A variational quantum algorithm for tackling multi-dimensional Poisson equations with inhomogeneous boundary conditions
Minjin Choi, Hoon Ryu · 2024
Read A variational quantum algorithm for tackling multi-dimensional Poisson equations with inhomogeneous boundary conditions by Minjin Choi, Hoon Ryu (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2411-03009.txt.
Description
A 2024 arXiv research paper by Minjin Choi, Hoon Ryu. We design a variational quantum algorithm to solve multi-dimensional Poisson equations with mixed boundary conditions that are typically required in various fields of computational science. Employing an objective function that is formulated with the concept of the minimal potential energy, we not only present in-depth discussion on the cost-efficient & noise-robust design of quantum circuits that are essential for evaluation of the objective function, but, more remarkably, employ the proposed algorithm to calculate bias-dependent spatial distributions of electric fields in
Catalog record
| CREATOR | Minjin Choi, Hoon Ryu |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 44 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2411-03009
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A variational quantum algorithm for tackling multi-dimensional Poisson equations with inhomogeneous boundary conditions Minjin Choi, Hoon Ryu Abstract We design a variational quantum algorithm to solve multi-dimensional Poisson equations with mixed boundary conditions that are typically required in various fields of computational science. Employing an objective function that is formulated with the concept of the minimal potential energy, we not only present in-depth discussion on the cost-efficient & noise-robust design of quantum circuits that are essential for evaluation of the objective function, but, more remarkably, employ the proposed algorithm to calculate bias-dependent spatial distributions of electric fields in semiconductor systems that are described with a two-dimensional domain and up to 10-qubit circuits. Extending the application scope to multi-dimensional problems with mixed boundary conditions for the first time, fairly solid computational results of this work clearly demonstrate the potential of variational quantum algorithms to tackle Poisson equations derived from physically meaningful problems. I Introduction The Poisson equation is a second-order partial differential equation that plays a significant role in analyzing physical phenomena in various fields of computational science such as fluid dynamics [ 1 , 2 ] , nanoelectronics [ 3 , 4 , 5 , 6 ] , computational electromagnetics [ 7 , 8 ] . In general, it is not easy to determine the solution of a Poisson equation analytically, so computer-based calculations are commonly employed with aids of numerical analysis, by transforming the master problem into a linear system problem A\mathbf{v}=\mathbf{b} where \mathbf{v} represents the solution of a Poisson system matrix A given by a right-hand-side (RHS) \mathbf{b} [ 9 ] . Consequently, the computational expense of a Poisson equation problem directly depends on the degrees of freedom (DOFs) of the system matrix that would generally become larger if we need to increase the accuracy of numerical solutions or to handle larger domains in a real space. Recently, quantum computing has obtained huge attention due to the potential of solving some computational problems more efficiently than its classical counterpart by leveraging quantum phenomena [ 10 , 11 , 12 , 13 , 14 , 15 ] , and, in particular, Harrow et al. proposed the Harrow-Hassidim-Lloyd (HHL) algorithm that can achieve exponential speed-up relative to classical algorithms for solving linear system problems [ 14 ] . Yudong et al. demonstrated that approximate solutions of Poisson equations can be obtained based on the HHL algorithm [ 16 ] , but its effectiveness in terms of computing cost hinges on fault-tolerant quantum computing that is not tangible yet. It is fair to say that the current status of quantum computing is in the noisy intermediate-scale quantum (NISQ) era [ 17 ] , where computing devices are subject to substantial noises and thus are limited in the number of quantum bits (qubits). Variational quantum algorithms (VQAs) [ 18 ] , which are executable in NISQ devices through a hybrid utilization of quantum and classical computing, have led to significant efforts to explore their utility in diverse numerical applications such as determination of low-lying eigenpairs of symmetric matrices [ 19 ] and solutions of linear systems [ 20 ] , combinatorial optimization [ 21 ] , and so on. VQAs dedicated to the Poisson equation have been also reported, showing their capability for securing solutions of simple one-dimensional problems [ 22 , 23 ] . Rigorous studies on their utility for the Poisson equation, however, are still necessary since physically meaningful problems are typically represented with multi-dimensional domains and mixed boundary conditions [ 3 , 2 , 6 ] , motivating comprehensive investigations with more complicated scenarios to examine the effectiveness of VQAs for prediction of physical phenomena with Poisson equations. So in this work, we design a VQA that can be useful to calculate multi-dimensional problems of inhomogeneous boundaries with a Poisson equation. To achieve cost-efficiency and noise-robustness of the algorithm, we begin the design process with adoption of strong points of the two previous studies that not only focus on one-dimensional problems, but are limited by impracticality in the objective (cost) function [ 22 ] and the circuit decomposition [ 23 ] . In more detail, we employ the cost function derived upon the concept of the minimal potential energy by Sato et al. [ 23 ] that is much simpler than the one employed by Liu et al. [ 22 ] . For evaluation of the cost function, however, Sato et al. utilize quantum circuits involving multi-controlled X logics that are not quite affordable in the NISQ regime, so here we design evaluation circuits with only one- & two-qubit gates as Liu et al. have done. The algorithm designed in this work can evaluate the cost function with circuits of polynomial complexities, and, more remarkably, the number of circuits required to get the cost function is independent of the dimensionality of target problems. Finally, the practicality of our algorithm is carefully examined with two-dimensional problems including the calculation of potential energy distributions in a silicon-based structure that are manipulated by electrical biases imposed on multiple electrodes. All the numerical simulations are conducted with the PENNYLANE software development kit [ 24 ] to emulate a gate-based quantum computer, and the optimizer based on L-BFGS-B algorithm [ 25 ] is used to classically update ansatz parameters. Overall results of this work clearly indicate that multi-dimensional Poisson equations of mixed boundary conditions can be solved with fairly sound accuracy, revealing the potential of VQA to address physically meaningful problems. II Methods II.1 The cost function employed to solve Poisson equations Let us consider the d -dimensional Poisson equation -\Delta u(\mathbf{x})=f(\mathbf{x}),\quad\mathbf{x}\in(0,1)^{d}, (1) where \Delta is the Laplace operator and f is a given smooth function [ 26 ] . Equation ( 1 ) can be then numerically transformed into a linear system problem A\mathbf{v}=\mathbf{b}, (2) where A is a m^{d}\times m^{d} system matrix describing the computational domain, \mathbf{b} is a m^{d}\times 1 vector that represents a perturbation given by the function f , and m is a number of grids employed for discretization of the domain along each axis. The solution \mathbf{v} of a linear system problem in Equation ( 2 ) then becomes the approximate solution of the Poisson equation. When a d -dimensional computational domain is discretized with a finite difference method [ 9 ] , the Poisson system matrix A becomes \displaystyle A \displaystyle= \displaystyle A_{1}\otimes I^{(m)}\otimes\cdots\otimes I^{(m)}+I^{(m)}\otimes A_{2}\otimes I^{(m)}\otimes\cdots\otimes I^{(m)} (3) \displaystyle+\cdots+I^{(m)}\otimes\cdots\otimes I^{(m)}\otimes A_{d}, where I^{(m)} is a m\times m identity matrix and A_{i} is a m\times m matrix representing the one-dimensional Poisson equation of the i -th axis. The one-dimensional Poisson matrix A_{i} depends on the boundary condition, so, for example, it is given by A_{D}\equiv\begin{pmatrix}2&-1&0&\cdots&0&0&0\\ -1&2&-1&\cdots&0&0&0\\ 0&-1&2&\cdots&0&0&0\\ &\vdots&&\ddots&&\vdots&\\ 0&0&0&\cdots&2&-1&0\\ 0&0&0&\cdots&-1&2&-1\\ 0&0&0&\cdots&0&-1&2\end{pmatrix}\in\mathbb{R}^{m\times m} (4) with Dirichlet (closed) boundaries, and A_{N}\equiv\begin{pmatrix}1&-1&0&\cdots&0&0&0\\ -1&2&-1&\cdots&0&0&0\\ 0&-1&2&\cdots&0&0&0\\ &\vdots&&\ddots&&\vdots&\\ 0&0&0&\cdots&2&-1&0\\ 0&0&0&\cdots&-1&2&-1\\ 0&0&0&\cdots&0&-1&1\end{pmatrix}\in\mathbb{R}^{m\times m} (5) with Neumann (zero-derivative) boundaries. Then, a Poisson system matrix A_{ND} that describes a two-dimensional domain having Neumann and Dirichlet boundaries along the x_{1} - and the x_{2} -axis, respectively, can be expressed as A_{ND}=A_{N}\otimes I^{(m)}+I^{(m)}\otimes A_{D}. (6) Since the Poisson matrix is always positive-definite, the linear system problem associated with Poisson equations can become an optimization problem. In more detail, for a positive-definite matrix A , the condition A\mathbf{v}_{0}=\mathbf{b} is