APMA 1690: Computational Probability & Statistics


Course Information

Lectures
    M/W/F  9:00 - 9:50 am  [lectures continue during reading period]
    Hunter Psychology Lab, Carmichael Auditorium (Room 130)

Instructor
    Matthew Harrison
    182 George St Room 327
    Office hours: Tuesdays 2-5pm
    Firstname_Lastname@brown.edu  [please use the discussion board whenever appropriate]

TAs
    Dahlia Nadkarni
    Office hours: Mondays 2:30-4:30pm; 182 George St Room 024
    Firstname_Lastname@brown.edu  [please use the discussion board whenever appropriate]

    Wei Wu
    Office hours: Friday 3-5pm; 180 George St Room 002
    Firstname_Lastname@brown.edu  [please use the discussion board whenever appropriate]

    Hoy Loper
    Office hours: Monday 12-2pm;  J. Walter Wilson Room 501
    Firstname_Lastname@brown.edu  [please use the discussion board whenever appropriate]

Computer TA [for basic Matlab help only]
    Seonmin Ahn
    Office hours: Wednesdays 3-5pm, Thursdays 9-11am; 37 Manning St Room 005
    Firstname_Lastname@brown.edu

Website
    http://mycourses.brown.edu/   [automatic access for enrolled students, guest access for anyone for the first few weeks, email me after that for access]

Prerequisites

Calculus-based undergraduate probability, such as APMA 1650 or MATH 1610, is required.  Familiarity with matrix algebra is helpful.  No prior computer programming experience is required.

Course Description

Probability and statistics are increasingly computational fields.  Students will be exposed to several topics that use computers to solve challenging problems in probability and statistics.  They will also use computers to develop intuitions about classical analytic results in probability and statistics.  The tentative list of topics is: 

1. Simulating randomness (pseudo-random number generation, transformation of random variables, rejection sampling)
2. Stochastic approximation (law of large numbers, central limit theorem, Monte Carlo integration, importance sampling)
3. Random walks (recurrence properties, exit times)
4. Graphical models (Markov random fields, Bayes nets, hidden Markov models, dynamic programming)
5. Dimensionality reduction (principle components analysis, independent components analysis)

Homework

The graded components of this course are (approximately 14) homework assignments (and nothing else).  The homework assignments are primarily computer experiments (Matlab) with the occasional pencil and paper problem.

You may collaborate on homework, but only after attempting the problems by yourself.  You must write your own computer code and submit your own homework.  You are not allowed to use solutions from previous years or from external sources.

Grading Policies

Late Homework

Any on-time homework submitted to the TAs will be graded, will be entered into the gradebook, and will receive feedback from the TAs.  Any late homework, for whatever reason, should be submitted to me directly.  I will look at all late homework at the end of the semester if it could make a difference in your final letter grade.  Depending on your excuse for submitting late, I may only give partial credit for late homework.  You will not get feedback on late homework and it will not be entered into the gradebook.

This policy is a very fair way of balancing the administrative hassle of late homework with the legitimate conflicts that occasionally affect each student.  Do not try to take advantage of it.  If you submit a lot of late homework, I am very unlikely to take it all into consideration when assigning final letter grades.  Incidentally, I do not adhere to the policy of dropping the lowest homework grade.  (Someone always asks this question.)  Homework is really the only way to learn the content of this course and the last thing I want to do is to encourage you to skip some of it.

Note that this policy includes partially completed homework.  If you cannot complete an assignment on time, you should submit whatever you have finished on time.  At least you will get some partial credit and perhaps some useful feedback.  Then you should finish the rest of the assignment and submit it to me directly.

If you have a serious or prolonged personal emergency during the semester, you should contact me and/or a dean as soon as possible.  We will make every effort to help you and these grading policies can, of course, be adjusted to accommodate your situation.

Homework Grading Problems

1.  My primary concern is content.  If there is something that you do not understand, come to office hours, post on the discussion board, etc.  We'll get to the bottom of it.

2.  My next concern is administrative grading errors.  If we added your points wrong, or typed your grade into the gradebook incorrectly, or if your homework got lost, please contact me immediately and we will correct the problem.  If we simply didn’t see part of your solutions, follow the directions for item 3 below (and also take a look at item 5).

3.  My least concern is grading disagreements.  There is very little chance that your final grade will be affected by a point here or there on some homework.  It is a big waste of everyone’s time to haggle about it.  Nevertheless, there is, of course, some slight chance that your final grade could be affected by a homework grading mistake, so here is how to handle it:

•    Copy the pages of your homework in question.
•    Attach a written explanation of your disagreement.
•    Put it in my mailbox or give it to me personally.
•    I will address the concern at the end of the semester if your letter grade is borderline and may have been affected by the disagreement.  Otherwise, I may never look at it.

4.  Do not contact or confront the TA about grading or late homework.  Direct all grading questions to me.  I will discuss systematic grading problems with the TA if necessary.

5.  Some general advice about homework and life.  Everything that you say or write is for an audience.  Sometimes the audience is you.  As far as homework grades are concerned, the audience for your APMA 1690 homework is a graduate TA who has at most a few minutes per problem.  I believe that it is your responsibility to adjust to the grader and his/her highly restrictive time constraints.  (So, for example, make it easy to find all the parts to your solutions and write clearly.)

Final Grades

Here are the minimum numerical grades needed to ensure each final letter grade.  I may lower these standards, but I will not raise them.

A      90%
B      75%
C      60%    

Pass  60%