### Preface - MIT Mathematics

```Preface
Differential equations and linear algebra are the two crucial courses in undergraduate
mathematics. This new textbook develops those subjects separately and together.
Separate is normal—these ideas are truly important. This book presents the basic course
on differential equations, in full :
Chapter 1 First order equations
Chapter 2 Second order equations
Chapter 3 Graphical and numerical methods
Chapter 4 Matrices and linear systems
Chapter 6 Eigenvalues and eigenvectors
I will write below about the highlights and the support for readers. Here I focus on the
option to include more linear algebra. Many colleges and universities want to move in
this direction, by connecting two essential subjects.
More than ever, the central place of linear algebra is recognized. Limiting a student
to the mechanics of matrix operations is over. Without planning it or foreseeing it, my
lifework has been the presentation of linear algebra in books and video lectures :
Introduction to Linear Algebra (Wellesley–Cambridge Press)
MIT OpenCourseWare (ocw.mit.edu, Mathematics 18:06 in 2000 and 2014).
Linear algebra courses keep growing because the need keeps growing. At the same time,
a rethinking of the MIT differential equations course 18.03 led to a new syllabus.
And independently, it led to this book.
The underlying reason is that time is short and precious. The curriculum for many
students is just about full. Still these two topics cannot be missed—and linear differential
equations go in parallel with linear matrix equations. The prerequisite is calculus, for a
single variable only—the key functions in these pages are inputs f .t/ and outputs y.t/.
For all linear equations, continuous and discrete, the complete solution has two parts :
One particular solution yp
All null solutions yn
Ayp D b
Ayn D 0
Those right hand sides add to b C 0 D b. The crucial point is that the left hand sides
add to A.yp C yn /. When the inputs add, and the equation is linear, the outputs add.
The equality A.yp C yn / D b C 0 tells us all solutions to Ay D b :
The complete solution to a linear equation is y D .one yp / C .all yn /:
v
vi
Preface
The same steps give the complete solution to dy=dt D f .t/, for the same reason.
We know the answer from calculus—it is the form of the answer that is important here :
dyp
D f .t/
dt
dyn
D0
dt
dy
D f .t/
dt
is solved by
is solved by
is completely solved by
yp .t/ D
Rt
f .x/ dx
0
yn .t/ D C .any constant/
y.t/ D yp .t/ C C
For every differential equation dy=dt D Ay C f .t/, our job is to find yp and yn :
one particular solution and all homogeneous solutions. My deeper purpose is to
build confidence, so the solution can be understood and used.
Differential Equations
The whole point of learning calculus is to understand movement. An economy grows,
currents flow, the moon rises, messages travel, your hand moves. The action is fast or
slow depending on forces from inside and outside : competition, pressure, voltage, desire.
Calculus explains the meaning of dy=dt, but to stop without putting it into an equation
(a differential equation) is to miss the whole purpose.
That equation may describe growth (often exponential growth e at ). It may describe
oscillation and rotation (with sines and cosines). Very frequently the motion approaches an
equilibrium, where forces balance. That balance point is found by linear algebra, when the
rate of change dy=dt is zero.
The need is to explain what mathematics can do. I believe in looking partly outside
mathematics, to include what scientists and engineers and economists actually remember
and constantly use. My conclusion is that first place goes to linear equations. The essence
of calculus is to linearize around a present position, to find the direction and the speed
of movement.
Section 1.1 begins with the equations dy=dt D y and dy=dt D y 2 . It is simply
wonderful that solving those two equations leads us here :
dy
Dy
dt
dy
D y2
dt
1
1
y D 1 C t C t2 C t3 C 2
6
y D et
y D 1 C t C t2 C t3 C y D 1=.1
t/
To meet the two most important series in mathematics, right at the start, that is pure
pleasure. No better practice is possible as the course begins.
vii
Preface
Important Choices of f .t/
Let me emphasize that a textbook must do more than solve random problems. We could
invent functions f .t/ forever, but that is not right. Much better to understand a small
number of highly important functions:
f .t/ D
f .t/ D
f .t/ D
f .t/ D
sines and cosines
exponentials
1 for t > 0
impulse
(oscillating and rotating)
(growing and decaying)
(a switch is turned on)
(a sudden shock)
The solution y.t/ is the response to those inputs—frequency response, exponential
response, step response, impulse response. These particular functions and particular
solutions are the best—the easiest to find and by far the most useful. All other solutions
are built from these.
I know that an impulse (a delta function that acts in an instant) is new to most students.
This idea deserves to be here ! You will see how neatly it works. The response is like the
inverse of a matrix—it gives a formula for all solutions. The book will be supplemented by
video lectures on many topics like this, because a visual explanation can be so effective.
Support for Readers
Readers should know all the support that comes with this book :
math.mit.edu/dela is the key website. The time has passed for printing solutions to
odd-numbered problems in the back of the book. The website can provide more detailed
solutions and serious help. This includes additional worked problems, and codes for numerical experiments, and much more. Please make use of everything and contribute.
ocw.mit.edu has complete sets of video lectures on both subjects (OpenCourseWare
is also on YouTube). Many students know about the linear algebra lectures for 18:06 and
18:06 SC. I am so happy they are helpful. For differential equations, the 18.03 SC videos
and notes and exams are extremely useful.
The new videos will be about special topics—possibly even the Tumbling Box.
Linear Algebra
I must add more about linear algebra. My writing life has been an effort to present
this subject clearly. Not abstractly, not with a minimum of words, but in a way that is
helpful to the reader. It is such good fortune that the central ideas in matrix algebra
(a basis for a vector space, factorization of matrices, the properties of symmetric and
orthogonal matrices), are exactly the ideas that make this subject so useful. Chapter 5
emphasizes those ideas and Chapter 7 explains the applications of AT A.
Matrices are essential, not just optional. We are constantly acquiring and organizing
and presenting data—the format we use most is a matrix. The goal is to see the relation
between input and output. Often this relation is linear. In that case we can understand it.
viii
Preface
The idea of a vector space is so central. Take all combinations of two vectors or two
functions. I am always encouraging students to visualize that space—examples are really
the best. When you see all solutions to v1 C v2 C v3 D 0 and d 2 y=dt 2 C y D 0, you
have the idea of a vector space. This opens up the big questions of linear independence and
basis and dimension—by example.
If f .t/ comes in continuous time, our model is a differential equation. If the input
comes in discrete time steps, we use linear algebra. The model predicts the output y.t/ this
is created by the input f .t/. But some inputs are simply more important than others—they
are easier to understand and much more likely to appear. Those are the right equations to
present in this course.
Notes to Faculty (and All Readers)
One reason for publishing with Wellesley-Cambridge Press can be mentioned here.
I work hard to keep book costs reasonable for students. This was just as important for
Introduction to Linear Algebra. A comparison on Amazon shows that textbook prices
from big publishers are more than double. Wellesley-Cambridge books are distributed by
SIAM inside North America and Cambridge University Press outside, and from Wellesley,
with the same motive. Certainly quality comes first.
I hope you will see what this book offers. The first chapters are a normal textbook
on differential equations, for a new generation. The complete book is a year’s course
on differential equations and linear algebra, including Fourier and Laplace transforms—
plus PDE’s (Laplace equation, heat equation, wave equation) and the FFT and the SVD.
This is extremely useful mathematics ! I cannot hope that you will read every word.
But why should the reader be asked to look elsewhere, when the applications can come
so naturally here ?
A special note goes to engineering faculty who look for support from mathematics. I
have the good fortune to teach hundreds of engineering students every year. My work with
finite elements and signal processing and computational science helped me to know what
students need—and to speak their language. I see texts that mention the impulse response
(for example) in one paragraph or not at all. But this is the fundamental solution from
which all particular solutions come. In the book it is computed in the time domain, starting
with e at , and again with Laplace transforms. The website goes further.
I know from experience that every first edition needs help. I hope you will tell me what
should be explained more clearly. You are holding a book with a valuable goal—to become
a textbook for a world of students and readers in a new generation and a new time, with
limits and pressing demands on that time. The book won’t be perfect. I will be so grateful
if you contribute, in any way, to making it better.
ix
Preface
Acknowledgments
So many friends have helped this book. In first place is Ashley C. Fernandes, my early
morning contact for 700 days. He leads the team at Valutone that prepared the LATEX files.
They gently allowed me to rewrite and rewrite, as the truly essential ideas of differential
equations became clear. Working with friends is the happiest way to live.
The book began in discussions about the MIT course 18.03. Haynes Miller and
David Jerison and Jerry Orloff wanted change—this is the lifeblood of a course. Think
more about what we are doing ! Their starting point (I see it repeated all over the world) was
to add more linear algebra. Matrix operations were already in 18.03, and computations
of eigenvalues—they wanted bases and nullspaces and ideas.
I learned so much from their lectures. There is a wonderful moment when a class gets
the point. Then the subject lives. The reader can feel this too, but only if the author does.
I guess that is my philosophy of education.
Solutions to the Problem Sets were a gift from Bassel Khoury and Matt Ko. The
example of a Tumbling Box came from Alar Toomre, it is the highlight of Section 3.3
(this was a famous experiment in his class, throwing a book in the air). Daniel Drucker
watched over the text of Chapters 1-3, the best mathematics editor I know. My writing
tries to be personal and direct—Dan tries to make it right.
The cover of this book was an amazing experience. Gonçalo Morais visited MIT from
Portugal, and we talked. After he went home, he sent this very unusual picture of a strange
attractor—a solution to the Lorenz equation. It became a way to honor that great and
humble man, Ed Lorenz, who discovered chaos. Gail Corbett and Lois Sellers are the artists
who created the cover—what they have done is beyond my thanks, it means everything.
At the last minute (every book has a crisis at the last minute) Shev MacNamara
saved the day. Figures were missing. Big spaces were empty. The S -curve in Section
1.7, the direction fields in Section 3.1, the Euler and Runge-Kutta experiments, those and
more came from Shev. He also encourages me to do an online course with new video
lectures. I will think more about a MOOC when readers respond.
Thank you all, including every reader.
Gilbert Strang
x
Outline of Chapter 1 : First Order Equations
1:3
1:4
1:5
1:6
1:7
1:8
Solve
Solve
Solve
Solve
Solve
Solve
dy=dt
dy=dt
dy=dt
dy=dt
dy=dt
dy=dt
D ay
D ay C q.t/
D ay C e st
D a.t/y C q.t/
D ay by 2
D g.t/=f .y/
Construct the exponential e at
Four special q.t/ and all q.t/
Growth and oscillation : s D a C i !
Integrating factor D 1/growth factor
The equation for z D 1=y is linear
R
R
Separate f .y/ dy from g.t/ dt
The key formula in 1.4 gives the solution y.t/ D e at y.0/ C
Rt
e a.t
s/
q.s/ds.
0
The website with solutions and codes and extra examples and videos is math.mit.edu/dela
Please contact [email protected] with questions and book orders and ideas.
```