Theoretical Computer Science Cheat Sheet

Definitions Series
f (n) = O(g(n)) i positive c, n0 such that
n
n
n
n(n + 1) n(n + 1)(2n + 1) n2 (n + 1)2
0 f (n) cg(n) n n0 . i= , i2 = , i3 = .
i=1
2 i=1
6 i=1
4
f (n) = (g(n)) i positive c, n0 such that
In general:
f (n) cg(n) 0 n n0 . [ ]
n
1 n
( )
f (n) = (g(n)) i f (n) = O(g(n)) and im = (n + 1)m+1 1 (i + 1)m+1 im+1 (m + 1)im
i=1
m+1 i=1
f (n) = (g(n)).
( )
1 m+1
n1 m
f (n) = o(g(n)) i limn f (n)/g(n) = 0. im = Bk nm+1k .
i=1
m + 1 k
k=0
lim an = a i > 0, n0 such that
n Geometric series:
|an a| < , n n0 .
n
cn+1 1 1 c
sup S least b R such that b s, ci = , c = 1, ci = , ci = , |c| < 1,
c1 1c 1c
s S. i=0 i=0 i=1

n

ncn+2 (n + 1)cn+1 + c c
inf S greatest b R such that b ici = , c = 1, ici = , |c| < 1.
s, s S. i=0
(c 1)2 i=0
(1 c)2
Harmonic series:
lim inf an lim inf{ai | i n, i N}.
n n n
1
n
n(n + 1) n(n 1)
Hn = , iHi = Hn .
lim sup an lim sup{ai | i n, i N}. i=1
i i=1
2 4
n ( ) ( )( )
n n
(n) n i n+1 1
k Combinations: Size k sub- Hi = (n + 1)Hn n, Hi = Hn+1 .
sets of a size n set. i=1 i=1
m m+1 m+1
[n] ( ) n ( ) ( ) ( )
k Stirling numbers (1st kind): n n! n n n
n
Arrangements of an n ele- 1. = , 2. =2 , 3. = ,
k (n k)!k! k k nk
ment set into k cycles. ( ) ( ) k=0 ( ) ( ) ( )
{n} n n n1 n n1 n1
4. = , 5. = + ,
k Stirling numbers (2nd kind): k k k1 k k k1
Partitions of an n element ( )( ) ( )( ) n ( ) ( )
n m n nk r+k r+n+1
set into k non-empty sets. 6. = , 7. = ,
n m k k mk k n
n ( ) ( ) n ( )( ) ( )
k=0
k 1st order Eulerian numbers: k n+1 r s r+s
Permutations 1 2 . . . n on 8. = , 9. = ,
m m+1 k nk n
{1, 2, . . . , n} with k ascents. ( )
k=0 ( ) k=0 { } { }
n n kn1 n n
2nd order Eulerian numbers. 10. = (1)k , 11. = = 1,
k k k 1 n
Cn Catalan Numbers: Binary { } { } { } { }
n n n1 n1
trees with n + 1 vertices. 12. = 2n1
1, 13. = k + ,
2 k k k1
[ ] [ ] [ ] [ ] { }
n n n n n
14. = (n 1)!, 15. = (n 1)!Hn1 , 16. = 1, 17. ,
1 2 n k k
[ ] [ ] [ ] { } [ ] ( ) n [ ] ( )
n n1 n1 n n n n 1 2n
18. = (n 1) + , 19. = = , 20. = n!, 21. Cn = ,
k k k1 n1 n1 2 k n+1 n
k=0
n n n n n n1 n1
22. = = 1, 23. = , 24. = (k + 1) + (n k) ,
0 n1 k n1k k k k1
{ ( )
0 1 if k = 0, n n n+1
25. = 26. = 2n n 1, 27. = 3n (n + 1)2n + ,
k 0 otherwise 1 2 2
n ( ) m ( ) { } n ( )
n x+k n n+1 n n k
28. xn = , 29. = (m + 1 k)n (1)k , 30. m! = ,
k n m k m k nm
k=0
n { }( ) k=0

k=0
n n nk n n
31. = (1) nkm
k!, 32. = 1, 33. = 0 for n = 0,
m k m 0 n
k=0

n
n n1 n1 n (2n)n
34. = (k + 1) + (2n 1 k) , 35. = ,
k k k1 k 2n
{ } n ( ) { } ( ){ } n {
k=0
}
x n x+n1k n+1 n k k
36. = , 37. = = (m + 1)nk ,
xn k 2n m+1 k m m
k=0 k k=0
 Theoretical Computer Science Cheat Sheet
Identities Cont. Trees
[ ] [ n ]( k ) n [
] n [ ] [ n ]
( )
n+1 k 1 k x n x+k Every tree with n
38. = = nnk = n! , 39. = , vertices has n 1
m+1 k m m k! m xn k 2n
{ } ( ){ k } k=0 k=0 [ ] [ k=0 ]( ) edges.
n n k+1 nk n n+1 k
40. = (1) , 41. = (1)mk , Kraft inequal-
m k m+1 m k+1 m
{ k
} m { } [ ]
k
m [ ] ity: If the depths
m+n+1 n+k m+n+1 n+k of the leaves of
42. = k , 43. = k(n + k) ,
m k m k a binary tree are
( ) { k=0}[ ] ( ) [ ]{ } k=0
n n+1 k n n+1 k d1 , . . . , d n :
44. = (1)mk , 45. (n m)! = (1)mk , for n m, n

{
m
}k (
k+1 m
)( )[ ]
m
[ k ]
k+1 m 2di 1,
n mn m+n m+k n (m n)(m + n){ m + k } i=1
46. = , 47. = ,
nm m+k n+k k nm m+k n+k k and equality holds
{ }( k ) { }{ }( ) [ ](k ) [ ][ ]( )
n +m k nk n n +m k nk n only if every in-
48. = , 49. = . ternal node has 2
+m m k +m m k
k k
sons.

Recurrences
Master method: ( ) Generating functions:
T (n) = aT (n/b) + f (n), a 1, b > 1 1 T (n) 3T (n/2) = n 1. Multiply both sides of the equa-
( )
3 T (n/2) 3T (n/4) = n/2 tion by xi .
If > 0 such that f (n) = O(nlogb a )
.. .. .. 2. Sum both sides over all i for
then
. . . which the equation is valid.
T (n) = (nlogb a ). ( )
3log2 n1
T (2) 3T (1) = 2 3. Choose a generatingfunction

If f (n) = (nlogb a ) then G(x). Usually G(x) = i=0 xi gi .
T (n) = (nlogb a log2 n). Let m = log2 n. Summing the left side 3. Rewrite the equation in terms of
we get T (n) 3m T (1) = T (n) 3m = the generating function G(x).
If > 0 such that f (n) = (nlogb a+ ), T (n) nk where k = log2 3 1.58496.
and c < 1 such that af (n/b) cf (n) 4. Solve for G(x).
Summing the right side we get
5. The coecient of xi in G(x) is gi .
for large n, then n
m1 ( )i
m1
T (n) = (f (n)). 3i
= n 3
. Example:
2 i 2
i=0 i=0
gi+1 = 2gi + 1, g0 = 0.
Substitution (example): Consider the
Let c = 23 . Then we have Multiply
and sum:
following recurrence
i

m1 ( m ) gi+1 xi = 2gi xi + xi .
Ti+1 = 22 Ti2 , T1 = 2. i c 1
n c =n i0 i0 i0
c1
Note that Ti is always a power of two. i=0
We choose G(x) = i0 xi gi . Rewrite
Let ti = log2 Ti . Then we have = 2n(clog2 n 1)
in terms of G(x):
ti+1 = 2i + 2ti , t1 = 1.
= 2n(c(k1) logc n 1) G(x) g0
= 2G(x) + xi .
Let ui = ti /2i . Dividing both sides of = 2nk 2n, x
i0
the previous equation by 2i+1 we get
ti+1 2i ti and so T (n) = 3n 2n. Full history re-
k Simplify:
= + i. G(x) 1
2 i+1 2 i+1 2 currences can often be changed to limited = 2G(x) + .
history ones (example): Consider x 1x
Substituting we find
ui+1 = 21 + ui , u1 = 12 ,
i1
Solve for G(x):
Ti = 1 + Tj , T0 = 1. x
G(x) = .
which is simply ui = i/2. So we find j=0
(1 x)(1 2x)
i1
that Ti has the closed form Ti = 2i2 . Note that Expand this (
using partial fractions:
)
Summing factors (example): Consider
i
2 1
the following recurrence Ti+1 = 1 + Tj . G(x) = x
1 2x 1 x
T (n) = 3T (n/2) + n, T (1) = 1. j=0

Subtracting we find
Rewrite so that all terms involving T
i
i1 = x 2 2i xi xi
are on the left side Ti+1 Ti = 1 + Tj 1 Tj i0 i0
T (n) 3T (n/2) = n.
j=0 j=0
= (2i+1 1)xi+1 .
Now expand the recurrence, and choose = Ti . i0
a factor which makes the left side tele-
scope And so Ti+1 = 2Ti = 2i+1 . So gi = 2i 1.
 Theoretical Computer Science Cheat Sheet

3.14159, e 2.71828, 0.57721, = 1+ 5
2 1.61803, = 1 5
2 .61803

i 2i pi General Probability
1 2 2 Bernoulli Numbers (Bi = 0, odd i = 1): Continuous distributions: If
b
2 4 3 B0 = 1, B1 = 12 , B2 = 16 , B4 = 30 1
, Pr[a < X < b] = p(x) dx,
3 8 5 B6 = 42 , B8 = 30 , B10 = 66 .
1 1 5 a

4 16 7 Change of base, quadratic formula: then p is the probability density function of

X. If
5 32 11 loga x b b2 4ac Pr[X < a] = P (a),
logb x = , .
6 64 13 loga b 2a
then P is the distribution function of X. If
7 128 17 Eulers number e: P and p both exist then
e = 1 + 12 + 16 + 24 1 1
+ 120 + a
8 256 19 ( )
x n P (a) = p(x) dx.
9 512 23 lim 1 + = ex .
n n
10 1,024 29 ( )n ( )n+1 Expectation: If X is discrete

1 + n1 < e < 1 + n1 .
11 2,048 31 ( ) E[g(X)] = g(x) Pr[X = x].
( )n e 11e 1 x
12 4,096 37 1 + n1 = e + 2
O .
2n 24n n3 If X continuous
then
13 8,192 41
Harmonic numbers:
14 16,384 43 E[g(X)] = g(x)p(x) dx = g(x) dP (x).
1, 32 , 11 25 137 49 363 761 7129
6 , 12 , 60 , 20 , 140 , 280 , 2520 , . . .

15 32,768 47 Variance, standard deviation:
16 65,536 53 ln n < Hn < ln n + 1, VAR[X] = E[X 2 ] E[X]2 ,
( )
17 131,072 59 1 = VAR[X].
Hn = ln n + + O .
18 262,144 61 n For events A and B:
19 524,288 67 Factorial, Stirlings approximation: Pr[A B] = Pr[A] + Pr[B] Pr[A B]
20 1,048,576 71 1, 2, 6, 24, 120, 720, 5040, 40320, 362880, ... Pr[A B] = Pr[A] Pr[B],
21 2,097,152 73 ( )n ( ( )) i A and B are independent.
n 1
22 4,194,304 79 n! = 2n 1+ . Pr[A B]
e n Pr[A|B] =
23 8,388,608 83 Pr[B]
Ackermanns
function and inverse: For random variables X and Y :
24 16,777,216 89 2j i=1
25 33,554,432 97 a(i, j) = a(i 1, 2) j=1 E[X Y ] = E[X] E[Y ],

26 67,108,864 101 a(i 1, a(i, j 1)) i, j 2 if X and Y are independent.

27 134,217,728 103 (i) = min{j | a(j, j) i}. E[X + Y ] = E[X] + E[Y ],

E[cX] = c E[X].
28 268,435,456 107 Binomial distribution:
( )
n k nk Bayes theorem:
29 536,870,912 109 Pr[X = k] = p q , q = 1 p, Pr[B|Ai ] Pr[Ai ]
30 1,073,741,824 113 k Pr[Ai |B] = n .
n ( ) j=1 Pr[Aj ] Pr[B|Aj ]
31 2,147,483,648 127 n k nk
E[X] = k p q = np. Inclusion-exclusion:
32 4,294,967,296 131 k [ n ] n
k=1
Poisson distribution: Pr Xi = Pr[Xi ] +
Pascals Triangle
e k i=1 i=1
1 Pr[X = k] = , E[X] = . [ ]
k!
n k
11 Normal (Gaussian) distribution: (1)k+1 Pr X ij .
k=2 ii <<ik j=1
121 1
e(x) /2 , E[X] = .
2 2
p(x) = Moment inequalities:
1331 2 [ ] 1
14641 The coupon collector: We are given a Pr |X| E[X] ,

1 5 10 10 5 1 random coupon each day, and there are n [  ] 1
dierent types of coupons. The distribu- Pr X E[X] 2 .
1 6 15 20 15 6 1 tion of coupons is uniform. The expected
1 7 21 35 35 21 7 1 Geometric distribution:
number of days to pass before we to col-
lect all n types is Pr[X = k] = pq k1 , q = 1 p,
1 8 28 56 70 56 28 8 1

1 9 36 84 126 126 84 36 9 1 nHn . 1
E[X] = kpq k1 = .
p
1 10 45 120 210 252 210 120 45 10 1 k=1
 Theoretical Computer Science Cheat Sheet
Trigonometry Matrices More Trig.
Multiplication: C
(0,1)
n
C = A B, ci,j = ai,k bk,j . a
b b h
(cos , sin ) k=1
C Determinants: det A = 0 i A is non-singular.
A
(-1,0) (1,0) A c B
det A B = det A det B, Law of cosines:
c a
(0,-1)
n
c2 = a2 +b2 2ab cos C.
B det A = sign()ai,(i) .
i=1 Area:
Pythagorean theorem:
C 2 = A2 + B 2 . 2 2 and 3 3 determinant:
 
a b A = 12 hc,
 
Definitions:  c d  = ad bc, = 12 ab sin C,
sin a = A/C, cos a = B/C,  
a b c        c2 sin A sin B
        = .
csc a = C/A, sec a = C/B, d e f  = gb c  ha c  + ia b 2 sin C
  e f  d f  d e
sin a A cos a B g h i  Herons formula:
tan a = = , cot a = = .
cos a B sin a A aei + bf g + cdh
Area, radius of inscribed circle: =
ceg f ha ibd. A = s sa sb sc ,
1 AB s = 12 (a + b + c),
2 AB, . Permanents:
A+B+C n
sa = s a,
Identities: perm A = ai,(i) .
i=1
sb = s b,
1 1
sin x = , cos x = , Hyperbolic Functions sc = s c.
csc x sec x
1 Definitions: More identities:
tan x = , sin2 x + cos2 x = 1,
cot x ex ex e +e x x
x 1 cos x
sinh x = , cosh x = , sin 2 = ,
1 + tan2 x = sec2 x, 1 + cot2 x = csc2 x, 2 2 2
e e
x x
1
( ) tanh x = x , csch x = , 1 + cos x
sin x = cos 2 x , sin x = sin( x), e + ex sinh x cos x2 = ,
1 1 2
( ) sech x = , coth x = .
cos x = cos( x), tan x = cot 2 x , cosh x tanh x 1 cos x
tan x2 = ,
Identities: 1 + cos x
cot x = cot( x), csc x = cot x2 cot x, 1 cos x
cosh2 x sinh2 x = 1, tanh2 x + sech2 x = 1, = ,
sin x
sin(x y) = sin x cos y cos x sin y,
sin x
coth2 x csch2 x = 1, sinh(x) = sinh x, = ,
cos(x y) = cos x cos y sin x sin y, 1 + cos x

cosh(x) = cosh x, tanh(x) = tanh x, 1 + cos x
tan x tan y cot x2 = ,
tan(x y) = , 1 cos x
1 tan x tan y sinh(x + y) = sinh x cosh y + cosh x sinh y,
1 + cos x
cot x cot y 1 = ,
cot(x y) = , cosh(x + y) = cosh x cosh y + sinh x sinh y, sin x
cot x cot y sin x
2 tan x sinh 2x = 2 sinh x cosh x, = ,
sin 2x = 2 sin x cos x, sin 2x =
1 + tan2 x
, 1 cos x
cosh 2x = cosh2 x + sinh2 x, eix eix
cos 2x = cos2 x sin2 x, cos 2x = 2 cos2 x 1, sin x = ,
2i
1 tan2 x cosh x + sinh x = ex , cosh x sinh x = ex , eix + eix
cos 2x = 1 2 sin2 x, cos 2x = , cos x = ,
1 + tan2 x
(cosh x + sinh x)n = cosh nx + sinh nx, n Z, 2
2 tan x cot2 x 1 eix eix
tan 2x = 2 , cot 2x = , 2 sinh2 x2 = cosh x 1, 2 cosh2 x2 = cosh x + 1. tan x = i ix ,
1 tan x 2 cot x e + eix
sin(x + y) sin(x y) = sin2 x sin2 y, e2ix 1
sin cos tan . . . in mathematics = i 2ix ,
e +1
cos(x + y) cos(x y) = cos2 x sin2 y. you dont under- sinh ix
0 0
1
0 sin x = ,
1 3 3 stand things, you i
Eulers equation: 6 2

2

3 just get used to
eix = cos x + i sin x, ei
= 1. 2 2
cos x = cosh ix,
4 2 2 1 them.
v2.02 1994
c by Steve Seiden

3 1
J. von Neumann tan x =
tanh ix
.
3 2 2 3 i
sseiden@acm.org
2 1 0
http://www.csc.lsu.edu/~seiden
 Theoretical Computer Science Cheat Sheet
Number Theory Graph Theory
The Chinese remainder theorem: There ex- Definitions: Notation:
ists a number C such that: Loop An edge connecting a ver- E(G) Edge set
tex to itself. V (G) Vertex set
C r1 mod m1 c(G) Number of components
Directed Each edge has a direction.
.. .. .. Simple Graph with no loops or G[S] Induced subgraph
. . .
multi-edges. deg(v) Degree of v
C rn mod mn (G) Maximum degree
Walk A sequence v0 e1 v1 . . . e v .
if mi and mj are relatively prime for i = j. Trail A walk with distinct edges. (G) Minimum degree
Path A trail with distinct (G) Chromatic number
Eulers function: (x) is the number of
vertices. E (G) Edge chromatic number
positive integersnless than x relatively Connected A graph where there exists Gc Complement graph
prime to x. If i=1 pei i is the prime fac-
a path between any two Kn Complete graph
torization of x then
n
vertices. Kn1 ,n2 Complete bipartite graph
(x) = pei i 1 (pi 1). Component A maximal connected
r(k, ) Ramsey number
i=1
subgraph. Geometry
Eulers theorem: If a and b are relatively Tree A connected acyclic graph.
prime then Projective coordinates: triples
Free tree A tree with no root.
1 a(b) mod b. (x, y, z), not all x, y and z zero.
DAG Directed acyclic graph.
Eulerian Graph with a trail visiting (x, y, z) = (cx, cy, cz) c = 0.
Fermats theorem:
each edge exactly once. Cartesian Projective
1 ap1 mod p.
Hamiltonian Graph with a cycle visiting (x, y) (x, y, 1)
The Euclidean algorithm: if a > b are in- each vertex exactly once. y = mx + b (m, 1, b)
tegers then Cut A set of edges whose re- x=c (1, 0, c)
gcd(a, b) = gcd(a mod b, b). moval increases the num- Distance formula, Lp and L
n ber of components. metric:
If i=1 pei i is the prime factorization of x
then Cut-set A minimal cut. (x1 x0 )2 + (y1 y0 )2 ,
n
pei i +1 1 Cut edge A size 1 cut. [ ]1/p
S(x) = d= . k-Connected A graph connected with |x1 x0 |p + |y1 y0 |p ,
pi 1 [ ]
d|x i=1 the removal of any k 1 lim |x1 x0 |p + |y1 y0 |p
1/p
.
Perfect Numbers: x is an even perfect num- vertices. p

ber i x = 2n1 (2n 1) and 2n 1 is prime. k-Tough S V, S = we have Area of triangle (x0 , y0 ), (x1 , y1 )
Wilsons theorem: n is a prime i k c(G S) |S|. and (x2 , y2 ):
 
(n 1)! 1 mod n. k-Regular A graph where all vertices  x1 x0 y1 y0 
1  .
2 abs  x x 
have degree k. 2 0 y2 y0
Mobius
inversion: k-Factor A k-regular spanning
1 if i = 1. Angle formed by three points:
subgraph.
0 if i is not square-free.
(i) = Matching A set of edges, no two of
r
(1) if i is the product of (x2 , y2 )
r distinct primes. which are adjacent.
2
Clique A set of vertices, all of
If which are adjacent.
G(a) = F (d), (0, 0) 1 (x1 , y1 )
Ind. set A set of vertices, none of
d|a
which are adjacent. (x1 , y1 ) (x2 , y2 )
cos = .
then (a) Vertex cover A set of vertices which 1 2
F (a) = (d)G . cover all edges. Line through two points (x0 , y0 )
d
d|a Planar graph A graph which can be em- and (x1 , y1 ):
 
beded in the plane.  x y 1
Prime numbers:  
ln ln n Plane graph An embedding of a planar  x0 y0 1  = 0.
pn = n ln n + n ln ln n n + n  
ln n graph.  x1 y1 1 
( )
n Area of circle, volume of sphere:
+O , deg(v) = 2m.
ln n A = r2 , V = 43 r3 .
vV
n n 2!n
(n) = + + If G is planar then n m + f = 2, so
ln n (ln n)2 (ln n)3 If I have seen farther than others,
( ) f 2n 4, m 3n 6.
n it is because I have stood on the
+O . Any planar graph has a vertex with de- shoulders of giants.
(ln n)4
gree 5. Issac Newton
 Theoretical Computer Science Cheat Sheet
Calculus
Wallis identity: Derivatives:
2 2 4 4 6 6
=2 d(cu) du d(u + v) du dv d(uv) dv du
1 3 3 5 5 7 1. =c , 2. = + , 3. =u +v ,
dx dx dx dx dx dx dx dx
Brounckers continued fraction expansion: ( du ) ( dv )
12 d(un ) du d(u/v) v dx u dx cu
d(e ) du
4. = nun1 , 5. = , 6. = cecu ,
4 =1+ 32 dx dx dx v2 dx dx
2+ 52
2+
2+ 72 d(cu ) du d(ln u) 1 du
2+
7. = (ln c)cu , 8. = ,
dx dx dx u dx
Gregrorys series:
4 =1 3 +
1 1 1
+ 1 d(sin u) du d(cos u) du
5 7 9 9. = cos u , 10. = sin u ,
dx dx dx dx
Newtons series:
d(tan u) du d(cot u) du
1 1 13 11. = sec2 u , 12. = csc2 u ,
dx dx dx dx
6 = 2 + 2 3 23 + 2 4 5 25 +

d(sec u) du d(csc u) du
Sharps series: 13. = tan u sec u , 14. = cot u csc u ,
dx dx dx dx
1 ( 1 1 1 ) d(arcsin u) 1 du d(arccos u) 1 du

= 1 1 + 2 3 + 15. = , 16. = ,
6
3 3 3 3 5 3 7 dx 1 u dx
2 dx 1 u2 dx
Eulers series: d(arctan u) 1 du d(arccot u) 1 du
17. = , 18. = ,
dx 1 + u2 dx dx 1 + u2 dx
2
6 = 1
12 + 1
22 + 1
32 + 1
42 + 1
52 + d(arcsec u) 1 du d(arccsc u) 1 du
2
19. = , 20. = ,
8 = 1
12 + 1
32 + 1
52 + 1
72 + 1
92 + dx u 1 u dx
2 dx u 1 u dx
2

2
= 1
1
+ 1
1
+ 1
d(sinh u) du d(cosh u) du
12 12 22 32 42 52 21. = cosh u , 22. = sinh u ,
dx dx dx dx
Partial Fractions d(tanh u) du d(coth u) du
23. = sech2 u , 24. = csch2 u ,
Let N (x) and D(x) be polynomial func- dx dx dx dx
tions of x. We can break down d(sech u) du d(csch u) du
N (x)/D(x) using partial fraction expan- 25. = sech u tanh u , 26. = csch u coth u ,
dx dx dx dx
sion. First, if the degree of N is greater
than or equal to the degree of D, divide d(arcsinh u) 1 du d(arccosh u) 1 du
27. = , 28. = ,
N by D, obtaining dx 1 + u dx
2 dx u 1 dx
2

N (x) N (x) d(arctanh u) 1 du d(arccoth u) 1 du

= Q(x) + , 29. = , 30. = 2 ,
D(x) D(x) dx 1 u2 dx dx u 1 dx
where the degree of N is less than that of d(arcsech u) 1 du d(arccsch u) 1 du
31. = , 32. = .
D. Second, factor D(x). Use the follow- dx u 1 u2 dx dx |u| 1 + u2 dx
ing rules: For a non-repeated factor: Integrals:
N (x) A N (x)
= + ,
(x a)D(x) xa D(x) 1. cu dx = c u dx, 2. (u + v) dx = u dx + v dx,
where [ ]
N (x) 1 1
A= . 3. n
x dx = xn+1 , n = 1, 4. dx = ln x, 5. ex dx = ex ,
D(x) x=a
n+1 x

For a repeated factor: dx dv du
6. = arctan x, 7. u dx = uv v dx,
N (x)
m1
Ak N (x) 1 + x2 dx dx
= + ,
(x a) D(x)
m (x a)mk D(x) 8. sin x dx = cos x, 9. cos x dx = sin x,
k=0

where [ ( )]
1 dk N (x) 10. tan x dx = ln | cos x|, 11. cot x dx = ln | cos x|,
Ak = .
k! dxk D(x) x=a
12. sec x dx = ln | sec x + tan x|, 13. csc x dx = ln | csc x + cot x|,
The reasonable man adapts himself to the
world; the unreasonable persists in trying
to adapt the world to himself. Therefore 14. arcsin xa dx = arcsin xa + a2 x 2 , a > 0,
all progress depends on the unreasonable.
George Bernard Shaw
 Theoretical Computer Science Cheat Sheet
Calculus Cont.

15. arccos x
a dx = arccos x
a a2 x2 , a > 0, 16. arctan xa dx = x arctan xa a
2 ln(a2 + x2 ), a > 0,

( ) ( )
17. sin2 (ax)dx = 1
2a ax sin(ax) cos(ax) , 18. cos2 (ax)dx = 1
2a ax + sin(ax) cos(ax) ,

19. sec2 x dx = tan x, 20. csc2 x dx = cot x,

sinn1 x cos x n 1 cosn1 x sin x n 1
21. sinn x dx = + sinn2 x dx, 22. cosn x dx = + cosn2 x dx,
n n n n

tann1 x cotn1 x
23. n
tan x dx = tan n2
x dx, n = 1, 24. cot x dx =
n
cotn2 x dx, n = 1,
n1 n1

tan x secn1 x n 2
25. secn x dx = + secn2 x dx, n = 1,
n1 n1

cot x cscn1 x n 2
26. cscn x dx = + cscn2 x dx, n = 1, 27. sinh x dx = cosh x, 28. cosh x dx = sinh x,
n1 n1

 
29. tanh x dx = ln | cosh x|, 30. coth x dx = ln | sinh x|, 31. sech x dx = arctan sinh x, 32. csch x dx = ln tanh x2  ,

33. 2
sinh x dx = 1
4 sinh(2x) 1
2 x, 34. 2
cosh x dx = 1
4 sinh(2x) + 1
2 x, 35. sech2 x dx = tanh x,

36. arcsinh x
a dx = x arcsinh x
a x 2 + a2 , a > 0, 37. arctanh xa dx = x arctanh xa + a
2 ln |a2 x2 |,

x
x arccosh x2 + a2 , if arccosh xa > 0 and a > 0,
arccosh xa dx = a
38.
x arccosh x + x2 + a2 , if arccosh x < 0 and a > 0,
a a
( )
dx
39. = ln x + a2 + x2 , a > 0,
a2 + x 2

dx 2 x2 dx = x
2
40. = 1
a arctan x
a , a > 0, 41. a 2 a2 x2 + a2 arcsin xa , a > 0,
a2 + x2
4
42. (a2 x2 )3/2 dx = x8 (5a2 2x2 ) a2 x2 + 3a8 arcsin xa , a > 0,
 
dx dx 1 a + x dx x
43. x
= arcsin a , a > 0, 44. = ln  , 45. = ,
a x
2 2 a x
2 2 2a  ax  (a x )
2 2 3/2
a a2 x 2
2
   
2   dx  
46. a2 x2 dx = x2 a2 x2 a2 ln x + a2 x2  , 47. = ln x + x2 a2  , a > 0,
x 2 a2
 
dx 1  x  2(3bx 2a)(a + bx)3/2
48. 2
= ln   , 49. x a + bx dx = ,
ax + bx a a + bx 15b2
 
a + bx 1 x 1  a + bx a 
50. dx = 2 a + bx + a dx, 51. dx = ln    , a > 0,
x x a + bx  a + bx 2 a + bx + a 
2  a + a2 x2 
a x2  
52. dx = a x a ln 
2 2 , 53. x a2 x2 dx = 13 (a2 x2 )3/2 ,
x  x 
 
 a + a2 x2 
4 dx  
54. x2 a2 x2 dx = x8 (2x2 a2 ) a2 x2 + a8 arcsin xa , a > 0, 55. = a1 ln  ,
a x
2 2  x 
2
x dx x dx 2
56. = a2 x2 , 57. = x2 a2 x2 + a2 arcsin a, x
a > 0,
a x
2 2
  a 2 x2

 a + a2 + x 2 
a2 + x 2   x 2 a2
58. dx = a2 + x2 a ln  , 59. dx = x2 a2 a arccos |x| a
, a > 0,
x  x  x
 
dx  x 
60. x x2 a2 dx = 13 (x2 a2 )3/2 , 61. = a1 ln  ,
x x2 + a2 a + a2 + x 2 
 Theoretical Computer Science Cheat Sheet
Calculus Cont. Finite Calculus

dx dx x 2 a2 Dierence, shift operators:
62. 1 a
= a arccos |x| , a > 0, 63. = ,
x x2 a2 x2 x2 a2 a2 x f (x) = f (x + 1) f (x),

x dx x 2 a2 (x2 + a2 )3/2 E f (x) = f (x + 1).
64. = x 2 a2 , 65. dx = ,
x 2 a2 
x4  3a2 x3 Fundamental Theorem:

 2ax + b b2 4ac 


1   f (x) = F (x) f (x)x = F (x) + C.
ln   , if b2 > 4ac,
dx b 4ac
2  2ax + b + b 4ac 
2
66. =
b
b1
ax2 + bx + c f (x)x = f (i).


2
arctan
2ax + b
, if b2 < 4ac, a
4ac b2 4ac b2 i=a
Dierences:
 
1   (cu) = cu, (u + v) = u + v,

ln 2ax + b + 2 a ax2 + bx + c , if a > 0,
dx a
67. = (uv) = uv + E vu,
ax2 + bx + c 1 2ax b
arcsin , if a < 0, (xn ) = nxn1 ,
a b2 4ac
(Hx ) = x1 , (2x ) = 2x ,
2ax + b 2 4ax b2 dx (x) ( x )
68. ax2 + bx + c dx = ax + bx + c + , (cx ) = (c 1)cx , m = m1 .
4a 8a 2
ax + bx + c
Sums:

x dx ax2 + bx + c b dx cu x = c u x,
69. = ,
2
ax + bx + c a 2a 2
ax + bx + c
(u + v) x = u x + v x,
 
1  2 c ax2 + bx + c + bx + 2c 

uv x = uv E vu x,

c ln   , if c > 0,
dx x  n 1

n+1
70. =
x x = xm+1 , x x = Hx ,
x ax + bx + c
2
1 bx + 2c
arcsin , if c < 0, x cx ( x
) ( x )
c |x| b2 4ac c x = c1 , m x = m+1 .
Falling Factorial Powers:
71. x3 x2 + a2 dx = ( 13 x2 15
2 2
a )(x2 + a2 )3/2 , xn = x(x 1) (x n + 1), n > 0,

x0 = 1,
72. xn sin(ax) dx = a1 xn cos(ax) + n
a xn1 cos(ax) dx,
1
xn = , n < 0,
(x + 1) (x + |n|)
73. n
x cos(ax) dx = 1 n
ax sin(ax) n
x n1
sin(ax) dx,
a xn+m = xm (x m)n .
Rising Factorial Powers:
xn eax
74. xn eax dx = n
xn1 eax dx,
a a xn = x(x + 1) (x + n 1), n > 0,
( )
ln(ax) 1 x0 = 1,
75. xn ln(ax) dx = xn+1 ,
n+1 (n + 1)2 1
xn = , n < 0,
xn+1
m (x 1) (x |n|)
76. xn (ln ax)m dx = (ln ax)m xn (ln ax)m1 dx.
n+1 n+1 xn+m = xm (x + m)n .
Conversion:
x1 = x1 = x1 xn = (1)n (x)n = (x n + 1)n
x2 = x2 + x1 = x2 x1 = 1/(x + 1)n ,
x3 = x3 + 3x2 + x1 = x3 3x2 + x1 xn = (1)n (x)n = (x + n 1)n
x4 = x4 + 6x3 + 7x2 + x1 = x4 6x3 + 7x2 x1 = 1/(x 1)n ,
n { } n { }
x5 = x5 + 15x4 + 25x3 + 10x2 + x1 = x5 15x4 + 25x3 10x2 + x1 n k n
xn = x = (1)nk xk ,
k k
x1 = x1 x1 = x1 k=1 k=1
n [ ]
2
x = 2
x +x 1 2
x = x x1
2
xn =
n
(1)nk xk ,
x3 = x + 3x2 + 2x1
3
x3 = x3 3x2 + 2x1 k
k=1
n [ ]
x4 = x4 + 6x3 + 11x2 + 6x1 x4 = x4 6x3 + 11x2 6x1 n n k
x = x .
x5 = x5 + 10x4 + 35x3 + 50x2 + 24x1 x5 = x5 10x4 + 35x3 50x2 + 24x1 k
k=1
 Theoretical Computer Science Cheat Sheet
Series
Taylors series: Ordinary power series:

(x a)i
(x a)2



f (x) = f (a) + (x a)f (a) + f (a) + = f (i) (a). A(x) = ai xi .
2 i=0
i!
i=0
Expansions:
Exponential power series:
1
= 1 + x + x2 + x3 + x4 + = xi ,
xi
1x i=0 A(x) = ai .
i!
1 i=0
= 1 + cx + c2 x2 + c3 x3 + = ci xi ,
1 cx i=0
Dirichlet power series:

1 ai
= 1 + x n
+ x 2n
+ x 3n
+ = xni , A(x) = .
1 xn i=1
ix
i=0
Binomial theorem:
x n ( )
= x + 2x + 3x + 4x +
2 3 4
= ixi , n nk k
(1 x)2 (x + y)n = x y .
n { }
i=0
k
n k!z k k=0
= x + 2 n 2
x + 3 n 3
x + 4 n 4
x + = in xi , Dierence of like powers:
k (1 z)k+1 i=0
k=0

n1
xi xn y n = (x y) xn1k y k .
ex = 1 + x + 12 x2 + 16 x3 + = ,
i=0
i! k=0

xi For ordinary power series:
ln(1 + x) = x 12 x2 + 13 x3 14 x4 = (1)i+1 ,

i
i=1 A(x) + B(x) = (ai + bi )xi ,

x i
1 i=0
ln = x + 12 x2 + 13 x3 + 14 x4 + = ,
1x i
i=1
xk A(x) = aik xi ,
x2i+1
sin x = x 3! x + 5! x 7! x + =
1 3 1 5 1 7
(1)i , i=k
(2i + 1)! k1

i=0

A(x) ai xi
x2i
i=0
= ai+k xi ,
cos x = 1 2! 1 2
x + 4! x 6!
1 4 1 6
x + = (1)i , xk i=0
i=0
(2i)!

A(cx) = ci ai xi ,
x2i+1
tan1 x = x 13 x3 + 15 x5 17 x7 + = (1)i , i=0
(2i + 1)
( )
i=0
n i A (x) = (i + 1)ai+1 xi ,
(1 + x)n = 1 + nx + n(n1) 2 x 2
+ = x, i=0
i
( )
i=0

1 (n+2) 2 i+n i xA (x) = iai xi ,
= 1 + (n + 1)x + 2 x + = x,
(1 x)n+1 i i=1
i=0


x Bi xi ai1
= 1 1
x + 1 2
x 1
x 4
+ = , A(x) dx = xi ,
e 1
x 2 12 720 i! i=1
i
i=0
( )
1 1 2i i A(x) + A(x)


(1 1 4x) = 1 + x + 2x + 5x +
2 3
= x, = a2i x2i ,
2x i+1 i 2
( )
i=0 i=0
1 2i i

= 1 + 2x + 6x2 + 20x3 + = x, A(x) A(x)
1 4x i = a2i+1 x2i+1 .
( )n 2
( )
i=0
1 1 1 4x (4+n) 2 2i + n i i
i=0
= 1 + (2 + n)x + 2 x + = x, Summation: If bi = j=0 ai then
1 4x 2x i=0
i
1
1 1 B(x) = A(x).
ln = x + 2 x + 6 x + 12 x + =
3 2 11 3 25 4
Hi xi , 1x
1x 1x Convolution:
( )2 i=1

Hi1 xi
1
ln
1
= 12 x2 + 34 x3 + 11 x 4
+ = , i

2 1x 24 i A(x)B(x) = aj bij xi .
i=2
i=0 j=0
x
= x + x 2
+ 2x 3
+ 3x 4
+ = Fi x i ,
1 x x2 i=0 God made the natural numbers;

Fn x all the rest is the work of man.
= F n x + F 2n x 2
+ F 3n x 3
+ = Fni xi .
1 (Fn1 + Fn+1 )x (1)n x2 i=0
Leopold Kronecker
 Theoretical Computer Science Cheat Sheet
Series Eschers Knot
Expansions:
( ) ( )n { }
1 1 n+i i 1 i
ln = (Hn+i Hn ) x, = xi ,
(1 x) n+1 1x i x n
[ ] { }
i=0 i=0
n i i n!xi
xn = x, (e 1)
x n
= ,
i n i!
( )n [ ]
i=0 i=0

1 i n!xi (4)i B2i x2i
ln = , x cot x = ,
1x i=0
n i! i=0
(2i)!

i1 2 (2 1)B2i x
2i 2i 2i1
1
tan x = (1) , (x) = x
,
i=1
(2i)! i=1
i

1 (i) (x 1) (i)
= x
, = ,
(x) i=1
i (x) i=1
ix
1
(x) = , Stieltjes Integration
p
1 px

If G is continuous in the interval [a, b] and F is nondecreasing then
d(i) b
2 (x) = where d(n) = d|n 1,
i=1
xi G(x) dF (x)
a

S(i) exists. If a b c then
(x)(x 1) = where S(n) = d|n d, c b c
i=1
xi
G(x) dF (x) = G(x) dF (x) + G(x) dF (x).
2 |B2n | 2n
2n1 a a b
(2n) = , n N, If the integrals involved exist
(2n)! b
( ) b b
x (4i 2)B2i x2i G(x) + H(x) dF (x) = G(x) dF (x) + H(x) dF (x),
= (1)i1 , a a a
sin x (2i)!
( )n
i=0 b ( ) b b

G(x) d F (x) + H(x) = G(x) dF (x) + G(x) dH(x),
1 1 4x n(2i + n 1)!
= xi , a a a
2x i!(n + i)! b
i=0 b b ( )
i/2 i c G(x) dF (x) = G(x) d c F (x) = c G(x) dF (x),
x 2 sin 4 i a a a
e sin x = x, b b
i!

i=1
G(x) dF (x) = G(b)F (b) G(a)F (a) F (x) dG(x).

1 1x (4i)!
a a
= xi , If the integrals involved exist, and F possesses a derivative F at every
x i=0
i
16 2(2i)!(2i + 1)! point in [a, b] then
( )2
b b
arcsin x 4i i!2
= x2i . G(x) dF (x) = G(x)F (x) dx.
x i=0
(i + 1)(2i + 1)! a a

Cramers Rule Fibonacci Numbers

00 47 18 76 29 93 85 34 61 52
If we have equations: 86 11 57 28 70 39 94 45 02 63
a1,1 x1 + a1,2 x2 + + a1,n xn = b1 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, . . .
95 80 22 67 38 71 49 56 13 04
a2,1 x1 + a2,2 x2 + + a2,n xn = b2 59 96 81 33 07 48 72 60 24 15
Definitions:
.. .. .. 73 69 90 82 44 17 58 01 35 26 Fi = Fi1 +Fi2 , F0 = F1 = 1,
. . . 68 74 09 91 83 55 27 12 46 30 Fi = (1)i1 Fi ,
an,1 x1 + an,2 x2 + + an,n xn = bn 37 08 75 19 92 84 66 23 50 41
( )
14 25 36 40 51 62 03 77 88 99
Fi = 15 i i ,
Let A = (ai,j ) and B be the column matrix (bi ). Then 21 32 43 54 65 06 10 89 97 78 Cassinis identity: for i > 0:
there is a unique solution i det A = 0. Let Ai be A
with column i replaced by B. Then
42 53 64 05 16 20 31 98 79 87
Fi+1 Fi1 Fi2 = (1)i .
det Ai Additive rule:
xi = . The Fibonacci number system:
det A Every integer n has a unique Fn+k = Fk Fn+1 + Fk1 Fn ,
representation F2n = Fn Fn+1 + Fn1 Fn .
Improvement makes strait roads, but the crooked n = Fk1 + Fk2 + + Fkm , Calculation by matrices:
roads without Improvement, are roads of Genius. ( ) ( )n
where ki ki+1 + 2 for all i, Fn2 Fn1 0 1
William Blake (The Marriage of Heaven and Hell) 1 i < m and km 2. = .
Fn1 Fn 1 1