CS 621: Fall 2005

| News | General Information | Syllabus | Assignment and Solutions

|Exams and Solutions |Book Errata | CVL Home | Problem of the Day |

 

News 

Dec 14. Final exam pick-up in 4130 Upson, 9-4pm. Nora or Bonnie can help you.

Dec 13. The final exam grading guide is available. Send me an email if you'd like your course grade. Pick up instructions will be posted soon.

Dec 12. The final is today, 9-11:30, in Olin 218.

Dec 7  I am not quite done grading Assignment 6, but I have put up solutions to the first two problems. I'll be all done grading this later today so they will be ready for pick-up starting Thursday.

Dec 6 Office Hours this week: WThF 2:30-4. Closed book final exam: Monday, Dec 12 9:00 -  11:30, Olin 218.

Nov 23 Class will meet today in 4130 Upson if you feel like it!

Nov 22  Assignment 5 solutions

Nov 22 Assignment 6 handout and test scripts are available. Slight change regarding Problem 1. I am providing a Test script. So INSTEAD of submitting your implementation electronically, run the test script and submit hardcopy of the output. The rest of the problem stays the same. There is also some confusion concerning U. The story is that the given T =  U'*A*U for some A. You then go on to find a different U that reverses the order of the two bumps in the given T:   S = U'*T*U. The problem is really about computing an orthonormal basis for the invariant subspace associated with the eigenvalues of T(3:4,3:4).

Nov 11 Assignment 4 solutions.

Nov 8 Assignment 5 test scripts are available.

Nov 7. Assignment 5 is available.

Nov 3. For problem 3, ignore the "not OKs' when n2 = 1. (That is because when I then check diag(S), diag returns a matrix and not the vector of diagonal elements. And when n2<p, you have p-n2 c's that are exactly 1. If you set these to one, no problem. But if you compute them, then some problems arise when i check the ordering. Moral: ignore the OK/notOK comments.

Nov 3 Matlab's QR may not give you positive R(i,i). But if you multiply a row in your R by -1 and the corresponding column in Q by -1, we will still have A = QR.

Oct 27 The condition estimator function MyCond is now available.

Oct 25 Assignment 4 test scripts are available. In these problems you are to make use of the Matlab functions svd and qr. These functions can be used in different ways; be efficient. For example if A is m-by-n then qr(A) is O(m^2n)  while qr(A,0) is O(mn^2). That's a big difference if m>>n.

Oct 24. Assignment 4 is available.

Oct 24 Solutions Assignment 3. You will find the solution to GTD3 and the O(n) solution to Problem 1 here. Take a look at the derivation of the O(n) solution.

Oct 21 Assignment 4 will be available Monday. A heads up about two of the problems in case you want to start working. (1) How would solve the TLS problem by simply using the QR factorization of [A b]? Hint. Think about condition estimation with the R matrix. (2) Look at section 4 of this brilliant paper:   

                       http://www.cs.cornell.edu/cv/ResearchPDF/Computing.Cs.Gen.Sing.Value.Decomp.pdf

 

Oct 19 Midterm solution.

Oct 14 Assignment 3 solutions.

Oct 13. The Friday 4pm due date for Assignment 3 is a hard deadline. Solutions to the two problems will be posted at that time so that you can study from them for the Monday Quiz. No late submissions.

Oct 12. There will be an in-class quiz Monday, October 17. All topics relevant to the first three assignments are fair game. Look at all Problems of the Day through Sept 30. A sample quiz is here.

Oct 6 Assignment 3 test scripts are here.

Oct 5  Solution scripts for Assignment 2 are available and so is a write-up of my method for the GTD problem. Regarding the grading of the latter, 2 points for the solution A0 = U*S*V'*Delta where Delta = diag(1,1,...,1,-1) and A = USV' is the SVD. Three points if you say something about it. 4 points if you go on to explain or show why this isn't optimal. 5 points if you have a method that produces the optimal A0.

Sept 30 Assignment 3 is here.

Sept 21 There was a typo in the P1 test script for Assignment 2. It has been corrected. Make sure you use the corrected script.

Sept 21 Assignment 1 solutions are available here. Some comments. Not enough attention was given to making informative plots. For example, since one expects FastProd to execute 25% faster than the naive algorithm, the plot should somehow comment on this expected result. Reporting raw times and expecting the reader to do the ratio is not good journalism!. Regarding the Kronecker product problems--never form an explicit Kronecker product. Always run our test scripts exactly. Never submit the output from a modified test script, unless it is accompanied by the results of our test scripts. For the GTD problems, I'm using a grading scale of zero-to-5. You are always welcome to discuss the grading of your assignments.

Sept 20 Assignment 2 test scripts available here.

Sept 6 If you order GVL from Amazon, then I am happy to reimburse you for tax and (regular) shipping. Just give me the receipt.

Sept 2 A1 handout is available. Test cripts P1 and P2 will be available next week.

Aug 29.  For Matlab review/reference, I recommend Getting Started with Matlab by Rudra Pratap. For linear algebra review/reference, some good books are mentioned at the top of page 52 in GVL.

General Information

Meeting Times: MWF 11:15-12:05, Upson 109

InstructorProfessor Charles Van Loan, 4130 Upson Hall, 255-7644, cv@cs.cornell.edu

Office Hours: Office Hours are posted here. There will be exceptions and I'll tell you in class. There will be extra times and I will tell you in class. And you can always schedule a special time by sending me email.

TA:   I am the TA !

Text: Matrix Computations, 3rd Edition by G.H. Golub and  C. Van Loan.

Grading: Six Matlab assignments (30%), three best Go-the-Distance problems (10%), Midterm (30%), and a Final Exam (30%).

 

Problem of the Day

Problems
Monday Wednesday Friday
 Aug 29  Aug 31  Sept 1
 Sept 5  Sept 7  Sept 9
 Sept 12  Sept 14  Sept 16
 Sept 19  Sept 21  Sept 23
 Sept 26  Sept 28  Sept 30
 Oct 3  Oct 5  Oct 7
 Break  Oct 12  Oct 14
 Oct 17  Oct 19  Oct 21
 Oct 24  Oct 26  Oct 28
 Oct 31  Nov 2  Nov 4
 Nov 7  Nov 9  Nov 11
 Nov 14  Nov 16  Nov 18
 Nov 21  Nov 23  Break
 Nov 28  Nov 30  Dec 2

 

 

Syllabus

Date

Topic

Reading

Events

Fri Aug 26 Fundamentals  Chapter 1  
Mon  Aug 29  "    
Wed  Aug 31  "    
Fri      Sept 2  "  Chapter 2 P1 Out
Mon  Sept 5  "    
Wed  Sept 7  "    
Fri    Sept  9   Ax = b              Chapter 3  
Mon  Sept 12 "    
Wed  Sept 14                P1 Due
Fri     Sept 16        "          P2 Out
Mon  Sept 19 Special Ax=b  Chapter 4  
Wed  Sept 21       "           
Fri   Sept 23 "    
Mon  Sept 26    "         
Wed  Sept 28 Least Squares  Chapter 5 P2 Due
Fri  Sept 30  "    P3 Out
Mon Oct 3      "        
Wed Oct 5  "    
Fri Oct 7  "    
Wed Oct 12  "    
Fri Oct 14 Constrained LS   12.1 P3 Due
Mon Oct 17 Total LS 12.3 Take-Home Midterm Out
Wed Oct 19 Subspaces,Updating 12.4,12.5  
Fri Oct 21  " 1  
Mon Oct 24 Symm Eigenprob 8.4 Midterm Due, P4 Out
Wed Oct 26 " 8.2, 8.1  
Fri  Oct 28  " 8.3  
Mon Oct 31  " 8.5  
Wed Nov 2  " 8.6  
Fri Nov 4 Unsymm Eigenprob 7.3 P4 due
Mon Nov 7 " 7.1,7.2 P5 Out
Wed Nov 9  " 7.4  
Fri Nov 11  " 7.5-7.6  
Mon Nov 14  " 11.1  
Wed Nov 16  " 11.2,11.3  
Fri Nov 18 Lanczos Methods 9.1  P5 Due
Mon Nov 21   " 9.2 P6 Out
Wed  Nov 23   Conjugate Gradients 10.1  
Mon  Nov 28   " 10.2  
Wed  Nov 30    " 10.3  
Fri   Dec 2    Review   P6 Due
       
Mon Dec 12 Final Exam   9:00 -  11:30, Olin 218

 

 Top

 

Assignments and Solutions

Assignment1        Handout Scripts Solution Go the Distance 1    Handout
Assignment 2       Handout Scripts Solution

GTD2

 
Assignment 3       Handout Scripts Solution  
Assignment 4       Handout Scripts Solution  
Assignment 5       Handout Scripts Solution  
Assignment 6       Handout Scripts Solution  

 

Top

Exams and Solutions

 

Midterm   Solution
Final   Solution

 Top

Courses

Requirements

Admin

Docs