Math 54 Cheat Sheet

Vector spaces
Subspace: If u and v are in W , then u + v are in W ,
and cu is in W
Nul(A): Solutions of Ax = 0. Row-reduce A.
Row(A): Space spanned by the rows of A: Row-reduce
A and choose the rows that contain the pivots.
Col(A): Space spanned by columns of A: Row-reduce
A and choose the columns of A that contain the pivots
Rank(A): = Dim(Col(A)) = number of pivots
Rank-Nullity theorem:
Rank(A) + dim(N ul(A)) = n, where A is m n
Linear transformation: T (u + v) = T (u) + T (v),
T (cu) = cT (u), where c is a number.
T is one-to-one if T (u) = 0 u = 0
T is onto if Col(T ) = Rm .
Linearly independence:
a1 v1 + a2 v2 + + an vn = 0 a1 = a2 =
= an = 0.
To show lin. ind, form the matrix of the vectors, and
show that N ul(A) = {0}
Linear dependence: a1 v1 + a2 v2 + + an vn = 0
for a1 , a2 , , an , not all zero.
Span: Set of linear combinations of v1 , vn
Basis B for V : A linearly independent set such that
Span (B) = V
To show sthg is a basis, show it is linearly independent
and spans.
To find a basis from a collection of vectors, form the
matrix A of the vectors, and find Col(A).
To find a basis for a vector space, take any element of
that v.s. and express it as a linear combination of
simpler vectors. Then show those vectors form a
basis.
Dimension: Number of elements in a basis.
To find dim, find a basis and find num. elts.
Theorem: If V has a basis of vectors, then every basis
of V must have n vectors.
Basis theorem: If V is an ndim v.s., then any lin. ind.
set with n elements is a basis, and any set of n elts.
which spans V is a basis.
Matrix of a lin. transf T with respect to bases B and C:
For every vector v in B, evaluate T (v), and express

T (v) as a linear combination of vectors in C. Put the

coefficients in a column vector, and then form the
matrix of the column vectors you found!
Coordinates: To find [x]B , express x in terms of the
vectors in B.
x = PB [x]B , where PB is the matrix whole columns
are the vectors in B.
Invertible matrix theorem: If A is invertible, then: A
is row-equivalent to I, A has n pivots, T (x) = Ax is
one-to-one and onto, Ax = b has a unique solution for
every b, AT is invertible, det(A) 6= 0, the columns of
A form a basis for Rn , N ul(A) = {0},
Rank(A) = n
1



a b
d
b
1
= adbc
c d
c
a




A | I I | A1
Change of basis: [x]C = PCB [x]B (think of C as the
new, cool basis)
[C | B] [I | PCB ]
PCB is the matrix whose columns are [b]C , where b
is in B

Diagonalization
Diagonalizability: A is diagonalizable if
A = P DP 1 for some diagonal D and invertible P .
A and B are similar if A = P BP 1 for P invertible
Theorem: A is diagonalizable A has n linearly
independent eigenvectors
Theorem: IF A has n distinct eigenvalues, THEN A is
diagonalizable, but the opposite is not always true!!!!
Notes: A can be diagonalizable

 even if its not
0 0
invertible (Ex: A =
). Not all matrices are
0 0


1 1
diagonalizable (Ex:
)
0 1
1
Consequence: A = P DP
An = P Dn P 1
How to diagonalize: To find the eigenvalues, calculate
det(A I), and find the roots of that.
To find the eigenvectors, for each find a basis for
N ul(A I), which you do by row-reducing
Rational roots theorem: If p() = 0 has a rational
root r = ab , then a divides the constant term of p, and b
divides the leading coefficient.
Use this to guess zeros of p. Once you have a zero that

works, use long division! Then A = P DP 1 , where

D= diagonal matrix of eigenvalues, P = matrix of
eigenvectors
Complex eigenvalues If = a + bi, and v is an
eigenvector, then A = P CP 1, where 


a
b
P = Re(v) Im(v) , C =
b a
p
by ,
C is a scaling of det(A)

 followed by a rotation
cos()
sin()
1

where:
C=
sin() cos()
det(A)

Orthogonality
u, v orthogonal
if u v = 0.

kuk = u u
{u1 un } is orthogonal if ui uj = 0 if i 6= j,
orthonormal if ui ui = 1
W : Set of v which are orthogonal to every w in W .
If {u1 un } is an orthogonal basis, then:
yu
y = c1 u1 + cn un cj = u uj
j

Orthogonal matrix Q has orthonormal columns!

Consequence:QT Q = I, QQT = Orthogonal
projection on Col(Q).
kQxk = kxk
(Qx) (Qy) = x y
Orthogonal projection: If {u1 uk } is a basis for
W , then
projection

 of y on W is:
 orthogonal
= uyuu1 u1 + + uyuu1 uk
y
1 1
k k
is orthogonal to y
, shortest distance btw y and
yy
W is ky y
k
Gram-Schmidt: Start with B = {u1 , un }. Let:
v1 = u1 

2 v1
v2 = u2 u
v1
v1 v1




u3 v1
3 v2
v3 = u3 v v
v1 u
v2
v v
1

Then {v1 vn } is an orthogonal basis for Span(B),

vi
and if wi = kv
, then {w1 wn } is an
ik
orthonormal basis for Span(B).
QR-factorization: To find Q, apply G-S to columns of
A. Then R = QT A
Least-squares: To solve Ax = b in the least
squares-way, solve AT Ax = AT b.
Least squares solution makes kAx bk smallest.
= R1 QT b, where A = QR.
x

R
Inner product spaces f g = ab f (t)g(t)dt. G-S
applies with this inner product as well.
Cauchy-Schwarz: |u v| kuk kvk
Triangle inequality: ku + vk kuk + kvk

Symmetric matrices (A = AT )
Has n real eigenvalues, always diagonalizable,
orthogonally diagonalizable (A = P DP T , P is an
orthogonal matrix, equivalent to symmetry!).
Theorem: If A is symmetric, then any two
eigenvectors from different eigenspaces are orthogonal.
How to orthogonally diagonalize: First diagonalize,
then apply G-S on each eigenspace and normalize.
Then P = matrix of (orthonormal) eigenvectors, D =
matrix of eigenvalues.
Quadratic forms: To find the matrix, put the
x2i -coefficients on the diagonal, and evenly distribute
the other terms. For example, if the x1 x2 term is 6,
then the (1, 2)th and (2, 1)th entry of A is 3.
Then orthogonally diagonalize A = P DP T .
Then let y = P T x, then the quadratic form becomes
2 , where are the eigenvalues.
1 y12 + + n yn
i
Spectral decomposition:
1 u1 u1 T + 2 u2 u2 T + + n un un T

Second-order and Higher-order

differential equations
Homogeneous solutions: Auxiliary equation: Replace
equation by polynomial, so y 000 becomes r3 etc. Then
find the zeros (use the rational roots theorem and long
division, see the Diagonalization-section). Simple
zeros give you ert , Repeated zeros (multiplicity m)
give you Aert + Btert + Ztm1 ert , Complex
zeros r = a + bi give you
Aeat cos(bt) + Beat sin(bt).
Undetermined coefficients: y(t) = y0 (t) + yp (t),
where y0 solves the hom. eqn. (equation = 0), and yp
is a particular solution. To find yp :
If the inhom. term is Ctm ert , then:
yp = ts (Am tm + A1 t + 1)ert , where if r is a
root of aux with multiplicity m, then s = m, and if r is
not a root, then s = 0.
If the inhom term is Ctm eat sin(t), then:
yp = ts (Am tm + A1 t + 1)eat cos(t) +

ts (Bm tm + B1 t + 1)ert sin(t), where s = m,

if a + bi is also a root of aux with multiplicity m
(s = 0 if not). cos always goes with sin and
vice-versa, also, you have to look at a + bi as one
entity.
Variation of parameters: First, make sure the leading
coefficient (usually the coeff. of y 00 ) is = 1.. Then
y = y0 + yp as above. Now suppose
yp (t) = v1 (t)y1 (t) + v2 (t)y2 (t), where y1 and y2
are
 your hom.
  solutions.
  Then

y1 y2 v10
0
=
. Invert the matrix and
0
0
0
f (t)
y1 y2 v2
solve for v10 and v20 , and integrate to get v1 and v2 , and
finally use: yp (t) = v1 (t)y1 (t) + v2 (t)y2 (t).



1
a b
d
b
1
Useful formulas:
= adbc
c d
c
a
R
R sec(t) = ln |sec(t) +Rtan(t)|,
2
R tan(t) = ln |sec(t)|, tan (t) = tan(x) x,
ln(t) = t ln(t) t
Linear independence: f, g, h are linearly independent if
af (t) + bg(t) + ch(t) = 0 a = b = c = 0. To
show linear dependence, do it directly. To show linear
independence,
form the Wronskian:


g(t)
f (t) = f0(t)
W
(for 2 functions),
0
f (t) g (t)

f (t)
g(t)
h(t)
0
0
0
f

g (t)
h (t) (for 3 functions).
W (t) = f (t)
f 00 (t) g 00 (t) h00 (t)
f (t0 )) is easy to
Then pick a point t0 where det(W
evaluate. If det 6= 0, then f, g, h are linearly
independent! Try to look for simplifications before you
differentiate.
Fundamental solution set: If f, g, h are solutions and
linearly independent.
Largest interval of existence: First make sure the
leading coefficient equals to 1. Then look at the domain
of each term. For each domain, consider the part of the
interval which contains the initial condition. Finally,
intersect the intervals and change any brackets to
parentheses. Harmonic oscillator:
my 00 + by 0 + ky = 0 (m = inertia, b = damping,
k = stiffness)

Systems of differential equations

To solve x0 = Ax:
x(t) = Ae1 t v1 + Be2 t v2 + e3 t v3 (i are your
eigenvalues, vi are your eigenvectors)
Fundamental matrix: Matrix whose columns are the
solutions, without the constants (the columns are
solutions and linearly independent)
Complex eigenvalues If = + i, and v = a + ib.


Then: x(t) = A et cos(t)a et sin(t)b +


B et sin(t)a + et cos(t)b
Notes: You only need to consider one complex
eigenvalue. For real eigenvalues, use the formula
1
above. Also, a+bi
= aabi
2 +b2
Generalized eigenvectors If you only find one
eigenvector v (even though there are supposed to be 2),
then solve the following equation for u:
(A I)(u) = v (one solution is enough).

Then: x(t) = Aet v + B tet v + et u
Undetermined coefficients First find hom. solution.
Then for xp , just like regular undetermined
coefficients, except that instead of guessing
xp (t) = aet +b cos(t), you guess aet + b cos(t),
a
where a = 1 is a vector. Then plug into
a2
x0 = Ax + f and solve for a etc.
Variation of parameters First hom. solution
xh (t) = Ax1 (t) + Bx2 (t). Then sps
xp (t) = v1 (t)x1 (t) + v2 (t)x2 (t), then solve
0


f (t) v10 = f , where W
f (t) = x1 (t) | x2 (t) .
W
v2

1
f (t)
Multiply both sides by W
, integrate and solve
for v1 (t), v2 (t), and plug back into xp . Finally,
x = xh + xp
P
An t n
Matrix exponential eAt =
n=0 n! . To calculate
eAt , either diagonalize:
A = P DP 1 eAt = P eDt P 1 , where eDt is a
diagonal matrix with diag. entries ei t . Or if A only
has one eigenvalue with multiplicity m, use
P
(AI)n tn
. Solution of x0 = Ax
eAt = et m1
n=0
n!
At
is then x(t) = e c, where c is a constant vector.

Coupled mass-spring system

Case N = 2


2
1
1
2
Proper frequencies: Eigenvalues of A are:
Equation: x00 = Ax, A =

= 1, 3, then proper frequencies i, 3i (

square roots of eigenvalues)

 " 3 #

sin 3

Proper modes: v1 =
= 23 ,
sin 2 3
2
"
#



3
sin 2 3

2

v2 =
=
sin 4 3
23
Case N = 3

2
1
0
2
1
Equation:
= Ax, A = 1
0
1
2
Proper frequencies: Eigenvalues of A:

= 2, 2 2, 2 + 2, then proper frequencies



p
p


2i,
2 + 2 i,
2 2 i
x00

Proper modes:

2

sin 4

sin 2 4

2

v1 = sin 2 4
= 1 , v2 = sin 4 4
=
2
sin 3 4
sin 6 4


2

2
sin 3 4

1
2
0 , v3 = sin 6 =
1

4


2
1
sin 9 4
2
General case (just in case!)
Equation: x00 = Ax,

2
1
0

0
1
2
1
0

0
1
2
1
0

A= .

.
.
.
.
.
..
..
..
..
..
..

0
0

1
2
1
0
0
0
0
1
2


k
Proper frequencies: 2i sin 2(N +1) ,
k = 1, 2, N

sin

 N +1 
sin 2k

N +1

Proper modes: vk =

..

.

N k
sin N +1

Partial differential equations

Full Fourier
on
(T, T ):



P series: f defined
mx
f (x) e
+ bm sin mx
,
m=0 am cos
T
T
where: R
T
1
a0 = 2T
T f (x)dx
R


T
am = T1 T f (x) cos mx
T
b0 = 0 R


T
bm = T1 T
f (x) sin mx
T
Cosine series:
f defined on (0,
T ):
P
mx
f (x) e
, where:
m=0 am cos
T
RT
2
a0 = 2T 0 f (x)dx (not a typo)
R


am = T2 0T f (x) cos mx
T
Sine series:
(0, T
):
P f defined on mx
f (x) e
, where:
m=0 bm sin
T
b0 = 0 R


bm = T2 0T f (x) sin mx
T
R
R
Tabular integration: (IBP: f 0 g = f g f g 0 ) To
R
integrate f (t)g(t)dt where f is a polynomial, make
a table whose first row is f (t) and g(t). Then
differentiate f as many times until you get 0, and
antidifferentiate as many times until it aligns with the 0
for f . Then multiply the diagonal terms and do + first
term second term etc.
Orthogonality formulas:
RT




cos mx
sin nx
dx = 0
T
T
RT




T
mx
nx
cos T
cos T dx = 0 if m 6= n
RT




T
mx
nx
sin
sin
dx = 0 if m 6= n
T
T
T
Convergence: Fourier series F goes to f (x) is f is
continuous at x, and if f has a jump at x, F goes to the
average of the jumps. Finally, at the endpoints, F goes
to average of the left/right endpoints.
Heat/Wave equations:
Step 1: Suppose u(x, t) = X(x)T (t), plug this into
PDE, and group X-terms and T -terms. Then

X 00 (x)
X(x)

= , so X 00 = X. Then find a differential

equation for T . Note: If you have an -term, put it
with T .
Step 2: Deal with X 00 = X. Use boundary
conditions to find X(0) etc. (if you have u
, you
x
might have X 0 (0) instead of X(0)).
Step 3: Case 1: = 2 , then
X(x) = Aex + Bex , then find = 0,
contradiction. Case 2: = 0, then X(x) = Ax + B,
then eihter find X(x) = 0 (contradiction), or find
X(x) = A. Case 3: = 2 , then
X(x) = A cos(x) + B sin(x). Then solve for ,
usually = m
. Also, if case 2 works, should find
T
cos, if case 2 doesnt work, should find sin.
Finally, = 2 , and X(x) = whatever you found in
2) w/o the constant.
Step 4: Solve for T (t) with the you found.
Remember that for the heat equation:
t
g
T 0 = T T (t) = A
m e . And for the wave
equation:
g
g
T 00 = T T (t) = A
m cos(t) + Bm sin(t).
P
Step 5: Then u(x, P
t) =
T
(t)X(x)
(if case 2
m=0
works), u(x, t) =
m=1 T (t)X(x) (if case 2 doesnt
work!)
Step 6: Use u(x, 0), and plug in t = 0. Then use
Fourier cosine or sine series or just compare, i.e. if
f2 = 4,
u(x, 0) = 4 sin(2x) + 3 sin(3x), then A
f3 = 3, and A
g
A
m = 0 if m 6= 2, 3.
u(x, 0):
Step 7: (only for wave equation): Use u
t
Differentiate Step 5 with respect to t and set t = 0.
Then use Fourier cosine or series or compare
Nonhomogeneous
heat equation:

2u
u
+
P (x)
=

t
x2
u(0,
t)
=
U
,
u(L, t) = U 2
1

u(x, 0) = f (x)
Then u(x,
w(x, t), where: i
h t) = v(x) +
R R
x
v(x) = U2 U1 + 0L 0z 1 P (s)dsdz L
+ U1
RxRz 1
P
(s)dsdz
and
w(x,
t)
solves
the
hom.
eqn:
0 0
w
2w

= x2
t
w(0, t) = 0,
w(L, t) = 0

u(x, 0) = f (x) v(x)

DAlemberts formula: ONLY works for wave

equation and < x < : u(x, t) =

R x+t
1
1
(f (x + t) + f (x t)) + 2
xt g(s)ds,
2
where
utt = 2 uxx , u(x, 0) = f (x), u
u(x, 0) = g(x).
t

The integral just means antidifferentiate and plug in.

Laplace equation:
Same as for Heat/Wave, but T (t) becomes Y (y), and

we get Y 00 (y) = Y (y). Also, instead of writing

y + B
y , write
g
g
Y (y) = A
me
me
g
g
Y (y) = Am cosh(y) + B
m sinh(y). Remember
cosh(0) = 1, sinh(0) = 0