7,⋯ ,2N−1)n_2\in(1,3,5,7,\cdots,2N-1). Owing to the translation symmetry in our circuit, the impedance between node 1 and its corresponding counterpart located at the opposite edge [i.e., (N,N)(N,N)] is given by Z□(N)=2V(1,1)=2V(N,N)Z_{\square}(N)=2V(1,1)=2V(N,N). The impedance between any pair of nodes denoted by magenta and cyan dashed circles in Figure 10(c) is identical, a consequence of the symmetrical voltage distribution. In Figure 10(d), we plot our numerical and analytical impedance calculation results for the ribbon circuit depicted in Figure 10(c). There is an exact correspondence between the two sets of results. Our results also show that the logarithmic dependence of the circuit impedance on the circuit size NN applies for two-dimensional circuits regardless of the lattice geometry [see Figure 10(d)].

Saturation in homogeneous circuits

To find the finite saturation values in homogeneous circuits when D>2D>2, let us first consider a homogeneous 3D cube circuit constructed by a typical resistor with the resistance RR in each principal direction. Now, it is essential to determine the electric potential at the opposite corner nodes whereby the corner-to-corner impedance as a function of the circuit size can be simply found by Z=2∣V∣/I=2GZ=2|V|/I=2G due to the translation symmetry implying that V(r)=−V(−r)V(\mathbf{r})=-V(-\mathbf{r}). To impose the boundary conditions, symmetrical current injection and extraction [for example, from Figure 3(c)] are performed such that the impedance between two opposite corner nodes in terms of the voltage distribution is written as

Z3D=2(G(0,0,0)+G(1,0,0)+G(0,1,0)+G(0,0,1)+G(1,1,0)+G(1,0,1)+G(0,1,1)+G(1,1,1)−G(N,N,N)−G(N+1,N,N)−G(N,N+1,N)−G(N,N,N+1)−G(N+1,N+1,N)−G(N+1,N,N+1)−G(N,N+1,N+1)−G(N+1,N+1,N+1))(44)Z_{\mathrm{3D}}=2\bigl(G(0,0,0)+G(1,0,0)+G(0,1,0)+G(0,0,1)+G(1,1,0)+G(1,0,1)+G(0,1,1)+G(1,1,1) -G(N,N,N)-G(N+1,N,N) -G(N,N+1,N)-G(N,N,N+1) -G(N+1,N+1,N)-G(N+1,N,N+1) -G(N,N+1,N+1)-G(N+1,N+1,N+1)\bigr) \tag*{(44)}

where G(n1,n2,n3)G(n_{1},n_{2},n_{3}) is the short notation of G(r)G(\mathbf{r}) [e.g., G(r)→G(0,0,0)G(\mathbf{r})\rightarrow G(0,0,0) or G(r+Nu+a1)→G(N+1,N,N)G(\mathbf{r}+N\mathbf{u}+\mathbf{a}_{1})\rightarrow G(N+1,N,N)], where r=∑i=1D=3niai\mathbf{r}=\sum_{i=1}^{D=3}n_{i}\mathbf{a}_{i} where nin_{i}s are the integers varying from 11 to 2N2N and where u=a1+a2+a3\mathbf{u}=\mathbf{a}_{1}+\mathbf{a}_{2}+\mathbf{a}_{3} being the unit vector. As accomplished in Sec. IV, the circuit Green’s function given by Eq. (23) is inserted into the above equation. Because the circuit nodes are connected by resistors with the admittance 1/R1/R, the corresponding circuit Laplacian is written as L3Dsat(k)=(2/R)(3−cos⁡k1−cos⁡k2−cos⁡k3)L^{\mathrm{sat}}_{\mathrm{3D}}(\mathbf{k})=(2/R)(3-\cos k_{1}-\cos k_{2}-\cos k_{3}) where k=∑i=1D=3kiai\mathbf{k}=\sum_{i=1}^{D=3}k_{i}\mathbf{a}_{i} and where ki=niπ/Nk_{i}=n_{i}\pi/N where ni∈{1,2,…,2N}n_{i}\in\{1,2,\ldots,2N\} and i=(1,2,3)i=(1,2,3). However, for the sake of simplicity, we henceforth set R=1ΩR=1\Omega in the derivation. By inserting the corresponding Laplacian substituted for G(k)−1G(\mathbf{k})^{-1}, we arrive

Z3D(N)=2(2N)3∑n1=12N∑n2=12N∑n3=12N(1−eiπ(n1+n2+n3))8cos⁡(n1π2N)cos⁡(n2π2N)cos⁡(n3π2N)cos⁡((n1+n2+n3)π2N)2(3−cos⁡(n1πN)−cos⁡(n2πN)−cos⁡(n3πN)).(45)Z_{\mathrm{3D}}(N)=\frac{2}{(2N)^{3}}\sum_{n_{1}=1}^{2N}\sum_{n_{2}=1}^{2N}\sum_{n_{3}=1}^{2N}\left(1-e^{i\pi(n_{1}+n_{2}+n_{3})}\right) \frac{8\cos\left(\frac{n_{1}\pi}{2N}\right)\cos\left(\frac{n_{2}\pi}{2N}\right)\cos\left(\frac{n_{3}\pi}{2N}\right)\cos\left(\frac{(n_{1}+n_{2}+n_{3})\pi}{2N}\right)} {2\left(3-\cos\left(\frac{n_{1}\pi}{N}\right)-\cos\left(\frac{n_{2}\pi}{N}\right)-\cos\left(\frac{n_{3}\pi}{N}\right)\right)}. \tag*{(45)}

Here, we obtain the analytical formula for the corner-to-corner impedance in the 3D homogeneous cube circuit. We can now generalize the corner-to-corner impedance in DD dimensions and write

ZD-dim(N)=1ND∑n1=12N⋯∑nD=12N(1−eiπ∑iDni)cos⁡(∑iDniπ2N)∏iDcos⁡(niπ2N)D−∑iDcos⁡(niπN).(46)Z_{\mathrm{D\text{-}dim}}(N)=\frac{1}{N^{D}}\sum_{n_{1}=1}^{2N}\cdots\sum_{n_{D}=1}^{2N}\left(1-e^{i\pi\sum_{i}^{D}n_{i}}\right) \frac{\cos\left(\sum_{i}^{D}\frac{n_{i}\pi}{2N}\right)\prod_{i}^{D}\cos\left(\frac{n_{i}\pi}{2N}\right)} {D-\sum_{i}^{D}\cos\left(\frac{n_{i}\pi}{N}\right)}. \tag*{(46)}

Here, DD refers to the circuit dimension, nin_{i}s are the Fourier components varying between 11 and 2N2N along each direction, and NN is the circuit size. Notice that because we have set R=1ΩR=1\Omega earlier, there is no term representing the unit admittance. However, one can multiply the denominator of Eq. (46) by a unit admittance if required because the admittance is a common factor in the Laplacian since the circuit nodes are linked by single-type components. The impedance between two opposite corner nodes calculated through the above expression saturates as NN approaches the continuum limit, i.e., ZD-dimsat≈lim⁡N→∞ZD-dim(N)Z_{\mathrm{D\text{-}dim}}^{\mathrm{sat}}\approx\lim_{N\rightarrow\infty}Z_{\mathrm{D\text{-}dim}}(N). The presence of saturation impedances in the large-NN limit in dimensions D≥3D\geq3 suggests a convergent integral expression for the corner-to-corner impedances in this regime. In the large-NN limit, the evenness and oddness of the indices should have negligible effects if the sum can be approximated by an integral, since that entails shifts of π/N\pi/N to the momenta. Noting that the factor 1−eiπ∑iDni1-e^{i\pi\sum_{i}^{D}n_{i}} effectively eliminates half of the terms, we have the approximation

D≥3D\geq3 suggests a convergent integral expression for the corner-to-corner impedances in this regime. In the large-NN limit, the evenness and oddness of the indices should have negligible effects if the sum can be approximated by an integral, since that entails shifts of π/N\pi/N to the momenta. Noting that the factor 1−eiπ∑iDni1-e^{i\pi\sum_{i}^{D}n_{i}} effectively eliminates half of the terms, we have the approximation

ZD-dim(N→∞)≈1πD∫[0,2π]Dcos⁡(∑iDki2)∏iDcos⁡ki2D−∑iDcos⁡ki dDk(47)Z_{\mathrm{D\text{-}dim}}(N\rightarrow\infty)\approx\frac{1}{\pi^{D}}\int_{[0,2\pi]^{D}}\frac{\cos\left(\sum_{i}^{D}\frac{k_{i}}{2}\right)\prod_{i}^{D}\cos\frac{k_{i}}{2}}{D-\sum_{i}^{D}\cos k_{i}}\,d^{D}k \tag*{(47)}

for D≥3D\geq3. In practice, a very small positive term may have to be added to the denominator to make the integral converge; physically, this corresponds to inevitable parasitic resistances. For instance, we obtain Z3-dim(N→∞)=1.44015 ΩZ_{\mathrm{3\text{-}dim}}(N\rightarrow\infty)=1.44015\ \Omega, Z4-dim(N→∞)=0.774964 ΩZ_{\mathrm{4\text{-}dim}}(N\rightarrow\infty)=0.774964\ \Omega and Z5-dim(N→∞)=0.542093 ΩZ_{\mathrm{5\text{-}dim}}(N\rightarrow\infty)=0.542093\ \Omega, which are very close to the values of Z3-dim(300)=1.43583 ΩZ_{\mathrm{3\text{-}dim}}(300)=1.43583\ \Omega, Z4-dim(50)=0.774739 ΩZ_{\mathrm{4\text{-}dim}}(50)=0.774739\ \Omega and Z5-dim(40)=0.542011 ΩZ_{\mathrm{5\text{-}dim}}(40)=0.542011\ \Omega computed from Eq. (46).

Dimensional reduction of impedance formula

Interestingly, the corner-to-corner impedance can be dimensionally reduced to a momentum integral in a lower number of dimensions, albeit with a seemingly more sophisticated integrand. To start, we separate out the last momentum coordinate kDk_{D} by writing the integrand of Eq. (47) as (∏iD−1cos⁡ki2)cos⁡kD2cos⁡(a+kD2)D−b−cos⁡kD\left(\prod_{i}^{D-1}\cos\frac{k_{i}}{2}\right)\frac{\cos\frac{k_{D}}{2}\cos\left(a+\frac{k_{D}}{2}\right)}{D-b-\cos k_{D}} where a=∑iD−1ki2a=\sum_{i}^{D-1}\frac{k_{i}}{2} and b=∑iD−1cos⁡kib=\sum_{i}^{D-1}\cos k_{i}. The kDk_{D}-dependent fraction on the right can be integrated as follows:

∫02πcos⁡k2cos⁡(a+k2)D−b−cos⁡k dk=π(D−b+1D−b−1−1)cos⁡a.\int_{0}^{2\pi}\frac{\cos\frac{k}{2}\cos\left(a+\frac{k}{2}\right)}{D-b-\cos k}\,dk =\pi\left(\sqrt{\frac{D-b+1}{D-b-1}}-1\right)\cos a.
∫[0,2π]D′cos⁡(∑iD′ki2)∏iD′cos⁡ki2 dD′k=πD′\int_{[0,2\pi]^{D'}}\cos\left(\sum_{i}^{D'}\frac{k_{i}}{2}\right)\prod_{i}^{D'}\cos\frac{k_{i}}{2}\,d^{D'}k=\pi^{D'}

for any D′D' (Here and below, we denote D′=D−1D'=D-1 for brevity), we obtain ZD-dim(N→∞)≈Z_{\mathrm{D\text{-}dim}}(N\rightarrow\infty)\approx

−1+∫[0,2π]D′cos⁡(∑iD′ki2)∏iD′cos⁡ki2πD′D+1−∑iD′cos⁡kiD−1−∑iD′cos⁡ki dD′k.(48)-1+\int_{[0,2\pi]^{D'}}\frac{\cos\left(\sum_{i}^{D'}\frac{k_{i}}{2}\right)\prod_{i}^{D'}\cos\frac{k_{i}}{2}}{\pi^{D'}}\sqrt{\frac{D+1-\sum_{i}^{D'}\cos k_{i}}{D-1-\sum_{i}^{D'}\cos k_{i}}}\,d^{D'}k. \tag*{(48)}

In general, we can continue with this dimensional reduction procedure to obtain ZD-dim(N→∞)≈Z_{\mathrm{D\text{-}dim}}(N\rightarrow\infty)\approx

∫[0,2π]D−dcos⁡(∑iD−dki2)∏iD−dcos⁡ki2πD−dfd(D−d−∑iD−dcos⁡ki) dD−dk.(49)\int_{[0,2\pi]^{D-d}}\frac{\cos\left(\sum_{i}^{D-d}\frac{k_{i}}{2}\right)\prod_{i}^{D-d}\cos\frac{k_{i}}{2}}{\pi^{D-d}}f_{d}\left(D-d-\sum_{i}^{D-d}\cos k_{i}\right)\,d^{D-d}k. \tag*{(49)}

where dd is the number of reduced dimensions and fd(x)f_d(x) is a weightage function that encapsulates the effects of dimensional reduction:

f0(x)=1x(50)f_0(x)=\frac{1}{x} \tag*{(50)}
f1(x)=1+2x−1(51)f_1(x)=\sqrt{1+\frac{2}{x}}-1 \tag*{(51)}
f2(x)=(x2+5x+8)K(−4x2+4x)−x(x+4)E(−4x2+4x)−2iK((x+2)2x(x+4))πx(x+4)+1πK(4(x+2)2)+2iK(x(x+4)(x+2)2)π(x+2)−1(52)f_2(x)=\frac{(x^2+5x+8)K\left(-\frac{4}{x^2+4x}\right)-x(x+4)E\left(-\frac{4}{x^2+4x}\right)-2iK\left(\frac{(x+2)^2}{x(x+4)}\right)}{\pi\sqrt{x(x+4)}}+\frac{1}{\pi}K\left(\frac{4}{(x+2)^2}\right)+\frac{2iK\left(\frac{x(x+4)}{(x+2)^2}\right)}{\pi(x+2)}-1 \tag*{(52)}

where E(y)=∫0π/21−ysin⁡2θ dθE(y)=\int_0^{\pi/2}\sqrt{1-y\sin^2\theta}\,\mathrm{d}\theta and K(y)=∫0π/211−ysin⁡2θ dθK(y)=\int_0^{\pi/2}\frac{1}{\sqrt{1-y\sin^2\theta}}\,\mathrm{d}\theta are elliptic integrals. Further expressions of fdf_d, d>2d>2 exist in principle, although they would be much more complicated. Note that f0f_0 is just the inverse Laplacian spectrum for the unbounded circuit; the other terms in the integrand keeps track of the source/sink as well as boundary effects, as we have obtained from the method of images. As an illustration,

Z3-dim(N→∞)≈1π∫02πcos⁡2k2 f2(1−cos⁡k) dk.(53)Z_{\text{3-dim}}(N\to\infty)\approx\frac{1}{\pi}\int_0^{2\pi}\cos^2\frac{k}{2}\,f_2(1-\cos k)\,\mathrm{d}k. \tag*{(53)}

References

References

  1. [1]J. Ningyuan, C. Owens, A. Sommer, D. Schuster, and J. Simon, Simulation of Topological Properties of Linear Circuit Lattices, Phys. Rev. X 5, 021031 (2015).
  2. [2]L. Song, H. Yang, Y. Cao, and P. Yan, Realization of the Square-Root Higher-Order Topological Insulator in Electric Circuits, Nano Lett. 20, 7566 (2020).
  3. [3]Y. Nakata, T. Okada, T. Nakanishi, and M. Kitano, Circuit model for hybridization modes in metamaterials and its analogy to the quantum tight-binding model: Circuit model for hybridization modes in metamaterials, Phys. Status Solidi B 249, 2293 (2012).
  4. [4]C. H. Lee, S. Imhof, C. Berger, F. Bayer, J. Brehm, L. W. Molenkamp, T. Kiessling, and R. Thomale, Topological Circuits, Commun. Phys. 1, 39 (2018).
  5. [5]S. Imhof, C. Berger, F. Bayer, J. Brehm, L. W. Molenkamp, T. Kiessling, F. Schindler, C. H. Lee, M. Greiter, T. Neupert, and R. Thomale, Topological-circuit realization of topological corner modes, Nature Physics 14, 925 (2018).
  6. [6]T. Helbig, T. Hofmann, C. H. Lee, R. Thomale, S. Imhof, L. W. Molenkamp, and T. Kiessling, Band structure engineering and reconstruction in electric circuit networks, Phys. Rev. B 99, 161114 (2019).
  7. [7]J. Bao, D. Zou, W. Zhang, W. He, H. Sun, and X. Zhang, Topological circuit outpost insulator with topologically protected corner states, Phys. Rev. B 100, 201406 (2019).
  8. [8]S. M. Rafi-Ul-Islam, Z. Bin Siu, and M. B. A. Jalil, Topoelectrical circuit realization of a Weyl semimetal heterojunction, Commun Phys 3, 72 (2020).
  9. [9]S. M. Rafi-Ul-Islam, Z. B. Siu, C. Sun, and M. B. A. Jalil, Realization of Weyl semimetal phases in topoelectrical circuits, New J. Phys. 22, 023025 (2020).
  10. [10]Y. Wang, H. M. Price, B. Zhang, and Y. D. Chong, Circuit implementation of a four-dimensional topological insulator, Nat Commun 11, 2356 (2020).
  11. [11]L. Li, C. H. Lee, and J. Gong, Emergence and full 3D-imaging of nodal boundary Seifert surfaces in 4D topological matter, Commun Phys 2, 135 (2019).
  12. [12]M. Ezawa, Electric circuits for non-Hermitian Chern insulators, Phys. Rev. B 100, 081401 (2019).
  13. [13]C. H. Lee, A. Sutrino, T. Hofmann, T. Helbig, Y. Liu, Y. S. Ang, L. K. Ang, X. Zhang, M. Greiter, and R. Thomale, Imaging nodal knots in momentum space through topoelectrical circuits, Nat Commun 11, 4385 (2020).
  14. [14]T. Helbig, T. Hofmann, S. Imhof, M. Abdelghany, T. Kiessling, L. W. Molenkamp, C. H. Lee, A. Szameit, M. Greiter, and R. Thomale, Generalized bulk-boundary correspondence in non-Hermitian topoelectrical circuits, Nat. Phys. 16, 747 (2020).
  15. [15]W. Zhang, D. Zou, J. Bao, W. He, Q. Pei, H. Sun, and X. Zhang, Topoelectrical-circuit realization of a four-dimensional hexadecacole insulator, Phys. Rev. B 102, 100102 (2020).
  16. [16]N. A. Olekhno, E. I. Kretov, A. A. Stepanenko, P. A. Ivanova, V. W. Yaroshenko, E. M. Puttina, D. S. Filonov, B. Cappello, L. Matekovits, and M. A. Gorlach, Topological edge states of interacting photon pairs emulated in a topoelectrical circuit, Nat Commun 11, 1436 (2020).
  17. [17]X. Ni, Z. Xiao, A. B. Khanikaev, and A. Alù, Robust Multiplexing with Topological Higher-Order Chern Insulators, Phys. Rev. Applied 13, 064031 (2020).
  18. [18]T. Hofmann, T. Helbig, C. H. Lee, M. Greiter, and R. Thomale, Chiral Voltage Propagation and Calibra- S. M. Rafi-Ul-Islam, Z. B. Siu, H. Sahin, C. H. Lee, and M. B. A. Jalil, System size dependent topological zero modes in coupled topoelectrical chains, Phys. Rev. B 106, 075158 (2022).
  19. [20]C. Shang, S. Liu, R. Shao, P. Han, X. Zhang, X. Zhang, K. N. Salama, W. Gao, C. H. Lee, R. Thomale, A. Manchon, S. Zhang, T. J. Cui, and U. Schwingenschlögl, Experimental identification of the second-order non-hermitian skin effect with physics-informed machine learning, Advanced Science n/a, 2202922 (2022).DOI
  20. [21]M. Ezawa, Electric circuits for universal quantum gates and quantum Fourier transformation, Phys. Rev. Research 2, 023278 (2020).DOI
  21. [22]M. Ezawa, Universal quantum gates, artificial neurons, and pattern recognition simulators, Phys. Rev. Research 3, 023051 (2021).DOI
  22. [23]N. Pan, T. Chen, H. Sun, and X. Zhang, Electric-Circuit Realization of Fast Quantum Search, Research 2021, 1 (2021).DOI
  23. [24]M. A. Quiroz-Juárez, C. You, J. Carrillo-Martínez, D. Montiel-Álvarez, J. L. Aragón, O. S. Magaña-Loaiza, and R. d. J. León-Montiel, Reconfigurable network for quantum transport simulations, Phys. Rev. Research 3, 013010 (2021).
  24. [25]T. Kotwal, F. Moseley, A. Stegmaier, S. Imhof, H. Brand, T. Kiebling, R. Thomale, H. Ronellenfitsch, and J. Dunkel, Active topoelectrical circuits, Proc. Natl. Acad. Sci. U.S.A. 118, e2106411118 (2021).
  25. [26]E. Kengne, W.-M. Liu, L. Q. English, and B. A. Malomed, Ginzburg–Landau models of nonlinear electric transmission networks, Physics Reports 982, 1 (2022).DOI
  26. [27]C. L. Kane and E. J. Mele, Z 2 Topological Order and the Quantum Spin Hall Effect, Phys. Rev. Lett. 95, 146802 (2005).
  27. [28]L. Jin and Z. Song, Solutions of P T -symmetric tight-binding chain and its equivalent Hermitian counterpart, Phys. Rev. A 80, 052107 (2009).DOI
  28. [29]K. Sun, Z. Gu, H. Katsura, and S. Das Sarma, Nearly Flatbands with Nontrivial Topology, Phys. Rev. Lett. 106, 236803 (2011).
  29. [30]M. Z. Hasan and C. L. Kane, Colloquium : Topological insulators, Rev. Mod. Phys. 82, 3045 (2010).
  30. [31]C. H. Lee and X.-L. Qi, Lattice construction of pseudopotential Hamiltonians for fractional Chern insulators, Phys. Rev. B 90, 085103 (2014).DOI
  31. [32]L. Chen, T. Mazaheri, A. Seidel, and X. Tang, The impossibility of exactly flat non-trivial Chern bands in strictly local periodic tight binding models, J. Phys. A: Math. Theor. 47, 152001 (2014).DOI
  32. [33]Y. Gu, C. H. Lee, X. Wen, G. Y. Cho, S. Ryu, and X.-L. Qi, Holographic duality between ( 2 + 1 ) -dimensional quantum anomalous Hall state and ( 3 + 1 )-dimensional topological insulators, Phys. Rev. B 94, 125107 (2016).DOI
  33. [34]S. M. Rafi-Ul-Islam, Z. B. Siu, H. Sahin, C. H. Lee, and M. B. A. Jalil, Unconventional skin modes in generalized topoelectrical circuits with multiple asymmetric couplings, Phys. Rev. Research 4, 043108 (2022).DOI
  34. [35]S. Kirkpatrick, Percolation and Conduction, Rev. Mod. Phys. 45, 574 (1973).
  35. [36]L. Lavavelli, The Resistive Net and Finite-Difference tion in a Topoelectrical Chern Circuit, Phys. Rev. Lett. 122, 247702 (2019).
  36. [37]F. J. Bartis, Let’s Analyze the Resistance Lattice, American Journal of Physics 40, 1246 (1972).
  37. [38]A. H. Zemanian, A classical puzzle: The driving-point resistances of infinite grids, IEEE Circuits Syst. Mag. 6, 7 (1984).DOI
  38. [39]J. Cserti, G. Dávid, and A. Piróth, Perturbation of infinite networks of resistors, American Journal of Physics 70, 153 (2002).
  39. [40]M. Q. Owaidat, A. Al-Badawi, and M. Abu-Samak, The two-point resistance on the diamond cubic lattice, Eur. Phys. J. Plus 133, 199 (2018).DOI
  40. [41]M. Q. Owaidat, R. S. Hijjawi, and J. M. Khalifeh, Perturbation theory of uniform tiling of space with resistors, Eur. Phys. J. Plus 129, 29 (2014).DOI
  41. [42]C. Koutschan, Lattice Green functions of the higher-dimensional face-centered cubic lattices, J. Phys. A: Math. Theor. 46, 125005 (2013).DOI
  42. [43]N. S. Izmailian and R. Kenna, A generalised formulation of the Laplacian approach to resistor networks, J. Stat. Mech. 2014, P09016 (2014).DOI
  43. [44]N. S. Izmailian, R. Kenna, and F. Y. Wu, The two-point resistance of a resistor network: a new formulation and application to the cobweb network, J. Phys. A: Math. Theor. 47, 035003 (2014).DOI
  44. [45]Z.-Z. Tan, J. Asad, and M. Qwaidat, Resistance formulae of a multipurpose n-step network and its application in LC network: Resistance Formulae of a Multipurpose n -Step Network, Int. J. Circ. Theor. Appl. 45, 1942 (2017).DOI
  45. [46]J. Cserti, G. Széchenyi, and G. Dávid, Uniform tiling with electrical resistors, J. Phys. A: Math. Theor. 44, 215201 (2011).DOI
  46. [47]G. Venezian, On the resistance between two points on a grid, American Journal of Physics 62, 1000 (1994).
  47. [48]M. Q. Owaidat, R. S. Hijjawi, and J. M. Khalifeh, Interstitial single resistor in a network of resistors application of the lattice Green’s function, J. Phys. A: Math. Theor. 43, 375204 (2010).
  48. [49]M. Owaidat, J. Asad, and Z.-Z. Tan, Resistance computation of generalized decorated square and simple cubic network lattices, Results in Physics 12, 1621 (2019).DOI
  49. [50]D. Atkinson and F. J. van Steenwijk, Infinite resistive lattices, American Journal of Physics 67, 486 (1999).
  50. [51]P. G. Doyle and J. L. Snell, Random Walks and Electric Networks, arXiv:math/0001057 (2000).DOI
  51. [52]Z.-Z. Tan and Z. Tan, Electrical properties of an m*n rectangular network, Phys. Scr. 95, 035226 (2020).DOI
  52. [53]Z.-Z. Tan, Two-point resistance of an m*n resistor network with an arbitrary boundary and its application in RLC network, Chinese Phys. B 25, 050504 (2016).
  53. [54]J. H. Asad, R. S. Hijjawi, A. J. Sakaji, and J. M. Khalifeh, Infinite network of identical capacitors by Green’s function, Int. J. Mod. Phys. B 19, 3713 (2005).
  54. [55]J. W. Essam and F. Y. Wu, The exact evaluation of the corner-to-corner resistance of an M*N resistor network: asymptotic expansion, J. Phys. A: Math. Theor. 42, 025205 (2009).
  55. [56]N. S. Izmailian and M.-C. Huang, Asymptotic expansion of the resistance between two maximally separated nodes on an M by N resistor network, Phys. Rev. E 82, 011125 (2010).
  56. [57]J. P. Clerc, G. Giraud, J. M. Luck, and T. Robin, Di- electric resonances of lattice animals and other fractal clusters, J. Phys. A: Math. Gen. 29, 4781 (1996).
  57. [58]J. Cserti, Application of the lattice Green’s function for calculating the resistance of an infinite network of resistors, American Journal of Physics 68, 896 (2000).
  58. [59]T. Morita, Useful Procedure for Computing the Lattice Green’s Function-Square, Tetragonal, and bcc Lattices, Journal of Mathematical Physics 12, 1744 (1971).DOI
  59. [60]J. Asad, A. Diab, M. Owaidat, and J. Khalifeh, Perturbed Infinite 3D Simple Cubic Network of Identical Capacitors, Acta Phys. Pol. A 126, 777 (2014).DOI
  60. [61]M. Q. Owaidat and J. H. Asad, Resistance calculation of three-dimensional triangular and hexagonal prism lattices, Eur. Phys. J. Plus 131, 309 (2016).DOI
  61. [62]M. Q. Owaidat, J. H. Asad, and Z.-Z. Tan, On the per-turbation of a uniform tiling with resistors, Int. J. Mod. Phys. B 30, 1650166 (2016).DOI
  62. [63]Z. Tan, Z.-Z. Tan, J. H. Asad, and M. Q. Owaidat, Electrical characteristics of the 2*n and □*n circuit network, Phys. Scr. 94, 055203 (2019).DOI
  63. [64]Z.-Z. Tan, Recursion-transform method for computing resistance of the complex resistor network with three arbitrary boundaries, Phys. Rev. E 91, 052122 (2015).
  64. [65]Y.-T. Zhang, X. Hu, H.-X. Chen, M.-Y. Wang, W.-J. Chen, X.-Y. Fang, and Z.-Z. Tan, Resistance Theory of General 2*n Resistor Networks, Adv. Theory Simul. 4, 2000255 (2021).DOI
  65. [66]H.-X. Chen, M.-Y. Wang, W.-J. Chen, X.-Y. Fang, and Z.-Z. Tan, Equivalent complex impedance of n-order RLC network, Phys. Scr. 96, 075202 (2021).DOI
  66. [67]Z.-z. Tan, H. Zhu, J. H. Asad, C. Xu, and H. Tang, Characteristic of the equivalent impedance for an m*n RLC network with an arbitrary boundary, Frontiers Inf Technol Electronic Eng 18, 2070 (2017).DOI
  67. [68]R. E. Aitchison, Resistance between Adjacent Points of Liebman Mesh, American Journal of Physics 32, 566 (1964).DOI
  68. [69]G. S. Joyce, Exact evaluation of the simple cubic lattice Green function for a general lattice point, J. Phys. A: Math. Gen. 35, 9811 (2002).DOI
  69. [70]M. Jeng, Random walks and effective resistances on toroidal and cylindrical grids, American Journal of Physics 68, 37 (2000).DOI
  70. [71]H.-X. Chen, L. Yang, and M.-J. Wang, Electrical characteristics of n-ladder network with internal load, Results in Physics 15, 102488 (2019).DOI
  71. [72]H.-X. Chen and L. Yang, Electrical characteristics of n-ladder network with external load, Indian J Phys 94, 801 (2020).DOI
  72. [73]C.-P. Chen and Z.-Z. Tan, Electrical characteristics of an asymmetric N-step network, Results in Physics 19, 103399 (2020).DOI
  73. [74]X.-Y. Fang and Z.-Z. Tan, Circuit network theory of n-horizontal bridge structure, Sci Rep 12, 6158 (2022).DOI
  74. [75]V. Černianová, J. Brenkuš, and V. Stopjaková, Non-Symmetric Finite Networks: The Two-Point Resistance, Journal of Electrical Engineering 65, 283 (2014).DOI
  75. [76]G. S. Joyce, Exact results for the diamond lattice Green function with applications to uniform random walks in a plane, J. Phys. A: Math. Theor. 50, 425001 (2017).DOI
  76. [77]J. H. Asad, A. A. Diab, M. Q. Owaidat, R. S. Hijjawi, and J. M. Khalifeh, Infinite Body Centered Cubic Network of Identical Resistors, Acta Phys. Pol. A 125, 60 (2014).
  77. [78]J. H. Asad, A. A. Diab, B. S. Hijjawi, and J. M. Khalifeh, Infinite face-centered-cubic network of identical resistors: Application to lattice Green’s function, Eur. Phys. J. Plus 128, 2 (2013).DOI
  78. [79]I. I. Guseinov and B. A. Mamedov, A unified treatment of the lattice Green function, generalized Watson integral and associated logarithmic integral for the d-dimensional hypercubic lattice, Philosophical Magazine 87, 1107 (2007).DOI
  79. [80]M. Mamedov, Calculation of two-point resistances for conducting media needs regularization of Coulomb singularities, Eur. Phys. J. Plus 134, 559 (2019).DOI
  80. [81]N. Zenine, S. Hassani, and J. M. Maillard, Lattice Green functions: the seven-dimensional face-centred cubic lattice, J. Phys. A: Math. Theor. 48, 035205 (2015).DOI
  81. [82]Z.-Z. Tan, Recursion-transform approach to compute the resistance of a resistor network with an arbitrary boundary, Chinese Phys. B 24, 020503 (2015).
  82. [83]Z.-Z. Tan, Recursion-Transform method to a non-regular m×n cobweb with an arbitrary longitude, Sci Rep 5, 11266 (2015).DOI
  83. [84]M. Q. Owaidat, J. H. Asad, and J. M. Khalifeh, Resistance calculation of the decorated centered cubic networks: Applications of the Green’s function, Mod. Phys. Lett. B 28, 1450252 (2014).DOI
  84. [85]M. Q. Owaidat, R. S. Hijjawi, J. H. Asad, and J. M. Khalifeh, Electrical networks with interstitial single capacitor, Mod. Phys. Lett. B 27, 1350123 (2013).DOI
  85. [86]M. Q. Owaidat, R. S. Hijjawi, and J. M. Khalifeh, Sub-situational single resistor in an infinite square lattice application to lattice Green’s function, Mod. Phys. Lett. B 24, 2057 (2010).
  86. [87]M. A. Jafarizadeh, R. Sufiani, and S. Jafarizadeh, Calculating two-point resistances in distance-regular resistor networks, J. Phys. A: Math. Theor. 40, 4949 (2007).
  87. [88]S. Giordano, Disordered lattice networks: general theory and simulations, Int. J. Circ. Theor. Appl. 33, 519 (2005).DOI
  88. [89]J. D. Jackson, Classical electrodynamics (American Association of Physics Teachers, 1999).
  89. [90]D. J. Griffiths, Introduction to electrodynamics (American Association of Physics Teachers, 2005).
  90. [91]K. F. Riley, M. P. Hobson, and S. J. Bence, Mathematical methods for physics and engineering (American Association of Physics Teachers, 1999).
  91. [92]R. Yang, J. W. Tan, T. Tai, J. M. Koh, Li. S. Longhi, and C. H. Lee, Designing non-Hermitian real spectra through electrostatics, Science Bulletin 67, 1865 (2022).DOI
  92. [93]M. Mamedov, Electrical resistance between pairs of vertices of a conducting cube and continuum limit for a cubic resistor network, J. Phys. Commun. 1, 035002 (2017).DOI
  93. [94]F. Y. Wu, Theory of resistor networks: the two-point resistance, J. Phys. A: Math. Gen. 37, 6653 (2004).
  94. [95]M. Q. Owaidat, Regular Resistor Networks in Two Dimensions (Archimedean Lattices), APR 6, p100 (2014).DOI
  95. [96]M. Q. Owaidat, Resistance calculation of the face-centered cubic lattice: Theory and experiment, American Journal of Physics 81, 918 (2013).DOI
  96. [97]E. Abrahams, P. W. Anderson, D. C. Licciardello, and T. V. Ramakrishnan, Scaling Theory of Localization: Absence of Quantum Diffusion in Two Dimensions, Phys. Rev. Lett. 42, 673 (1979).
  97. [98]W. J. Tzeng and F. Y. Wu, Theory of impedance networks: the two-point impedance and LC resonances, J. Phys. A: Math. Gen. 39, 8579 (2006).
  98. [99]M. Mamode, Revisiting the discrete planar Laplacian: exact results for the lattice Green function and continuum limit, Eur. Phys. J. Plus 136, 412 (2021).DOI
  99. [100]R. P. Feynman, R. B. Leighton, and M. Sands, The feynman lectures on physics: vol. II, Vol. 33 (American Association of Physics Teachers, 1965) pp. 750–752.
  100. [101]L. Li and C. H. Lee, Non-Hermitian pseudo-gaps, Science Bulletin 67, 685 (2022).DOI
  101. [102]S. M. Rafi-Ul-Islam, H. Sahin, Z. B. Siu, and M. B. A. Jalil, Interfacial skin modes at a non-Hermitian heterojunction, Phys. Rev. Research 4, 043021 (2022).DOI
  102. [103]G. C. F. Pantoja and W. S. Elias, Green’s functions and method of images: an interdisciplinary topic usually cast aside in physics textbooks (2021), arXiv:2006.09999 [physics].arxiv.org/abs/2006.09999
  103. [104]L. R. Mead, Resistance of a square lamina by the method of images, American Journal of Physics 77, 259 (2009).DOI
  104. [105]H. Hohmann, T. Hofmann, T. Helbig, S. Imhof, H. Brand, L. K. Upreti, A. Stegmaier, A. Fritzsche, T. Müller, U. Schwingenschlögl, C. H. Lee, M. Greiter, L. W. Molenkamp, T. Kießling, and R. Thomale, Observation of conoidal wave localization in non-linear topoelectric circuits, arXiv e-prints , arXiv:2206.09931 (2022), arXiv:2206.09931 [cond-mat.mes-hall].DOI
  105. [106]P. M. Lenggenhager, A. Stegmaier, L. K. Upreti, T. Hofmann, T. Helbig, A. Vollhardt, M. Greiter, C. H. Lee, S. Imhof, H. Brand, T. Kießling, I. Boettcher, T. Neupert, R. Thomale, and T. Bzdušek, Simulating hyperbolic space on a circuit board, Nat Commun 13, 4373 (2022).
  106. [107]H. Zhang, T. Chen, L. Li, C. H. Lee, and X. Zhang, Electrical circuit realization of topological switching for the non-hermitian skin effect, Physical Review B 107, 085426 (2023).
  107. [108]S. Katsura, T. Morita, S. Inawashiro, T. Horiguchi, and Y. Abe, Lattice Green’s Function. Introduction, Journal of Mathematical Physics 12, 892 (1971).DOI
  108. [109]A. J. Guttmann, Lattice Green’s functions in all dimensions, J. Phys. A: Math. Theor. 43, 250205 (2010).
  109. [110]G. S. Joyce and W. C. Price, On the simple cubic lattice green function, Philosophical Transactions of the Royal Society of London. Series A, Mathematical and Physical Sciences 273, 583 (1973).
  110. [111]E. W. Montroll and G. H. Weiss, Random Walks on Lattices. II, Journal of Mathematical Physics 6, 167 (1965).
  111. [112]F. Perrier and F. Girault, Symmetries, recurrence, and explicit expressions of two-point resistances in 2*n globe resistor networks, Eur. J. Phys. 42, 055201 (2021).
  112. [113]But see the experiment in Ref. [137], which demonstrates that such passive RLC elements can bring about non-local impedance responses in suitably designed circuits.
  113. [114]C. H. Lee, Y. Wang, Y. Chen, and X. Zhang, Electromagnetic response of quantum Hall systems in dimensions five and six and beyond, Phys. Rev. B 98, 094434 (2018).
  114. [115]In higher-dimensional systems where non-reciprocity also accompanies resistive non-Hermiticity, the non-local response can alter the effective dimensionality of the entire system [138, 139].
  115. [116]L. Li, C. H. Lee, S. Mu, and J. Gong, Critical non-Hermitian skin effect, Nat Commun 11, 5491 (2020).
  116. [117]L. Li, C. H. Lee, and J. Gong, Impurity induced scale-free localization, Commun Phys 4, 42 (2021).
  117. [118]S. M. Rafi-Ul-Islam, Z. B. Siu, H. Sahin, C. H. Lee, and M. B. A. Jalil, Critical hybridization of skin modes in coupled non-Hermitian chains, Phys. Rev. Research 4, 013243 (2022).
  118. [119]Z. B. Siu, S. M. Rafi-Ul-Islam, and M. B. Jalil, Critical non-Hermitian Skin Effect in a Single Closed non-Hermitian Chain, preprint (In Review, 2022).
  119. [120]S. M. Rafi-Ul-Islam, Z. B. Siu, H. Sahin, and M. B. A. Jalil, Valley hall effect and kink states in topoelectrical circuits, Phys. Rev. Res. 5, 013107 (2023).
  120. [121]D. Vanderbilt and R. Resta, Chapter 5 Quantum Electrostatics of Insulators: Polarization, Wannier Functions, and Electric Fields, in Contemporary Concepts of Condensed Matter Science, Vol. 2 (Elsevier, 2006) pp. 139–163.
  121. [122]X.-L. Qi, T. L. Hughes, and S.-C. Zhang, Topological field theory of time-reversal invariant insulators, Phys. Rev. B 78, 195424 (2008).
  122. [123]A. A. Soluyanov and D. Vanderbilt, Wannier representation of Z 2 topological insulators, Phys. Rev. B 83, 035108 (2011).
  123. [124]C. H. Lee and P. Ye, Free-fermion entanglement spectrum through Wannier interpolation, Phys. Rev. B 91, 085119 (2015).
  124. [125]C.-X. Liu, S.-C. Zhang, and X.-L. Qi, The Quantum Anomalous Hall Effect: Theory and Experiment, Annu. Rev. Condens. Matter Phys. 7, 301 (2016).
  125. [126]M. Jürgensen and M. C. Rechtsman, Chern Number Governs Soliton Motion in Nonlinear Thouless Pumps, Phys. Rev. Lett. 128, 113901 (2022).
  126. [127]J. Wang, S. Valligatla, Y. Yin, L. Schwarz, M. Medina-Sánchez, S. Baunack, C. H. Lee, R. Thomale, S. Li, V. M. Fomin, et al., Experimental observation of berry phases in optical möbius-strip microcavities, Nature Photonics 17, 120 (2023).
  127. [128]C.-Z. Chang, J. Zhang, X. Feng, J. Shen, Z. Zhang, M. Guo, K. Li, Y. Ou, P. Wei, L.-L. Wang, Z.-Q. Ji, Y. Feng, S. Ji, X. Chen, J. Jia, X. Dai, Z. Fang, S.-C. Zhang, K. He, Y. Wang, L. Lu, X.-C. Ma, and Q.-K. Xue, Experimental Observation of the Quantum Anomalous Hall Effect in a Magnetic Topological Insulator, Science 340, 167 (2013).
  128. [129]L. M. Nash, D. Kleckner, A. Read, V. Vitelli, A. M. Turner, and W. T. M. Irvine, Topological mechanics of gyroscopic metamaterials, Proc. Natl. Acad. Sci. U.S.A. 112, 14495 (2015).
  129. [130]Y. Ding, Y. Peng, Y. Zhu, X. Fan, J. Yang, B. Liang, X. Zhu, X. Wan, and J. Cheng, Experimental Demonstration of Acoustic Chern Insulators, Phys. Rev. Lett. 122, 014302 (2019).
  130. [131]J. Motruk and I. Na, Detecting Fractional Chern Insulators in Optical Lattices Through Quantized Displacement, Phys. Rev. Lett. 125, 236401 (2020).
  131. [132]J. M. Koh, T. Tai, and C. H. Lee, Simulation of Interaction-Induced Chiral Topological Dynamics on a Digital Quantum Computer, Phys. Rev. Lett. 129, 140502 (2022).
  132. [133]X. Zhang, B. Zhang, H. Sahin, Z. B. Siu, S. M. Rafi-Ul-Islam, J. F. Kong, B. Shen, M. B. A. Jalil, R. Thomale, and C. H. Lee, Anomalous fractal scaling in two-dimensional electric networks, Communications Physics 6, 151 (2023).
  133. [134]Z. Gong, Y. Ashida, K. Kawabata, K. Takasan, S. Higashikawa, and M. Ueda, Topological Phases of Non-Hermitian Systems, Phys. Rev. X 8, 031079 (2018).
  134. [135]K. Kawabata, K. Shiozaki, M. Ueda, and M. Sato, Symmetry and Topology in Non-Hermitian Physics, Phys. Rev. X 9, 041015 (2019).
  135. [136]L. Li, C. Yang, and S. Chen, Winding numbers of phase transition points for one-dimensional topological systems, EPL 112, 10004 (2015).

Paper details

vlnc:W_9strwq36
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents