Origin released the first quantum computational fluid dynamics simulation software in China

Recently, OriginQ officially released the first quantum computational fluid dynamics simulation software "OriginQ QCFD", which is based on the quantum finite volume method and can realize the sub linear acceleration of CFD simulation in principle. In the future, the software can carry out more efficient numerical simulation experiments in aerospace, thermal power, environmental engineering and other fields.

Computational fluid dynamics (CFD) is an interdisciplinary subject integrating fluid mechanics, mathematics and computer science. With the powerful computing power of computer and discrete mathematical method, it can numerically solve the flow field information. At present, it has been widely used in aerospace, marine engineering, energy engineering and other fields, which greatly reduces the test cost and speeds up the industrial design process. However, the traditional computer will feel "inadequate" when facing some situations with strict requirements for calculation accuracy and huge amount of grid, which seriously hinders the efficiency of industrial design.

In recent years, with the development of quantum computing, researchers began to gradually explore the possibility of using quantum computers with powerful computing power to solve hydrodynamic problems. In 2019, Airbus held the quantum computing global challenge. Benyuan quantum was lucky to be invited to participate and won the single champion of quantum computing fluid power project. In February this year, the joint research team of OriginQ and China University of science and technology proposed a quantum finite volume method that can support the input and output of classical data, which can reduce the linear complexity of classical algorithms to sub linearity, which indicates that the cross research of quantum computing and hydrodynamics is gradually moving towards reality from the theoretical level. (the results have been published in arXiv: https://arxiv.org/pdf/2102.03557.pdf ）

The main discretization methods of traditional CFD include finite difference, finite element and finite volume method. The finite volume method is often used in commercial software and open source software because of its good conservation and grid adaptability. However, in the above methods, the large consumption of computing resources and computing time caused by the increase of the number of grids can not be avoided. The quantum finite volume method that can support classical data input and output proposed by the joint research team of Benyuan quantum and China University of science and technology in February this year is to solve the above problems. This method uses the quantum linear solver as a sub process, combined with the self-designed quantum random memory (QRAM) and quantum data structure to reduce the linear complexity of the classical algorithm to sub linear. "originQ QCFD" was developed based on this.

Using quantum linear solver instead of traditional solver, OriginQ QCFD can theoretically realize the sub linear acceleration of CFD simulation calculation based on the operation of quantum virtual machine (which can be directly connected to real quantum computer in the future). The quantum linear solver includes HHL and fom-hhl algorithms. In order to reduce the condition number of coefficient matrix and accelerate the calculation speed, a variety of preprocessors including static sparse approximate inverse are provided; The source quantity yu is suitable for solving incompressible and compressible flows, steady and unsteady flows.

OriginQ QCFD operation interface

(1) Four functions
■ quantum linear solver
■ multiple preprocessors
■ compressible and incompressible flow
■ steady flow and unsteady flow

(2) Advantages of quantum computing
The algorithm complexity of classical linear solver (conjugate gradient method) is directly proportional to matrix dimension N, but the time complexity of quantum algorithm (HHL) is directly proportional to the logarithm of matrix dimension n. Therefore, when the matrix dimension is large, the acceleration effect of quantum algorithm really appears.

Comparison of time complexity between classical conjugate gradient method (CG) and quantum HHL algorithm

OriginQ QCFD is a software that simulates the flow problem by using quantum calculation. It can numerically simulate the inviscid and laminar flow of single-phase flow. The software flow is mainly divided into three parts: pre-processing, calculation and post-processing.

The pre-processing module includes the functions of data import and export, simple structure grid generation, grid inspection and so on.

The calculation setting module mainly includes physical model selection, numerical method, control parameter setting function and so on.

Although the post-processing of calculation only supports scalar cloud images showing calculation results, the source volume supports exporting other data format files, such as VTK format, which is convenient for users to use other post-processing software for related operations.

OriginQ QCFD operation interface

Application example

As a classical hydrodynamics problem, the flow around a cylinder has rich experimental results, which are often used to verify the effectiveness of calculation methods and models; In addition, the study of the flow around a cylinder is also of great practical significance to the aviation field and marine engineering.

Under the same grid and configuration parameters, the classical solver and quantum solver are used to calculate the flow around a half cylinder; In order to facilitate analysis and comparison, the cloud images of the two calculation results are placed symmetrically, and the upper and lower parts correspond to the calculation results of quantum solver and classical solver respectively, as shown in Fig. 1 and Fig. 2; It can be seen from the figure that the flow field distribution of the two is basically the same, which shows the accuracy of quantum calculation.