Deep Neural Operator
1
2
Burgers’ Equation
Korteveg-de Vries Equation
Kuramoto-Sivashinsky Equation
Navier-Stokes Equation
Limitation of PINN: Single Instance Solver
Euler Beam
4
Euler Beam
5
Euler Beam
6
function
function
Transition to Deep Neural Operators
7
What is Operator?
8
Can Neural Networks Learn Operators?
9
Neural Operators
10
Neural Operators
11
DeepONet
Multiscale DeepONet
MIONet
PI-DeepONet
Fourier-DeepONet
FNO
AFNO
MFNO
FNO-Attention
GFNO
Fourier Neural Operator
Deep Operator Network
Deep Operator Network�(DeepONet)
12
Basic Concept of DeepONet (1/3)
13
Basic Concept of DeepONet (2/3)
14
Basic Concept of DeepONet (3/3)
15
Euler Beam
16
From Function to Operator
17
Input vector
Output vector
From Function to Operator
18
From Function to Operator
19
From Function to Operator
20
From Function to Operator
21
From Function to Operator
22
From Function to Operator
23
DeepONet
24
DeepONet as Basis and Coefficients Representation
25
DeepONet as Basis and Coefficients Representation
26
Slide from Prof. Taejin Kim (Jeonbuk National University)
DeepONet as Basis and Coefficients Representation
27
Slide from Prof. Taejin Kim (Jeonbuk National University)
DeepONet as Basis and Coefficients Representation
28
Slide from Prof. Taejin Kim (Jeonbuk National University)
DeepONet as Basis and Coefficients Representation
29
Slide from Prof. Taejin Kim (Jeonbuk National University)
DeepONet as Basis and Coefficients Representation
30
Slide from Prof. Taejin Kim (Jeonbuk National University)
DeepONet as Basis and Coefficients Representation
31
Slide from Prof. Taejin Kim (Jeonbuk National University)
DeepONet as Basis and Coefficients Representation
32
Slide from Prof. Taejin Kim (Jeonbuk National University)
DeepONet is Fundamentally Data-Driven
33
Beyond Data-Driven: Physics-Informed DeepONet
34
Physics-informed DeepONet (PI-DeepONet)
35
Physics-informed DeepONet (PI-DeepONet)
36
Scientific Machine Learning (1/3)
37
PINNs
Physics-informed DeepONet
Deep Operator Network
Inputs | Neural Network | Output |
| Single network | |
Scientific Machine Learning (2/3)
38
PINNs
Physics-informed DeepONet
Deep Operator Network
Inputs | Neural Network | Output |
PDE conditions | Branch net | Basis coefficients |
Domain | Trunk net | Basis functions |
Scientific Machine Learning (3/3)
39
PINNs
Physics-informed DeepONet
Deep Operator Network
Inputs | Neural Network | Output |
PDE conditions | Branch net | Basis coefficients |
Domain | Trunk net | Basis functions |
Lab 1: Heat Transfer in 1D
40
Heat Transfer in 1D
41
Note
42
Heat Transfer in 1D
43
Heat Transfer in 1D
44
Heat Transfer in 1D
45
Heat Transfer in 1D
46
Heat Transfer in 1D
47
Heat Transfer in 1D
48
Training Data Generation
49
BranchNet and TruckNet
50
Result for Test
51
Lab 2: Euler Beam
52
Lab 2: Euler Beam
53
Input-Output Data Pairs
54
Result
55
Lab 3: Transient Heat Conduction
56
Problem Setup
57
Training Data Sets and DeepONet Structure
58
Three Initial Conditions
One Initial Condition Diffusing
Masking Function
59
Result
60
Result
61
Lab 4: A Vibrating String
62
Problem Setup
63
Training Data Sets and DeepONet Structure
64
One Initial Shape Vibrating
Three Initial Shapes
Masking Function and Fourier Features in Time
65
Result
66
Result
67
Adding the Physics: PI-DeepONet
68
Result
69
PI-DeepONet
70
Summary
71