solving pde related problems using deep-learning

31
Solving PDE related problems using deep-learning Adar Kahana Tel Aviv University Under the supervision of Eli Turkel (TAU) and Dan Givoli (Technion) Shai Dekel (TAU) Waves seminar, UC Merced February 4 th , 2021 1

Upload: others

Post on 16-Jan-2022

6 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Solving PDE related problems using deep-learning

Solving PDE related

problems using

deep-learning

Adar KahanaTel Aviv University

Under the supervision of

Eli Turkel (TAU) and

Dan Givoli (Technion)

Shai Dekel (TAU)

Waves seminar,

UC Merced

February 4th , 2021

1

Page 2: Solving PDE related problems using deep-learning

Agenda

Motivation

Data driven problems

Obstacle identification and deep-learning

Dealing with CFL instability using deep-learning

2

Page 3: Solving PDE related problems using deep-learning

Underwater acoustics

Sonar imaging

3

Page 4: Solving PDE related problems using deep-learning

The wave problem

4

แˆท๐‘ข ิฆ๐‘ฅ, ๐‘ก = ๐‘‘๐‘–๐‘ฃ ๐‘ ิฆ๐‘ฅ 2๐›ป๐‘ข ิฆ๐‘ฅ, ๐‘ก , ิฆ๐‘ฅ โˆˆ ฮฉ, t โˆˆ (0, ๐‘‡]

๐‘ข ิฆ๐‘ฅ, 0 = ๐‘ข0 ิฆ๐‘ฅ , ิฆ๐‘ฅ โˆˆ ฮฉ

แˆถ๐‘ข ิฆ๐‘ฅ, 0 = แˆถ๐‘ข0 ิฆ๐‘ฅ , ิฆ๐‘ฅ โˆˆ ฮฉ

๐‘ข ิฆ๐‘ฅ, ๐‘ก = ๐‘“ ิฆ๐‘ฅ, ๐‘ก , ิฆ๐‘ฅ โˆˆ ๐œ•ฮฉ1, t โˆˆ 0, ๐‘‡

๐›ป๐‘ข ิฆ๐‘ฅ, ๐‘ก = ๐‘” ิฆ๐‘ฅ, ๐‘ก , ิฆ๐‘ฅ โˆˆ ๐œ•ฮฉ2, t โˆˆ [0, ๐‘‡]

๐œ•ฮฉ1 โˆช ๐œ•ฮฉ2 = ๐œ•ฮฉ

where แˆถ๐‘ข0 ิฆ๐‘ฅ = 0 and ๐‘“ ิฆ๐‘ฅ, ๐‘ก = ๐‘” ิฆ๐‘ฅ, ๐‘ก = 0

Page 5: Solving PDE related problems using deep-learning

Ill-posed problems

In an experiment, we store the pressure at a small

number of sensors for all time steps

We wish to find the properties of the source or

obstacle from the data stored at these sensors

where the number of sensors << mesh

This is an inverse problem which is highly ill-posed

Hence, one cannot usually reconstruct the initial

conditions perfectly

Can we solve these types of ill-posed problems

with learning?

5

Page 6: Solving PDE related problems using deep-learning

Partial information โ€œRecordingโ€ the solution at a small set of

sensors placed in the domain ิฆ๐‘ฅs๐‘› ๐‘›=1

๐พโˆˆ ฮฉ

Data โ€“

๐‘ข ิฆ๐‘ฅs1 , t

๐‘ข ิฆ๐‘ฅs2 , t

โ‹ฎ๐‘ข ิฆ๐‘ฅsn , t

+ ๐’ฉ ๐œ‡, ๐œŽ2

The ill-posedness raises sensitivity to noise at

the sensors

6

Page 7: Solving PDE related problems using deep-learning

Agenda

Motivation

Data driven problems

Obstacle identification and deep-learning

Dealing with CFL instability using deep-learning

7

Page 8: Solving PDE related problems using deep-learning

Data driven problems

Supervised learning

Input data

Output labels

Training

Prediction (testing)

Drawback - sensitivity

8

Page 9: Solving PDE related problems using deep-learning

Deep-learning

Training โ€œweightsโ€ to learn connections in the data

Hidden multi-dimensional embeddings

Convolutions, Fully connected

โ€œDeepโ€ and non-linear

Loss

9

1 0 10 1 01 0 1

โ†’

๐‘ค1 ๐‘ค2 ๐‘ค3

๐‘ค4 ๐‘ค5 ๐‘ค6

๐‘ค7 ๐‘ค8 ๐‘ค9

Page 10: Solving PDE related problems using deep-learning

Physically-informed NN

Input: set of points from the initial and boundary

conditions

Output: solution in the domain

Loss: the problem

10

M. Raissi, P. Perdikaris, G.E. Karniadakis, Journal of computational Physics, 2018

Page 11: Solving PDE related problems using deep-learning

Results

11

Page 12: Solving PDE related problems using deep-learning

Agenda

Motivation

Data driven problems

Obstacle identification and deep-learning

Dealing with CFL instability using deep-learning

12

Page 13: Solving PDE related problems using deep-learning

Problem definitionGiven the position of the source/s and data at a few

sensors but many time slices find the location, size and

shape of the unknown scatterers

Input: Sensors recordings (๐‘๐‘ ๐‘Ž๐‘š๐‘๐‘™๐‘’๐‘  ร— ๐‘๐‘ก๐‘ ๐‘ก๐‘’๐‘๐‘  ร— ๐‘๐‘ ๐‘’๐‘›๐‘ ๐‘œ๐‘Ÿ๐‘ )

Output: Obstacle?

13

Page 14: Solving PDE related problems using deep-learning

Prior work

Location ิฆ๐‘ฅ

Shape and size โ€“

Circles: Radius

Rectangles: Height and Width

Complex shapes: need to be parametrized

โ€œSoftโ€ obstacles โ€“

Semi-penetrable

Multiphysics

14

Page 15: Solving PDE related problems using deep-learning

Labels solution - segments

15

Labels are ๐‘š ร— ๐‘›binary matrices

Predictions will be

๐‘š ร— ๐‘› probability

matrices

Loss: NLL

Page 16: Solving PDE related problems using deep-learning

Spatio-temporal architecture

16

Page 17: Solving PDE related problems using deep-learning

Loss diagram

17

Probability

map

Page 18: Solving PDE related problems using deep-learning

Physically informed loss

Using the segmentation network and output เทจ๐‘‚

Define a loss component based on:

Solve: ๐‘ข๐‘ก๐‘ก = 1 โˆ’ เทจ๐‘‚ ๐‘ฅ ๐‘2(๐‘ฅ) ฮ”๐‘ข

Get sensor data: ๐‘ข๐‘˜ ๐‘ฅ๐‘ ๐‘– ๐‘–=1

#๐‘ ๐‘’๐‘›๐‘ ๐‘œ๐‘Ÿ๐‘ for each sample

Calculate MSE between ground truth

๐‘ข๐‘˜ ๐‘ฅ๐‘ ๐‘– ๐‘–=1

#๐‘ ๐‘’๐‘›๐‘ ๐‘œ๐‘Ÿ๐‘ and the prediction as component ๐‘™2

Define the loss function for our network as:

๐›ผ โ‹… ๐‘™1 + 1 โˆ’ ๐›ผ โ‹… ๐‘™2such that ๐‘™1 is the NLL loss described earlier

18

Page 19: Solving PDE related problems using deep-learning

Numerical experiments Dirichlet BC

Compact Gaussian initial condition

Arbitrary polygonal obstacles โ€“

Generate number of edges

Generate edge length and angle

Generate location (๐‘ฅ0, ๐‘ฆ0, ๐‘ง0)

โ– Enormous samples space

Generated only 25,000 samples

19

Page 20: Solving PDE related problems using deep-learning

Probability images

20

Page 21: Solving PDE related problems using deep-learning

Neural network โ€“ results Intersection over union: 0 โ‰ค ๐ผ๐‘‚๐‘ˆ ๐ด, ๐ต =

๐ดโˆฉ๐ต

๐ดโˆช๐ตโ‰ค 1

Up to 66% IOU score

21

A.K., E. Turkel, D. Givoli, S. Dekel, Journal of computational Physics, 2020

Page 22: Solving PDE related problems using deep-learning

Agenda

Motivation

Data driven problems

Obstacle identification and deep-learning

Dealing with CFL instability using deep-learning

22

Page 23: Solving PDE related problems using deep-learning

Explicit schemes and CFL One-dimensional wave equation

CFL condition for stability: ๐›ผ =๐‘ฮ”๐‘ก

ฮ”๐‘ฅโ‰ค 1

FDCD: ๐‘ข๐‘–๐‘›+1 = 2๐‘ข๐‘–

๐‘› โˆ’ ๐‘ข๐‘–๐‘›โˆ’1 + ๐›ผ2 ๐‘ข๐‘–+1

๐‘› โˆ’ 2๐‘ข๐‘–๐‘› + ๐‘ข๐‘–โˆ’1

๐‘›

23

Page 24: Solving PDE related problems using deep-learning

Network architecture

Input: ๐‘ข ๐‘›โˆ’1 ๐‘š, ๐‘ข๐‘›๐‘š

Output: ๐‘ข ๐‘›+1 ๐‘š

Spatio-temporal architecture

Non-linear activation (PReLU)

Loss: MSE between ๐‘ข ๐‘›+1 ๐‘š and เทœ๐‘ข ๐‘›+1 ๐‘š

24

Page 25: Solving PDE related problems using deep-learning

Network diagram

25

Page 26: Solving PDE related problems using deep-learning

Physics informed loss

Use ๐‘ข ๐‘›โˆ’1 ๐‘š, ๐‘ข๐‘›๐‘š to predict เทœ๐‘ข ๐‘›+1 ๐‘š

Inside the loss:

Use ๐‘ข ๐‘›+1 ๐‘š, ๐‘ข ๐‘›+1 ๐‘š+1 to calculate ๐‘ข ๐‘›+1 ๐‘š+๐‘—

Use เทœ๐‘ข ๐‘›+1 ๐‘š, ๐‘ข ๐‘›+1 ๐‘š+1 to predict เทœ๐‘ข ๐‘›+1 ๐‘š+๐‘—

Calculate the MSE between ๐‘ข ๐‘›+1 ๐‘š+๐‘— and เทœ๐‘ข ๐‘›+1 ๐‘š+๐‘—

Network loss is the linear combination of the two

MSE losses

26

Page 27: Solving PDE related problems using deep-learning

Numerical experiments Dirichlet BC

Data:

Linear combinations with random coefficients

created from the basis sin ๐œ‹๐‘˜๐‘ฅ ๐‘˜=120

1250 different initial condition and 397 time-steps for

each one, total of 496,250 samples

Samples created with CFL = 0.875 and only each

10th sample was taken to get CFL = 8.75

27

Page 28: Solving PDE related problems using deep-learning

Results

28

Page 29: Solving PDE related problems using deep-learning

Results29

O. Ovadia, A. K, E. Turkel, S. Dekel, Journal of computational physics, submitted

Page 30: Solving PDE related problems using deep-learning

Summary and future work

Obstacle location and identification

Investigating source location

High measurement noise

Stability

Extending to 2,3 dimensions

Dispersion relation problem โ€“ optimized kernels

Experimental data

30

Page 31: Solving PDE related problems using deep-learning

Thanks!

31