ELEMENTARY NUMERICAL ANALYSIS-22M:072/22C:036
FALL SEMESTER 1998
FINAL EXAMINATION
Tuesday, December 15, 7:00-9:00 PM in Room 158 Van Allen.
The final test will be mainly on Chapters 8 and 9,
with some other questions from other chapters, so
consider it as being comprehensive. The test will be done
with open books and open notes. Bring a scientific calculator.
Only under exceptional circumstances will a student be permitted
to shift the time of this examination.
GENERAL INFORMATION
MATLAB
FORTRAN
COMPUTER LABORATORY
UNIX CLASS ACCOUNTS
- Class directory: /group/class/m72002. I will put
the codes by chapter, e.g., the codes for chapter 1 are under
/group/class/m72002/chap1.
- Class submit directory: /group/submit/m72002.
You must submit course work via the
coursework submission tools.
- Group you belong to and class mailing list: m72002.
- If you do not know your password go to 303A MLH.
- Accounts for students who are not majors within the Division
of Mathematical Sciences will be deleted after the term has ended!
Note the expiration date at the top of your Unix Account
Information form.
- The directory for this class will be removed before the
beginning of the next term! If there is anything in this directory
that you would like to save, you must archive it to your own
home directory.
TEACHING ASSISTANT
HOMEWORK ASSIGNMENTS
- Due Friday, September 4: Problems 1.1.2ac, 1.1.6., 1.2.2, and 1.2.11.
- Due Friday, September 11: Modify the Matlab code
/group/class/m72002/chap1/Gregory_Taylor.m by using
nested iterations (Horner's algorithm) and minimize
the number of arithmetic operations within the loop.
You may have to use one
of the following functions in Matlab: ceil, fix, floor, mod, or round.
Submit your work into /group/submit/m72002/homework2
via the
coursework submission tools.
- Due Friday, September 18:
Problems 2.1.1.b, 2.1.3.bc, 2.1.5. In problem 2.2.10 write the
given single precision Fortran code and run it. Modify the code
to double precision and run it. Rewrite the same code in
Matlab and run it. Comment on your results.
- Due Friday, September 25:
Problems 3.1.1.d, 3.1.6.b, 3.1.14.a, 3.2.5, 3.3.1.
For problem 3.1.14.a, do not submit your work via the
coursework submission tools, but plot the graphs (Matlab
is recommended). In problem 3.3.1 write a Fortran or
Matlab code and submit your work into /group/submit/m72002/homework4
via the
coursework submission tools.
- Due Friday, October 9:
Use the subroutine
BISECT to solve 4.1.1ad, 4.1.6. Solve the same problems, but this
time using the subroutine
NEWTON. Do problems 4.2.3abc, 4.2.9.
- Due Monday, October 19:
Do problems 4.3.7, 4.5.1, 5.1.1abc, 5.1.7.
Use the subroutine
DIVDIF to solve 5.2.7. Print out your codes and the results.
- Due Friday, October 30:
Do problems 5.3.4, 5.3.7 (use Matlab, print out your code and the pictures),
5.4.3, 5.4.9. Using the Fortran code
SIMPSON
compute the following integrals: e^{-x^2} between -1 and 1; x^3 between 4 and 8;
x^4 between 4 and 8; 1/(1+cos(x)^2) between 0 and pi, print out just your results.
- Due Wednesday, November 11:
Do problems 7.1.6a, 7.1.8, 7.2.1ab, 7.2.5ac, 7.2.12,
7.3.5, 7.4.1a, and 7.4.2 (repeat only 7.4.1a).
- Due Friday, December 4:
Do problem 8.1.1. Use the double precision routines DGETRF, DGETRS from LAPACK
(see copies to be distributed in class) on HP machines
to solve this problem 8.1 for the data (-1,1), (0,2), (1,1), (3,2). To link
LAPACK and BLAS routines in Fortran when compiling, type for example the
command f90 myprogram.f -L/usr/pkg/lapack/LAPACK/ -llapack -lblas
to compile the program myprogram.f under UNIX. Solve 8.2.7b using
these same routines. Do problems 8.3.1ac, 8.3.19a, 8.4.6a, 8.5.3a (consider
the 1-norm only for this exercise), 8.8.1b (iterate until the error
is less than 0.01 in the (Euclidean) 2-norm).
- Due Friday, December 11:
Do problem 9.1.1ad, 9.1.2, 9.2.3. Apply the code
DOPRI5
in FORTRAN to find y(1) the solution at x=1 of the initial
value problem y'=sin(y), y(0)=0.1. You can also apply instead the
Matlab routine ode23 to solve this problem.
SOME INTERESTING LINKS
Two
disasters due to floating-point errors.
The Flight 501 Failure
(more
technical details on that).
Laurent O. Jay
Department of Mathematics
14 MacLean Hall
The University of Iowa
Iowa City, IA 52242-1419
USA
Tel: (319)-335-0898
Fax: (319)-335-0627
E-mail: ljay@math.uiowa.edu
This page was last modified on .