孫美玲
小參數(shù)對(duì)流擴(kuò)散方程在最優(yōu)分層網(wǎng)格的一致收斂有限元計(jì)算
孫美玲
(南通職業(yè)大學(xué) 公共教學(xué)部,江蘇 南通 226007)
面向小參數(shù)的奇異攝動(dòng)對(duì)流擴(kuò)散方程,構(gòu)建分層網(wǎng)格自適應(yīng)地刻畫(huà)邊界層對(duì)應(yīng)的離散結(jié)點(diǎn),應(yīng)用有限元計(jì)算以期在特殊網(wǎng)格上得到優(yōu)化結(jié)果.分層網(wǎng)格無(wú)需復(fù)雜計(jì)算,僅根據(jù)遞推關(guān)系可形成隨機(jī)剖分?jǐn)?shù)的優(yōu)化網(wǎng)格,實(shí)現(xiàn)更好地捕捉邊界層.?dāng)?shù)值算例驗(yàn)證了方法的魯棒性,獲得了完全獨(dú)立于小攝動(dòng)參數(shù)、一致收斂的有限元高精度數(shù)值結(jié)果.
對(duì)流擴(kuò)散方程;小參數(shù);分層網(wǎng)格;有限元法;一致收斂
奇異攝動(dòng)問(wèn)題是當(dāng)前科學(xué)工程計(jì)算的熱點(diǎn),廣泛地出現(xiàn)在彈性力學(xué)、流體計(jì)算、生物系統(tǒng)和運(yùn)籌控制等眾多領(lǐng)域.在進(jìn)行數(shù)學(xué)物理方程描述時(shí),小參數(shù)會(huì)作為系數(shù)出現(xiàn)在方程的最高階導(dǎo)數(shù)項(xiàng),精確有效地求解奇異攝動(dòng)問(wèn)題漸進(jìn)解或近似解一直是相關(guān)研究的目標(biāo)[1-2].但如果常規(guī)處理簡(jiǎn)單地將小參數(shù)設(shè)為零,將導(dǎo)致原方程降階從而不能得到所有的近似解.奇異攝動(dòng)來(lái)源于問(wèn)題存在多個(gè)尺度,為了得到每個(gè)尺度的有效近似解,可以將方程用不同尺度分解以得到新方程.
本文考慮一維奇異攝動(dòng)問(wèn)題
本文研究含小參數(shù)問(wèn)題(1)的精確高效解,應(yīng)用有限元計(jì)算,結(jié)合自適應(yīng)生成的分層網(wǎng)格精確模擬解的邊界層部分和光滑部分,得到與真解相比精度非常高、完全獨(dú)立于攝動(dòng)系數(shù)大小的收斂解.
將每一單元形成的單元?jiǎng)偠染仃嚴(yán)奂拥剿袉卧纬煽倓偠染仃?,求解?duì)應(yīng)的大規(guī)模稀疏線性方程組,即得到有限元數(shù)值解.之后還可以對(duì)計(jì)算結(jié)果進(jìn)行誤差分析與后處理,通過(guò)列表和畫(huà)圖判斷數(shù)值結(jié)果的精確程度與收斂性,進(jìn)一步提升方法的校正和改進(jìn)性能.
這種分片等距Shishkin網(wǎng)格較之前的分段等距一致網(wǎng)格,能較好地捕捉解的部分奇性.不過(guò)Shishkin網(wǎng)格也存在一個(gè)缺點(diǎn),即在某些情形當(dāng)剖分?jǐn)?shù)為奇數(shù)時(shí),可能造成不穩(wěn)定.
基于一致Uniform網(wǎng)格和Shishkin網(wǎng)格存在的不足,本文給出分層Graded網(wǎng)格(簡(jiǎn)記G網(wǎng)格)算法.
表1 有限元計(jì)算在3種網(wǎng)格得到的最大模誤差與收斂階
圖1 真解與數(shù)值解的對(duì)比
本文所提方法能正確高效地模擬小參數(shù)引起的奇異攝動(dòng)邊界層問(wèn)題,采用分層網(wǎng)格的自適應(yīng)生成方案并運(yùn)用有限元計(jì)算,獲得了完全不依賴于參數(shù)大小且能保證高精度與快收斂的數(shù)值模擬結(jié)果.
[1] Miller J J,Oriordan E,Shishkin G I.Fitted numerical methods for singular perturbation problems[M].Singapore:World Scientific,2012
[2] 張偉江,周明儒,林武忠,等.奇異攝動(dòng)導(dǎo)論[M].北京:科學(xué)出版社,2014
[3] Reddy J N.An introduction to the finite element method[M].3th ed.New York:McGraw Hill,2006
[4] Roos H G,TeofanovL,Uzelac Z.Graded meshes for higher order FEM[J].J Comput Math,2015,33(1):1-16
[5] 王建云,田智鯤.定常非線性薛定諤方程的有限元方法超收斂估計(jì)[J].湘潭大學(xué)自然科學(xué)學(xué)報(bào),2018,40(1):24-26
[6] Qiu C X,Zhao W J,Song L N.A balanced finite element method of least-squares formulation for singularly perturbed reaction-diffusion problems[J].Chinese J Engineer Math,2016,33(3):309-318
[7] Jiang S,Presho M,Huang Y Q.An adapted Petrov-Galerkin multiscale finite element method for singularly perturbed reaction-diffusion problems[J].Inter J Comput Math,2016,93(7):1200-1211
[8] Jiang S,Sun M L,Yang Y.Reduced multiscale computation on adapted grid for the convection-diffusion Robin problem[J].J Appl Anal Comput,2017,7(4):1488-1502
[9] Chen C J,Chen Y P,Zhao X.A posteriori error estimates of two-grid finite volume element methods for nonlinear elliptic problems[J].Comput Math Appl,2018,75:1756-1766
[10] 尹云輝,祝鵬,楊宇博.流線擴(kuò)散有限元方法在分層網(wǎng)格上的收斂性分析[J].計(jì)算數(shù)學(xué),2015,37(1): 83-91
Uniformly convergent finite element computation on optimally graded meshes for the convection-diffusion equation with a small parameter
SUN Meiling
(Department of Public Course,Nantong Vocational University,Nantong 226007,China)
As for a singularly perturbed convection-diffusion equation with a small parameter,a graded mesh is built adaptively to describe the discrete nodes of the boundary layers,and the finite element computation is applied to solve the optimal results on special meshes.The graded mesh is free of complicated operations,it can be realized from a recursive formulation to build the optimal mesh with random partition numbers,which is qualified to capture the boundary layers.The robustness of the method is shown through numerical experiments,the high precision numerical results of finite element are obtained,which are completely independent of small perturbation parameters and uniformly convergent.
convection-diffusion equation;small parameter;graded mesh;finite element method;uniformly convergent
O241.81
A
10.3969/j.issn.1007-9831.2020.06.001
1007-9831(2020)06-0001-04
2020-03-19
國(guó)家自然科學(xué)基金面上項(xiàng)目(11771224);南通職業(yè)大學(xué)自然科學(xué)研究項(xiàng)目(1512105)
孫美玲(1981-),女,江蘇南通人,副教授,博士,從事微分方程數(shù)值解及其應(yīng)用研究.E-mail:sunmeiling81@163.com