一、本文先简单地介绍一下Green 函数,
第一部分内容来自于文献
[0]BI-GreenNet: Learning Green’s Functions by Boundary Integral Network
[1] Evans, L.C.: Partial Differential Equations. American Mathematical Society, Providence, R.I. (2010)
[2]Learning Elliptic Partial Differential Equations with Randomized Linear Algebra
[3] BINet: learning to solve partial differential equations with boundary integral networks
Interior problem(内点问题)
Exterior problem(外点问题)
此处
L
\mathcal{L}
L 是微分算子。为了方便,将方程(2.1)和(2.2)总结为:
格林函数
本文中, 我们关注泊松方程和亥姆霍兹方程, i.e., L = − Δ \mathcal{L}=-\Delta L=−Δ or L = − Δ − k 2 \mathcal{L}=-\Delta-k^2 L=−Δ−k2 , 此处 k k k是波数。
根据文献[1],我们可以使用如下的格林函数
G
:
Ω
×
Ω
⟶
R
+
∪
{
∞
}
G:\Omega\times \Omega\longrightarrow\mathbb{R}^{+}\cup\{\infty\}
G:Ω×Ω⟶R+∪{∞}, 得到方程 (2.3)的解析解
此处,
G
(
x
,
y
)
G(x,y)
G(x,y) 是一个二维函数,满足
当
g
=
0
g=0
g=0时,那么
(
f
,
u
)
(f,u)
(f,u)是个输入输出对,
f
f
f为源。
Seeking G, as opposed to L, has several theoretical benefits[2]:
第二部分内容来自:Learning Green’s functions associated with time-dependent partial differential equations
第三部分内容来自文献: MOD-Net: A Machine Learning Approach via Model-Operator-Data Network for Solving PDEs
第四部分内容来自文献:Neural Green’s function for Laplacian systems