
This web page is for people who hated/feared/wereconfusedby math in
school, but have decided they want to learn more on their own.
I have no particular credentials for writing this page, except:
 I did ok until differential equations, and then bombed out.
Years later I went back and through self study learned enough math to
enjoy getting an EE degree. After that I dabbled in quantum
mechanics, quantum field theory, and chaos theory.
 My daughter was confused by fractions in 5th grade. Through
home study she was ready for calculus by 7th grade.
The keys in both cases were:
 Find and use the very best tools.
We'll discuss these below. I typically looked at perhaps a dozen
texts or tools for each one selected.
In addition to specific primary texts, I like the Schaum Outline
Series. They are cheap; they give a nononsense description of the
technique; they provide clearly worked examples; and they provide lots
of answered problems.
 Find a way to make regular sessions fun 
and to stop each session before it becomes drudgery.
For my daughter, we set a timer for 30 minutes. Totally focus for the
30 minutes. Stop in mid problem when the timer dings. For myself, I
tend do a section at a time. Either way, the idea is to concentrate
intensely for a short time, and then get away before you burn out or
slow down.
 Practice each skill until it is trivial.
This is a problem for both group classes and selfstudy. Math is
defined and taught with very little redundancy. So if you miss
something early on, you are in serious trouble later.
My rule of thumb is this: Do all the problems available in the text.
Repeat any that you can't solve as fast as you can write.
 Move on immediately after making the skill trivial. Don't
keep pounding on it.
Again, because math is built up as a house of cards, the lower level
skills need to be fresh when needed for higher levels.
This is mainly a problem in group classes. The best students get
bored waiting for the rest of the class to catch up  and then can't
remember the material when it is finally needed.
For selfstudy, the implication is that if you do take time off, you
probably need to refresh on the more basic material.
 Keep your work  and make it readable.
Keep a notebook with all the problems and solutions worked out. These
represent your brain's approach to the problem domain. Years later
you can look at it and say "Hmmm. So that's how I do Laplace
transforms." Redoing a few of the problems can bring you back up to
speed.
Making it readable includes:
 Date each page
 For complex problems (e.g., "story problems"), do 1 per page.
 Clearly define the problem, drawing diagrams and
labeling parameters as needed.
 Describe how you will attack the problem. Convert this to
mathematical form (e.g., sets of equations)
 Solve the math symbolically, then insert parameters and solve
for the specific case
 Convert back to reality as needed.
 Draw a rectangle around the answer(s).
There is no way you will study hard enough to learn math if it isn't
fun. Part of the fun is mastering a difficult topic. Part is simply
knowing the topic exists. I find it helps to mix history/biography
with raw math and application problems.
Another possibility is to build the study program around a hobby. The
classic story is the boy who learns math to do baseball box scores.
That can be carried on to the physics of bat hitting ball,
aerodynamics of a curve ball, etc. Or it can be built around
fingernail polish or horses, or ...
All that matters is that you are interested, and stay interested for
several hundred hours (in 30 minute blocks) over several years.
I never found a text appropriate for this phase, so I've written down
my own approach. It assumes a teacher/coach (typically a parent) is
there every single minute, in an intense oneonone partnership.
See: K6
Here we assume you have a gut feel for and casual comfort with:
 Naturals (0...n)
 Integers (m...0...n)
 Rationals (a/b)
 Reals, in decimal form (0.125, 0.333333, etc.)
 Enumeration, identity, associative, commutative, distributive laws
 Multidigit addition, subtraction, multiplication, division
 Rote skill for single digit addition, subtraction,
multiplication, and division.
 Introduction to exponents
A graphing calculator, or a computer program which does graphs.
Swokowski's Precalculus [swok90] (whatever the
current edition happens to be).
Make photocopies of the insides of the covers (various formulas). On
the page with trigonometry, add:
periodic: y = a cos(bx+c) + d
amplitude= a period = 2pi/b frequecy = 1/period
shift = c/b offset = d
d+a. .
 . .  .
d  .   .     
 . .  .
da . .
++
c/b (2pi/b)c/b
Put these copied pages in plastic page protectors, and put these in
your math notebook. These cover pages are the whole story. The text
itself just gives you practice in using the formulas. You can use
those cover pages for many years after you finish the book.
The trick is to really understand each topic as you go. Don't say
"Oh, I'll figure it out when I need it later." To really understand,
do all the odd numbered problems (i.e., the ones with answers). At
least do them up to the story problems.
Story problems are a special case. You don't need them to learn the
math technique itself. But if you want to actually do something with
math (instead of just play with abstractions), you need practice
converting reality into math and math into reality. In a way, you are
really learning physics, chemistry, surveying, engineering,
etc. instead of math at that point. So do the story problems in each
section, but don't panic if you can't see how to get a few of them
turned into math.
Also, once you figure out the realitytomath transformation, it is
usually obvious. So if it was a hard one, throw away your work paper
and redo the problem on a fresh sheet  it will flow smoothly and you'll
be proud of yourself.
The biggest problem with calculus is that people think it is complex.
So they spend a year or so on all the nuances of beginning calculus,
get discouraged, and never get any further.
The truth is, you need to do a few difference limits to see what's
going on, memorize a few formulas for derivatives, and have access to
a table of integrals. In real life, you use computer programs which
do symbolic and numerical differentiation and integration.
So how can you learn just the essentials?
Paper and pencil, plus the graphing calculator.
Hahn et al Hurricane Calculus [hahn95].
If you need more examples, get Ayres's Calculus (Schaum Outline).
But do NOT try to do all the problems in it  they are overwhelming.
Just look at examples when you don't undertand a particular topic in
Hurricane Calculus.
Read the material, then solve the Additional_Problems (no more than
15 per section). If you are stuck, try this:
Go back and write out at least some of the worked examples, trying to
understand each step as you go. It is like a beginning novelist
typing out classics or a painter copying the masters  it helps you
focus on the mechanics of the process, which in turn helps you see the
big picture.
 Limits. If you like finding abstract pattens, do all the limit
stuff up front. If you just want to use the stuff, skip to section
3.2 "The Easy Way".
 Derivatives. The table in 3.2 is the whole story on
derivatives. The cover pages for Kreszig (see below) have a nice
listing of the formulas. Learn to use the formulas, but don't bother
trying to memorize them.
 Integrals. The essential point is this: Integration, except in
special cases found in calculus texts, is hard or impossible. In real
life we use lookup tables, or use a computer program which does
symbolic and/or numerical integration.
DO NOT BOG DOWN HERE. Read Ch 8 (Techniques for Solving
MindBending Integrals), be impressed that someone thought of those
tricks, and keep moving.
 Sequences and Series. As noted above, solving integrals
directly is difficult or impossible in many cases. Instead we make
approximations of the formula as a series of eversmaller correction
factors. You just keep calculating the corrections until the answer is
good enough. In some cases you can get terms to cancel, so the series
gives an elegant solution. That's what math texts are for. In most
cases you just have to grind it out. That is what computers are for.
Either way, you are responsible for assuring the series converges (and
how quickly), so take this section seriously.
 Remaining chapters. Skip them. Kreyszig does a better job.
That's it. We took one of the smallest calculus books round, and
chopped out major sections. The key is to keep moving  don't get
stuck in the quicksand.
This is the bulk of what scientists and engineers do in terms of
math. It assumes fluency in the previous tools, so just jumping here
doesn't help. You really do need to know algebra and exponentials and
such from Precalculus, plus derivatives and series approximations from
Calculus. But this time they are applied to more complex things, such
as differential equations, matrices, and vectors.
Paper and pencil; calculator.
Kreyszig's Advanced Engineering Mathematics [ krey88].
I've had various editions  just get the latest one available.
Copy the inside cover pages, put them in plastic page protectors, and
save them in your notebook.
Ok, if you are working through Kreyszig, you have nothing to be
ashamed of. You can leave it on your desk top and impress friends,
family, and coworkers. Of course, you actually have to do the
problems...
In fact, you should plan on doing all the oddnumbered problems.
They build nicely from simple cases, to tricky cases, to applications.
If you jump right in to the story problems, you will be wasting time
even if you solve them. Think of it this way: The fastest way to
build a yacht is to build its dinghy first (to learn the process) and
then build the yacht. By the time you hit the story problems for a
section, the mechanics should be rote.
This could take years. So what? You are doing it for fun and
personal development. Is TV really better than being able to say "So
that's what Laplace Transforms are about."?
 Differential Equations. Plow on through to Ch 3 (systems of
differential equations). Many real world problems translate to
systems of differential equations. (These are often converted to
finite difference equations, which can then be solved by computer.)
If you like math for its own sake, work through the power series
material. You can see all the old algebra skills coming into play at
a higher level. But it might be interesting to sci/engr folks only if
their subject domains need the formulas.
The Laplace Transform is a winner. If you can get your problem stated
in the Laplace form, you are down to straight algebra. Control
systems are generally defined and analyzed this way.
 Linear algebra and vectors. You've already covered much of the
material in Precalculus, ch 8. So either do some review, or skip
directly to vector differential calculus. There, you are sure to bog
down in grad, div, and curl. Just take your time and do the problems.
[I don't have a shortcut. I have trouble with left and right in the
first place, much less the "right hand rule". I think Kreyszig does
as good a job of explaining and demonstrating as anyone.]
 Fourier analysis. After a lot of gyrations, you end up with
variations on Fourier Transforms. In other words, translate from
waves to spectra and back. With any luck you can fit the problem into
a FFT (Fast Fourier Transform), and use existing software to solve it.
It's more fun if you can apply it to sound or image processing.
If you are a glutton for punishment, go look at wavelets. Fourier
transforms define events in terms of continuous waves, but that leaves
inconvenient side bands. Wavelets use little blips of waves as the
basis, allowing much tidier modeling of single events.
 Complex analysis. Here you get to do everything you've learned
so far all over again, but with complex numbers. To be honest, I
didn't complete all the sections. I do remember completing all of Ch
12 in 1 day (and have the solutions in my notebook to prove it.)
Unfortunately, I have no idea what it's about right now...
 Numerical Methods. Everyone needs to know about this, but in
real life you use prebuilt libraries (or at least wellknown
algorithms). So scan the text, then implement your own routines as a
learning exercise. E.g., my m3na (source:
m3na.tgz ). Then go use serious libraries such as
atlas, blas, lapack, gsl, fftw.
 Optimization. Generally this is taught as Operations Research
in MBA classes. You might want to implement simplex. But use an
algorithm text.
 Graphs. Again, an important topic. But get it from an
algorithm text.
 Statistics. Another important topic. Kreyszig does a good job.
[I also have some notes I'll put on the web sometime which takes a
single problem and evolves statistics as the answer becomes more
sophisticated.]
Suggested algorithm texts:
 Algorithms [sedge88]
 Introduction to Algorithms [cor90]
 Foundations of Data Structures in C++ [horo95]
Kreyszig covers most of the needed material except tensors and groups.
There is a text (M. Boas: Mathematical Methods in the Physical
Sciences, 2nd Ed) which ought to be appropriate here. But it doesn't
have answers to problems, so it is hard to use for selfstudy.
However, you can use it to see another view on topics in Kreyszig.
Someimes seeing two different people cover the same ground helps fill
in the missing links.
For tensors, I came up with Kay's Tensor Calculus
[kay88]. Ok, I didn't really internalize this. I kept
thinking: I need a computer program to keep track of all the pieces.
But even just doing some of the problems is cool. Heck, just knowing
about metric tensors is cool.
I haven't yet found a good text for selfstudy on groups. The best I
have on my shelf is ch 2, of Kaku's Quantum Field Theory
[kaku93] There is a Shaum Outline for it you might try.
By this point you are choosing your own topics, and finding your own
texts and tools. In my case, I'm moderately interested in:
 Logic. esp. in the context of logic programming and AI
 Category Theory. esp. in the context of programming "types"
When you can't handle the material, you can turn to biographies,
history, and popularizations. They at least give a flavor of what's
going on.
A good start point is Kramer's
The Nature and Growth of Modern Mathematics [kram81].
Next check out Davis's Mathematical_Experience
[davis99] to understand vicariously how it feels to do
math. Then look at Mathematical_People (out of print) for
interviews with professional mathematicians.
A book often cited is Peterson's The Mathematical Tourist
[peter88].
 Cor90
 T.H. Cormen, C.E. Leiserman, R.L. Rivest. Introduction to Algorithms
MIT Press, 1990. ISBN 0262031419
 Das92
 S.K. Das. Deductive Data Bases and Logic Programming
AddisonWesley, 1992. ISBN 0201568977
 Davis99
 P.J. Davis, R. Hersh. The Mathematical Experience
Houghton Mifflin Co,1999. ISBN: 0395929687
 Hahn95
 J.B. Hahn, T.J. Dunlap, S.P. Matyus. Hurricane Calculus
Prometheus Enterprises, 1995. ISBN 1886783004
 Horo95
 E. Horowitz, S. Sahni, D. Mehta. Foundations of Data Structures in C++
W.H. Freeman, 1995. ISBN 0716782828
 Kaku93
 M. Kaku. Quantum Field Theory
Oxford University Press, 1993. ISBN 019507652
 Kay88
 D.C. Kay. Tensor calculus
Schaum Outline Series, 1988. ISBN 0070334846
 Kram81
 E.E. Kramer. The Nature and Growth of Modern Mathematics
Princeton University Press, 1981. ISBN 0691023727
 Krey88
 E. Kreyszig. Advanced Engineering Mathematics
John Wiley and sons, 1988. ISBN 0471858242
 Peter88
 I. Peterson. The Mathematical Tourist
I. Peterson, 1998. ISBN 0716719533
 Sedge88
 R. Sedgewick. Algorithms , 2nd ed.
AddisonWesley, 1988. ISBN 0201066734
 Strogatz94
 S. H. Strogatz. Nonlinear Dynamics and Chaos Westview Press, 1994.
Paperback 2000. ISBN13 9780738204536.
The reviews say it is exceptionally clear, and that fits. Assumes
upper class or 1st year graduate level understanding of math and
physics. Works from 1D linear systems to nonlinear, and then to 2D
nonlinear, and then on to chaos. Lots of solved problems, with
examples from physics, biology, chemistry, and engineering.
 Swok90
 E.W. Skowkowski. Precalculus: functions and graphs 6th ed.
PWSKent Publishing, 1990. ISBN 0534920861

