趙麗志, 馮曉莉
(西安電子科技大學(xué) 數(shù)學(xué)與統(tǒng)計(jì)學(xué)院,西安 710126)
隨著科學(xué)技術(shù)的快速發(fā)展,反問(wèn)題在環(huán)境科學(xué)、能源開(kāi)發(fā)、流體力學(xué)、醫(yī)學(xué)、金融等領(lǐng)域有了越來(lái)越廣泛的應(yīng)用.所謂反問(wèn)題就是指用解的一些已知數(shù)據(jù)去重構(gòu)問(wèn)題中的未知數(shù)據(jù)[1-3].由于隨機(jī)偏微分方程在腦磁成像[4]、光聲成像[5]、超聲成像[6]、天線(xiàn)設(shè)計(jì)與合成[7-8]等方面有著重要的應(yīng)用,所以受到了廣大學(xué)者的關(guān)注.特別地,隨機(jī)反源問(wèn)題作為隨機(jī)反問(wèn)題中的一類(lèi),目前已有很多學(xué)者做了相關(guān)研究,比如文獻(xiàn)[9] 通過(guò)Carleman 估計(jì)證明了隨機(jī)源項(xiàng)的唯一性;文獻(xiàn)[10]研究了一維隨機(jī)熱方程模擬的無(wú)限桿中的反源問(wèn)題,并通過(guò)求解Fredholm 積分方程重構(gòu)了隨機(jī)源項(xiàng)的均值和方差.
近年來(lái),帶有Hurst 參數(shù)(H∈(0,1))的分?jǐn)?shù)階Brown 運(yùn)動(dòng)在科學(xué)和工程領(lǐng)域中有著廣泛的應(yīng)用,目前關(guān)于帶有不同類(lèi)型隨機(jī)源項(xiàng)的時(shí)間分?jǐn)?shù)階擴(kuò)散方程的反源問(wèn)題已有一些成果.對(duì)于H=1/2的情形,文獻(xiàn)[11]討論了帶有離散隨機(jī)噪聲的時(shí)間分?jǐn)?shù)階擴(kuò)散方程的反源問(wèn)題;文獻(xiàn)[12] 運(yùn)用終止時(shí)刻的數(shù)據(jù)u(x,T,ω)的統(tǒng)計(jì)信息確定了時(shí)間分?jǐn)?shù)階擴(kuò)散方程的源項(xiàng)f(x)h(t)+g(x)ω˙(t)中 的f(x)和 |g(x)|,更多相關(guān)研究可參考文獻(xiàn)[13-17]. 對(duì)于H∈(0,1)的 情形,文獻(xiàn)[18]研究了帶有f(x)h(t)+g(x)B˙H(t)隨機(jī)源項(xiàng)的時(shí)間分?jǐn)?shù)階擴(kuò)散方程,并根據(jù)終止時(shí)刻的數(shù)據(jù)u(x,T,ω)重 構(gòu)了f(x)和 |g(x)|;類(lèi)似地,文獻(xiàn)[19]考慮了另一個(gè)時(shí)間分?jǐn)?shù)階擴(kuò)散方程;文獻(xiàn)[20]為分?jǐn)?shù)階Gauss 噪聲驅(qū)動(dòng)下的隨機(jī)非線(xiàn)性分?jǐn)?shù)階擴(kuò)散方程的數(shù)值分析提供了一個(gè)統(tǒng)一的框架.目前,由于對(duì)帶有分?jǐn)?shù)階Brown 運(yùn)動(dòng)的隨機(jī)偏微分方程的討論還處于研究初期,并且關(guān)于對(duì)流擴(kuò)散方程還沒(méi)有相關(guān)的研究,因此,本文將討論如下由分?jǐn)?shù)階Brown 運(yùn)動(dòng)驅(qū)動(dòng)的隨機(jī)對(duì)流擴(kuò)散方程:
本節(jié)將在以下假設(shè)成立的條件下討論問(wèn)題(1)的適定性.
假設(shè)1 設(shè)H∈(0,1)并 且f, σ,g∈L?2(D). 假設(shè)h∈L∞(0,T)是一個(gè)非負(fù)函數(shù)并且有一個(gè)正的下界,即h≥Ch>0.
由式 (7)得
所以
證明 根據(jù)式(12)、(20)和(22)易得結(jié)論.
本節(jié)將通過(guò)終止時(shí)刻的數(shù)據(jù)u(x,T,ω)的 一些統(tǒng)計(jì)量來(lái)重構(gòu)源項(xiàng)中的f(x)和 σ2(x),并且分別討論它們的唯一性與不穩(wěn)定性,其中
由式(7)可得
3.2.1 反演f(x)的不穩(wěn)定性
由積分中值定理,可知
圖1 H =0.9,δ=0.04 時(shí) f 和σ 2的相對(duì)誤差Fig. 1 The relative errors of the reconstruction for f and σ 2 with respect to H =0.9 andδ=0.04
圖2 H =0.4,N=6 時(shí)的 f 和σ2Fig. 2 The reconstruction of f and σ 2 for the inverse problem with H =0.4 andN=6
圖3 H =0.5,N=6 時(shí)的 f 和σ2Fig. 3 The reconstruction of f and σ 2 for the inverse problem with H =0.5 andN=6
圖4 H =0.9,N=6 時(shí)的 f 和σ2Fig. 4 The reconstruction of f and σ 2 for the inverse problem with H =0.9 andN=6
在本文中,我們討論了帶有分?jǐn)?shù)階Brown 運(yùn)動(dòng)隨機(jī)源項(xiàng)的一維隨機(jī)對(duì)流擴(kuò)散方程.在正問(wèn)題部分通過(guò)對(duì)溫和解的期望的討論,證明了其適定性.在反隨機(jī)源部分,給定T時(shí)刻的數(shù)據(jù)來(lái)反演源項(xiàng),證明了反演的唯一性與不穩(wěn)定性.最后通過(guò)有限差分法和截?cái)嗾齽t化方法進(jìn)行數(shù)值模擬證明了理論部分的合理性,并且得出了Hurst 參數(shù)H越大,反演效果越好的結(jié)論.雖然本文只討論了一維隨機(jī)對(duì)流擴(kuò)散方程,但是關(guān)于高維的情形也可以類(lèi)似討論.
參考文獻(xiàn)( References ) :
[1] 李曉曉, 郭亨貞, 萬(wàn)詩(shī)敏, 等. 一類(lèi)對(duì)流-擴(kuò)散方程熱源識(shí)別反問(wèn)題[J]. 蘭州理工大學(xué)學(xué)報(bào), 2012, 38(3): 147-149. (LI Xiaoxiao, GUO Hengzhen, WAN Shimin, et al. A class of inverse problem of identification of heat source term in convection-diffusion equation[J].Journal of Lanzhou University of Technology, 2012, 38(3) : 147-149.( in Chinese))
[2] 賈現(xiàn)正, 張大利, 李功勝, 等. 空間-時(shí)間分?jǐn)?shù)階變系數(shù)對(duì)流擴(kuò)散方程微分階數(shù)的數(shù)值反演[J]. 計(jì)算數(shù)學(xué), 2014, 36(2):113-132. (JIA Xianzheng, ZHANG Dali, LI Gongsheng, et al. Numerical inversion of the fractional orders in the space-time fractional advection-diffusion equation with variable coefficients[J].Mathematica Numerica Sinica,2014, 36(2): 113-132.(in Chinese))
[3]FURATI K M, IYIOLA O S, KIRANE M. An inverse problem for a generalized fractional diffusion[J].Applied Mathmatics and Computation, 2014, 249: 24-31.
[4]AMMARI H, BAO G, FLEMING J L. An inverse source problem for Maxwell’s equations in magnetoencephalography[J].SIAM Journal on Applied Mathematics, 2002, 62(4): 1369-1382.
[5]ANASTASIO M A, ZHANG J, MODGIL D, et al. Application of inverse source concepts to photoacoustic tomography[J].Inverse Problems, 2007, 23(6): S21-S35.
[6]DEVANEY A J. Inverse source and scattering problems in ultrasonics[J].IEEE Transaction on Sonics and Ultrasonics, 1983, 30(6): 355-363.
[7]MARENGO E A, DEVANEY A J. The inverse source problem of electromagnetics: linear inversion formulation and minimum energy solution[J].IEEE Transaction on Antennas and Propagation, 1999, 47(2): 410-412.
[8]MARENGO E A, KHODJA M R, BOUCHERIF A. Inverse source problem in nonhomogeneous background media, Ⅱ: vector formulation and antenna substrate performance characterization[J].SIAM Journal on Applied Mathematics, 2008, 69(1): 81-110.
[9]Lü Q. Carleman estimate for stochastic parabolic equations and inverse stochastic parabolic problems[J].Inverse Problems, 2012, 28(4): 045008.
[10]CHEN S L, WANG Z W, CHEN G L. Cauchy problem of non-homogenous stochastic heat equation and application to inverse random source problem[J].Inverse Problems and Imaging, 2021, 15(4): 619-639.
[11]TUAN N H, NANE E. Inverse source problem for time-fractional diffusion with discrete random noise[J].Statistics and Probability Letters, 2017, 120: 126-134.
[12]NIU P P, HELIN T, ZHANG Z D. An inverse random source problem in a stochastic fractional diffusion equation[J].Inverse Problems, 2020, 36(4): 045002.
[13]LIU C. Reconstruction of the time-dependent source term in stochastic fractional diffusion equation[J].Inverse Problems and Imaging, 2020, 14(6): 1001-1024.
[14]FU S B, ZHANG Z D. Application of the generalized multiscale finite element method in an inverse random source problem[J].Journal of Computational Physics, 2021, 429: 110032.
[15]GONG Y X, LI P J, WANG X, et al. Numerical solution of an inverse random source problem for the time fractional diffusion equation via PhaseLift[J].Inverse Problems, 2021, 37: 045001.
[16]LI P J, WANG X. An inverse random source problem for Maxwell’s equations[J].SIAM Journal on Multiscale Modeling and Simulation, 2021, 19(1): 25-45.
[17]LI P J, WANG X. An inverse random source problem for the one-dimensional Helmholtz equation with attenuation[J].Inverse Problems, 2021, 37(1): 015009.
[18]FENG X L, LI P J, WANG X. An inverse random source problem for the time fractional diffusion equation driven by a fractional Brownian motion[J].Inverse Problems, 2020, 36(4): 045008.
[19]NIE D X, DENG W H. An inverse random source problem for the time-space fractional diffusion equation driven by fractional Brownian motion[EB/OL]. (2021-06-02)[2022-02-12]. https://arxiv.org/abs/2106.00917.
[20]NIE D X, DENG W H. A unified convergence analysis for the fractional diffusion equation driven by fractional Gaussion noise with Hurst indexH∈(0, 1)[J].SIAM Journal on Numerical Analysis, 2022, 60(3): 1548-1573.