Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watmath!clyde!rutgers!husc6!cmcl2!beta!unm-la!unmvax!hi!jedi!sundc!hadron!cos!howard
From: howard@cos.UUCP
Newsgroups: comp.edu
Subject: Re: Teaching Math in the Late Twentieth Century
Message-ID: <315@cos.COM>
Date: Fri, 12-Jun-87 11:38:43 EDT
Article-I.D.: cos.315
Posted: Fri Jun 12 11:38:43 1987
Date-Received: Sun, 14-Jun-87 22:02:48 EDT
References: <3797@udenva.UUCP>
Organization: Corporation for Open Systems, McLean, VA
Lines: 71
Summary: CS majors have too much math

In article <3797@udenva.UUCP>, skajihar@udenva.UUCP ("Lord of Sith" Kajihara) writes:
> 
> I graduated this past Saturday (June 6) with neither computer science or math
> as a major. One complaint that I had about
> some of the [CS] classes of the above curriculum was the emphasis that they had on
> the discrete mathematics.  How useful is it to learn graph theory and
> algorithms for polynomials in a linear algebra course?  Is not the usual path
> one covering matrices and vector spaces?

Graph theory is intensely useful to computer science practitioners involved
with data bases and communications; it is less relevant to mathematicians.

> 
> Another complaint was the eventual phasing out of true math courses and
> increasing the offering of computer science courses. 

I hope we can learn from the physicists and others and recognize the need
for mathematical tool courses (e.g., differential equations for physics)
as distinct from mathematical theory courses.  It may be that there fewer
pure mathematics programs can be justified, if the fewer programs can be
of higher quality.

> 
> Last, but not least, is it that cruel to expect that the computer science
> people have at least a basic understanding of the math that they will use?
> Granted, as I have been told, this is not always necessary for the fields
> that many of them take up, but who knows what these people will take up?
> There are people out in the world who do scientific programming for
> researchers; if the program crashes, they have no idea of how their code
> relates to the original equations that they were given.

Yes, I think it is cruel, if the math they are taught is not the math
they will use.  "Scientific programmers," as I believe you see
them, are increasingly a minority, as better software and personal computers
are available to more computer-literate scientists.  My experience is that
the more sophisticated numerical algorithms are more apt to be written
by physical scientists and engineers than "scientific programmers."

My flame here on "the math they will use" concerns what I feel is an
emphasis on analysis and theoretical abstract algebra, as opposed to
useful discrete mathematics, statistics, and operations research.  I hae
spent about 20 years in mostly state-of-the-art operating systems,
networks, and online applications.  I have found it necessary to study
more discrete math (graph theory, groups, number theory as applied to 
coding, etc.) and statistics/OR, and have rarely needed calculus or
differential equations (even with appreciable hardware work!)

Even in communications traffic engineering, founded on queueing theory,
I find little need to, say, derive a Poisson or Erlang distribution.
For most purposes, there are adequate packages or subroutine libraries;
for special purposes, specialists are needed.  If I want a complex
queueing model developed, where special distributions are involved,
I would no more try to do it than have a "scientific programmer"
design the code for a network control center!  There are enough subtleties
involved in modern queueing theory that it is not a job for dilettantes.

> Moreover, theyteach material at junior/senior level that should have
>been taught in high
> school.  I have been flamed by some of these instructors that I am out of the
> targeted group, but I wonder if they are not aiming their expectations too low.

Without knowing your background and current status, I wonder if you may
be equating "mathematical maturity" with "software maturity?"  In a related
context, I have taken graduate CS courses where material belonging in
an undergraduate or even high school honors programs was being taught,
such as extended discussions of singly linked lists in an alleged graduate
CS course in data structures.

Howard
(howard @ cos.com  via hadron, seismo->hadron, sundc, hqda-ai)
(703) 883-2812