Ge 193
Special Topics in Geophysics

Course Description

Topics

Practicals

Homework

Reading assignments

Software & Scripts

Bibliography

Instructor

Course Archives (Current & Previous Years)


Ge 193: An introduction to Geophysical Inverse Problems

Description

The purpose of the course is to introduce students to the principles and techniques of inversion and data inference. To give them an understanding of the underlying theory and an ability to recognize and tackle inverse problems in the geosciences or more generally. Areas to be covered include: The theory behind Linear, Linearized and Nonlinear inverse problems; Techniques for fitting models to data; Ill-posed inverse problems and methods of solution; Making sense of a solution: resolution, error propagation and goodness of fit. Nonlinear optimization and algorithms; Probabilistic inference theory and techniques; Fully nonlinear inversion: Monte Carlo based approaches to exploring parameter spaces.

The lectures will be supported by numerous examples and most likely MATLAB based exercises. Prerequisites: The course is aimed primarily at students with background in Physics and Mathematics, although some background mathematical material will be covered where necessary. Exposure to university level mathematics, particularly linear algebra and vector calculus will be an advantage.

Units: 3-0-6 (first term)
Regular class time: Tuesday and Thursday 1 PM – 2:30 PM, 162 S. Mudd
Grading policy: 3-4 homeworks (60%), 2-3 tutorials (30%), Class participation (i.e. discussion, answering and asking questions) (10%). No final exam.


Topics

Background material for this class is available HERE.

COMPLETE LECTURE NOTES for this class are available HERE [42 MB PPT File!].

Lecture

Day

Topic(s)

Reading

1

10/02

Class introduction and overview [PPT] [PDF]

-

2

10/04

Linear Discrete Inverse Problems [PPT] [PDF]

-

3

10/09

NO CLASS - But see Reading Assignment.

-

4

10/11

NO CLASS - But see Reading Assignment.

-

6

10/16

Linear Discrete Inverse Problems CONTINUED [see Lecture 2]

-

7

10/18

PRACTICAL 1 (Bring your laptops to class!)

-

8

10/23

Linearized Inverse Problems [PDF]

-

9

10/25

Singular Value Decomposition - I [PDF]

-

10

10/30

Singular Value Decomposition - II [PDF]

-

11

11/01

Regularizing Inverse Problems [PDF (up to p.160)]

-

12

11/06

Cross-Validation [PDF (p.160 onwards)]

-

13

11/08

Practicals 2: SVD (See below)

-

14

11/13

NO CLASS (Tectonics Observatory Meeting)

-

15

11/15

Iterative Methods [PDF]          [Buffon's Needle Problem 1, 2]

-

16

11/20

Iterative Methods - Continued

-

17

11/22

NO CLASS - Thanksgiving

-

18

11/27

Iterative Methods - Continued
Probabilistic Inference [PDF]

-

19

11/29

Probabilistic Inference - Continued

-

20

12/04

Probabilistic Inference - Continued

-

21

12/06

CLASS PRESENTATIONS

Complete Class Notes


Practicals

Practicals will be held two or three times during the quarter, on dates that will be decided in class. Additional software and tutorials useful for the practicals are HERE.

#

Date

Topic(s)

Links

1

10/17
(REGULAR CLASS TIME)

Linear Inverse Problems

Practical 1: Seismic Profiling

Additional File 1: LOCTIM.TXT

Additional File 2: BORDER.XY

2

10/24
(3:30 PM)

Linear Inverse Problems (CONTINUED)

See Practical 1 above

3

11/08
(REGULAR CLASS TIME)

Ill-conditioned problems: SVD

Practical 2: Singular Value Decomposition

Additional File 1: BOOKFONTS.M

Additional File 2: PLOTCONST.M

Additional File 3: SHAW.M

4

11/28
(4:00 PM)

Exploring Frequentist vs. Bayesian Approaches

Lighthouse Problem


Homework

Late homeworks will be penalized 10% per day, unless there is a clear problem with any given assignment or there is some kind of real emergency. Discussing homeworks with fellow students is fine, but what is turned in should represent your work, not a group effort. Please do not request extensions on the same day the homework is due. Software and tutorials useful for completing your homeworks are HERE.

#

Date Due (@BOC)

Topic(s)

Homework Links

1

11/01

Linear Inverse Problems

HW 1
Solutions





*BOC = beginning of class


Reading assignments

As part of this class, we will read selected important papers/textbook chapters. The reading material will be handed out during the class or available as PDFs from the class website.

Your paper summaries/comparisons should be two pages long, max - UNLESS OTHERWISE STATED BELOW. In general, please treat this assignment like any other scientific writing (i.e., papers, reviews, and proposals) and conform to a format with single or double-spaced lines, 1 inch margins, and 12 pt font. Use a concise style of prose (but do not slip into outline form). Choose an active style over a passive style, avoid chatty colloquialisms (this is not a diary), watch that you have some kind of logical order of presentation, and be precise with your language.


#

Date Due (@BOC)

Papers

Notes

1

10/18

(1) Aster et al., (2003), Chapters 1-2

&

(2) Parker (1977) OR
      Aster et al., (2003), Chapter 3


For (1) - Writing: 2-page summary of Chapter 2 (follow guidelines above)

For (2) - Reading: The point is to have a basic understanding of the Backus and Gilbert theory
for continuous linear inverse problems. So, reading either reference is fine.

2

11/29

(1) Aster et al., (2003), Chapter 6

&

(2) Rawlinson & Sambridge (2003)

Reading: Chapter 6 of (1) and Pages 33-45 of (2)

Writing: CRITICAL comparision of any THREE different iterative optimization algortihms from the above reading. Your critique must be detailed, and AT LEAST 3 PAGES long.

*BOC = beginning of class


Bibliography

All textbooks below should be available in the Geology Library Reference by October 8th.



Instructor

Teaching Assistant

Ombudsperson


Last updated by Ravi Kanda in December, 2007