MA1521 CALCULUS FOR COMPUTING

Wang Fei

matwf@nus.edu.sg

Department of Mathematics
Office: S17-06-16
Tel: 6516-2937

Polynomial Equations 2
Roots of Quadratic Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Roots of Cubic Equations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Roots of Quartic Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Roots of Quintic Equations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Complex Numbers 10
Complex Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Complex Functions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Stereographic Projection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Matrix Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Evaluation of π 19
Liu Hui’s Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

Series 25
Power Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Ordinary Differential Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Term by Term Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Some Special Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

1
Polynomial Equations 2 / 34

Roots of Quadratic Equations

 Solve the quadratic equation: x2 + bx + c = 0.
2
b2

b
1. x+ − + c = 0.
2 4
2
b2 − 4c

b
2. x+ = .
2 4
√
b b2 − 4c
3. x + = ± .
2 2
√
b ± b2 − 4c
4. x = − .
2
 Remove the linear term by substitution.
b b
 Let t = x + , i.e., x = t − .
2 2
2
 The equation becomes t + d = 0.

3 / 34

Roots of Cubic Equations

 Solve the cubic equation: x3 + bx2 + cx + d = 0.
1. Use Cardano’s method (1545): x = t − b/3.
 t3 + pt + q = 0.
2. Set t = u + v :

 (u3 + v 3 ) + (3uv + p)(u + v) + q = 0.

3. Suppose 3uv + p = 0.

 u3 + v 3 = −q , u3 v 3 = −p3 /27.
4. u3 and v 3 are roots of
 z 2 + qz − p3 /27 = 0.
5. Solve the equation above to get u and v .

 x = t − b/3 = u + v − b/3.
4 / 34

2
Roots of Cubic Equations
 Solve the cubic equation: x3 + bx2 + cx + d = 0.
1. Set x = t − b/3.

 t3 + pt + q = 0.
2. Solve z 2 + qz − p3 /27 = 0.

z1 = u3 and z2 = v 3 .
√ √
3. x = u + v − b/3 = 3 z1 + 3 z2 − b/3.

 t3 + pt + q = 0 is called the depressed form.

 The discriminant ∆ = 4p3 + 27q 2 .
 ∆ > 0: 3 distinct real roots.
 ∆ = 0: multiple real roots.
 ∆ < 0: 1 real and 2 nonreal conjugate roots.

5 / 34

Example
 Solve x3 − 6x2 + 9x − 4 = 0.
1. b = −6. Set x = t − b/3 = t + 2.
 Depressed form: t3 − 3t − 2 = 0.

2. p = −3 and q = −2. Solve z 2 + qz − p3 /27 = 0:

z 2 − 2z + 1 = 0 ⇒ z1 = z2 = 1.
√ √
3. u = 3 z1 = 1 and v = 3 z2 = 1.

 x1 = u + v + 2 = 1 + 1 + 2 = 4.
4. Factorize x3 − 6x2 + 9x − 4 = (x − 4)(x2 − 2x + 1).
5. Solve x2 − 2x + 1 = 0: x2 = x3 = 1.
6. Therefore, the roots are x1 = 4, x2 = 1, x3 = 1.

 x3 − 6x2 + 9x − 4 = (x − 4)(x − 1)2 .

6 / 34

3
Roots of Quartic Equations
 Quartic equation: x4 + bx3 + cx2 + dx + e = 0.
1. Ferrari method (1522–1565): Set x = t − b/4.

 Depressed form: t4 + pt2 + qt + r = 0.

2. Solve a cubic equation in z :

 3
pr q 2

5 2
3 2 p
 z + pz + (2p − r)z + − − = 0.
2 2 2 8
s 
√

2q
± p + 2z ± − 3p + 2z± √
p + 2z b
3. x = − .
2 4
 The first and the third ± are both positive or negative.

7 / 34

Example
 Solve x4 − 12x3 + 43x2 − 24x − 80 = 0.
1. Set x = t + 3.

 t4 − 11t2 + 18t − 8 = 0.
2. Solve a cubic equation in z :
55 2
 z3 − z + 250z − 750 = 0.
2
 Use Cardano’s method to get z = 15/2.
p
±2 ± −(−18±18)
3. z = .
2
 z = 1, 1, −4, 2.

8 / 34

4
Roots of Quintic Equations
 Quintic equation: x5 + bx4 + cx3 + dx2 + ex + f = 0.
 Évariste Galois (1811 – 1832) French
 Niels Henrik Abel (1802 – 1829) Norwegian
 Abel’s Impossibility Theorem.
 There is no general algebraic solution
√
 that is, expression using +, −, ×, ÷, n ,
to polynomial equations of degree ≥ 5.
 Remarks.
 Approximated solution may be found using Newton-Raphson’s method.
 Johann Carl Friedrich Gauss (1777 – 1855)
 Fundamental Theorem of Algebra (1799): A polynomial equation of degree n has n roots in
complex numbers.

9 / 34

Complex Numbers 10 / 34

Complex Numbers
 Complex numbers: C = {x + iy | x, y ∈ R}, i2 = −1.
 C is identified with the Cartesian plane R2 :
 x + iy ↔ (x, y).
 Polar coordinate: z = r(cos θ + i sin θ).
 z = reiθ , r = |z|, θ = arg z .

z = x + iy
y 2
y
p x2 +
r=

θ
O x

11 / 34

5
Complex Functions
 For any real number x,
 eix = cos x + i sin x,
 e−ix = cos(−x) + i sin(−x) = cos x − i sin x.
We solve that
eix + e−ix eix − e−ix
 cos x = and sin x = , x ∈ R.
2 2i
 Definition. For any complex number z ,define
eiz + e−iz eiz − e−iz
 cos z = and sin z = .
2 2i
We can verify that all the trigonometric identities still hold:
 cos2 z + sin2 z = 1;
 sin 2z = 2 sin z cos z ;
 cos 2z = cos2 z − sin2 z ;
 ................................................................................................
12 / 34

Complex Functions
eiz + e−iz eiz − e−iz
 Recall that cos z = and sin z = .
2 2i
 iz 2  iz 2
2 2 e + e−iz e − e−iz
 cos z + sin z = + .
2 2i
 iz 2
e + e−iz ei2z + e−i2z + 2
 = .
2 4
 iz 2
ei2z + e−i2z − 2

e − e−iz
 = .
2i −4
 cos2 z + sin2 z = 1.

sin z i(eiz − e−iz )

 Definition. tan z = = iz .
cos z e + e−iz
cos z 1 1
 cot z = , sec z = , csc z = .
sin z cos z sin z
13 / 34

6
Complex Functions
eiz + e−iz eiz − e−iz
 Recall that cos z = and sin z = .
2 2i
ei·i + e−i·i e−1 + e e + e−1
 cos i = = = ≈ 1.543.
2 2 2
ei·i − e−i·i e−1 − e i(e − e−1 )
 sin i = = = ≈ 1.175i.
2i 2i 2
 The exponential function ax is extendable for any a, x ∈ C.
 Examples.
 (−1)i = e−π ; ii = e−π/2 .
 The logarithmic function ln z is also extendable to C \ {0}.
 Examples.
πi πi
 log(−1) = πi; log i = ; log(−i) = − .
2 2
14 / 34

Example
 Let z = cos x + i sin x = eix . Let
1 − z n+1
 S = 1 + z + · · · + zn = .
1−z
1 − [cos(n + 1)x + i sin(n + 1)x]
S=
1 − (cos x + i sin x)
(n+1)x
1 1 sin 2 sin (n+1)x
2
sin nx
2
= + + i
2 2 sin x2 sin x2
Re(S) = 1 + cos x + cos 2x + · · · + cos nx
(n+1)x
1 1 sin 2
= +
2 2 sin x2
Im(S) = sin x + sin 2x + · · · + sin nx
sin (n+1)x
2
sin nx
2
= .
sin x2
15 / 34

7
Stereographic Projection
 For any z = x + iy ↔ (x, y), connect z to N(0, 0, 1).
 Line Nz intersects the unit sphere x2 + y 2 +Nz 2 = 1 at Z .
b

Z
b z
b
O
b

z
Z b

 S 2 := {(x, y, z) | x2 + y 2 + z 2 = 1} the unit sphere.

 C ↔ S 2 \ {N} via z ↔ Z .
 C ∪ {∞} ↔ S 2 via z ↔ Z and ∞ ↔ N(0, 0, 1).
16 / 34

Matrix Representation
 A complex number can be identified with a 2 × 2 real matrix:
 
x −y
 z = x + iy ↔ = Mz .
y x
 All the arithmetic properties are preserved:
 Addition: Let z1 = x1 + iy1 and z2 = x2 + iy2 .
 z1 + z2 = (x1 +
 x2 ) + i(y1 + y2 ). 
x1 + x2 −(y1 + y2 )
 Mz1 + Mz2 = = Mz1 +z2
y1 + y2 x1 + x2
 Multiplication: Let z1 = x1 + iy1 and z2 = x2 + iy2 .
 z1 z2 = (x1 x2 − y1 y2 ) + i(x1 y2 + x2 y1 ).
  
x1 −y1 x2 −y2
Mz1 Mz2 =
y1 x1 y2 x2
 
x1 x2 − y1 y2 −(x1 y2 + y1 x2 )
= = Mz1 z2
x1 y2 + y1 x2 x1 x2 − y1 y2
17 / 34

8
 Matrix Representation
 A complex number can be identified with a 2 × 2 real matrix:
 
x −y
 z = x + iy ↔ = Mz .
y x
 All the arithmetic properties are preserved:
 Modulus: |z|2 = x2 + y 2 = 
det(Mz )
.
x y
 Conjugate: z ∗ = x − iy ↔ = (Mz )T .
−y x
x − iy
 Quotient: Let z = x + iy . Then 1/z = z −1 = .
x2 + y 2
 
1 x y
 Mz −1 = 2 = · · · = (Mz )−1 .
x + y2 −y x
 Power: Mz n = (Mz )n .
 ................................................................................................
18 / 34

Evaluation of π 19 / 34

Liu Hui’s Algorithm

 Recall that Archimedes, Liu Hui and Zu Chongzhi used regular polygons to approximation π , the
area of unit circle.
 Liu Hui’s (220 – ?) algorithm for π :
1. Let MN = length of the side of regular N -sided polygon inscribed in the unit circle. M6 = 1.
2. Let AN = area of the regular N -sided polygon inscribed in the unit circle. A6 ≈ 3M6 = 3.
2
3. Let L(n) = 2 − M3×2n.

p
 L(n + 1) = 2 + L(n), L(1) = 2 − 12 = 1.
4. π ≈ A3×2n ≈ 3 × 2n−1 × M3×2n .
 Remark.
 Liu Hui evaluated up to 96-sided polygon, and used a shortcut to generate the result for
1536-sided polygon.

20 / 34

9
Liu Hui’s Algorithm
 Liu Hui’s Algorithm for π :

Iteration Sides Approximation of π

1 6 3.
2 12 3.1
3 24 3.13
4 48 3.14
5 96 3.141
9 1536 3.14159
12 12288 3.1415926
15 98304 3.141592653
20 1572864 3.141592653589
30 1610612736 3.141592653589793238

 Remarks: π ≈ 22/7 and π ≈ 355/113.

21 / 34

Integration
Z b Z b
 If f (x) ≤ g(x) on [a, b], then f (x) dx ≤ g(x) dx.
a a
1 1
 For all 0 ≤ x ≤ 1,≤ ≤ 1.
2 1 + x2
1 4 4 x4 (1 − x)4
 x (1 − x) ≤ 2
≤ x4 (1 − x)4 .
2 1+x
1 1 1
x4 (1 − x)4 x4 (1 − x)4
Z Z Z
 dx ≤ dx ≤ x4 (1 − x)4 dx.
0 2 0 1 + x2 0
1 22 1
 ≤ −π ≤ .
1260 7 630
22 1
∴ is a bigger approximation of π with error ≤ < 0.0016.
7 630
22 / 34

10
Integration
1 Z 1 8 Z 1 8
x8 (1 − x)8 x (1 − x)8 x (1 − x)8
Z
 dx ≤ 2
dx ≤ dx.
0 8 0 4(1 + x ) 0 4
1 47171 1
 ≤π− ≤ .
1750320 15015 875160
47171
∴ ≈ 3.14159174 · · · is a smaller approximation for π
15015
1
with error ≤ ≈ 10−6 .
875160
1 1 1
x12 (1 − x)12 x12 (1 − x)12 x12 (1 − x)12
Z Z Z
 dx ≤ dx ≤ dx.
0 32 0 16(1 + x2 ) 0 16
1 431302721 1
 ≤ −π ≤ .
2163324800 137287920 1081662400
431302721
∴ ≈ 3.141592654 · · · is a bigger approximation for π
137287920
1
with error ≤ ≈ 10−9 .
1081662400
23 / 34

Integration
 Approximate π using
1 1
x4n (1 − x)4n x4n (1 − x)4n 1
Z Z
 dx ≤ dx < 10n−2
0 4n−1 (1 + x2 ) 0 4n−1 2

n Fraction Decimal
22
1 3.14
7
431302721
3 3.141592654
137287920
26856502742629699
5 3.141592653589793
8548690331301120
10 89293478252053341114758995682016773
28422996899365886608045972478361600
3.141592653589793238462643383279

24 / 34

11
Series 25 / 34

Power Series
 Recall that a power series has the form:
∞
X
 cn xn = c0 + c1 x + c2 x2 + · · · + cn xn + · · · .
n=0
It plays an important role in approximation theory.

 Examples.
x2 x3 x4
 ex = 1 + x + + + + ···.
2! 3! 4!
1 1 1
 e ≈ 1+ 1+ + +···+ ≈ 2.718281801.
2! 3! 10!
x3 x5 x7
 sin x = x − + − + ···.
3! 5! 7!
23 25 27 29
 sin 2 ≈ 2 − + − + ≈ 0.9093474427.
3! 5! 7! 9!
26 / 34

Integration
Z 1
 Approximate sin(x2 ) dx.
0
x3 x5 x7 x9 x11
1. sin x = x − + − + − + ···.
3! 5! 7! 9! 11!
x6 x10 x14 x18 x22
2. sin(x2 ) ≈ x2 − + − + − .
3! 5! 7! 9! 11!
Z 1
3. Approximate sin(x2 ) dx by
0
1 1 1
x6 x22
Z Z Z
2
 x dx − dx + · · · − dx
0 0 3! 0 11!
Z 1
4. sin(x2 ) dx ≈ 0.3102683017174579.
0

 The exact value is 0.31026830172338110 · · · .

27 / 34

12
Ordinary Differential Equation
 Suppose f (x) = c0 + c1 x + c2 x2 + c3 x3 + · · · + cn xn + · · · .
 Term by term differentiation: f ′ (x):
c1 + 2c2 x + 3c3 x2 + · · · + ncn xn−1 + · · · .


dy
 Example. Suppose = y and y = 1 at x = 0.
dx
 Let f (x) = y = c0 + c1 x + c2 x2 + · · · + cn xn + · · · .
 f ′ (x) = c1 + 2c2 x + 3c3 x2 + · · · + ncn xn−1 + · · · .
 Compare coefficients:
 c0 = 1, c1 = c0 , 2c2 = c1 , 3c3 = c2 , 4c4 = c3 , . . . .
 c0 = 1, c1 = 1, c2 = 1/2, c3 = 1/6, c4 = 1/24, . . .
x2 x3 x4
 y = f (x) = 1 + x + + + + ···.
2 6 24
28 / 34

Term by Term Integration

 Suppose f (x) = c0 + c1 x + c2 x2 + c3 x3 + · · · + cn xn + · · · .
Z
 Term by term integration: f (x) dx:
c1 2 c2 3 cn n+1
 c0 x + x + x +···+ x +···
2 3 n+1
 Examples.
1
 = 1 + x + x2 + x3 + x4 + · · · + xn + · · · .
1−x
1
Z
 dx = − ln |1 − x| + C .
1−x
x2 x3 x4 xn xn+1
 x+ + + +···+ + + ···.
2 3 4 n n+1
x2 x3 x4 xn xn+1
 − ln |1 − x| = x + + + +···+ + + ···.
2 3 4 n n+1
29 / 34

13
Term by Term Integration
 Suppose f (x) = c0 + c1 x + c2 x2 + c3 x3 + · · · + cn xn + · · · .
Z
 Term by term integration: f (x) dx:
c1 2 c2 3 cn n+1
 c0 x + x + x +···+ x +···
2 3 n+1
 Examples.
1
 = 1 − x2 + x4 − x6 + · · · + (−1)n x2n + · · · .
1 + x2
1
Z

2
dx = tan−1 x + C .
1+x
x3 x5 x7 x2n+1
 x− + − + · · · + (−1)n + ···.
3 5 7 2n + 1
x3 x5 x7 x2n+1
 tan−1 x = x − + − + · · · + (−1)n +···
3 5 7 2n + 1
30 / 34

Some Special Series

x2 x3 xn
 − ln |1 − x| = x + + +···+ + · · · , R = 1.
2 3 n
 Let x = 1/2. Then
1 1 1 1
 ln 2 = + 2
+ 3
+···+ +···
2 2·2 3·2 n · 2n
1 1 1
 ln 2 ≈ + 2
+···+ = 0.6931471 · · ·
2 2·2 20 · 220
 Let x = −1. Then
1 1 1 (−1)n
 − ln 2 = −1 + − + − · · · + + ···.
2 3 4 n
n−1
1 1 1 (−1)
 ln 2 = 1 − + − + · · · + + ···.
2 3 4 n
∴ The alternating Harmonic series converges to ln 2.
Warning: The convergency of alternating Harmonic series is very slow!

31 / 34

14
Some Special Series
x3 x5 x7 x2n+1
 tan−1 x = x − + − + · · · + (−1)n +···
3 5 7 2n + 1
1
 Note that R = 1. Take x = √ :
3
π 1 1 1 1
 =√ − √ + √ − √ +···
6 3 3( 3)3 5( 3)5 7( 3)7
π 1 1 1 1
 ≈√ − √ + √ +···+ √
6 3 3( 3)3 5( 3)5 41( 3)41
π
 ≈ 0.52359877559927 · · ·
6
 π ≈ 3.14159265359563 · · ·
∞
√ X (−3)−n
 π=2 3 is the Madhava-Leibniz series.
n=0
2n + 1
 It is efficient in evaluating π .

32 / 34

Some Special Series

x3 x5 x7 x2n+1
 tan−1 x = x − + − + · · · + (−1)n +···
3 5 7 2n + 1
 Note that R = 1. Take x = 1:
π 1 1 1 1 (−1)n
 = 1− + − + −···+ + ···.
4 3 5 7 9 2n + 1
This is known as the Leibniz formula for π .
 Warning: The convergency is very slow.
 Srinivasa Ramanujan (1887–1920) Indian mathematician.
√ ∞
1 2 2 X (4n)!(1103 + 26390n)
 = .
π 9801 n=0 (n!)4 3944n
√
1 2 2
 The first term gives ≈ · 1103.
√ π 9801
9801 2
 π≈ = 3.14159273 · · · .
4412
33 / 34

15
Some Special Series
x3 x5 x7 x2n+1
 tan−1 x = x − + − + · · · + (−1)n +···
3 5 7 2n + 1
 Note that R = 1. Take x = 1:
π 1 1 1 1 (−1)n
 = 1− + − + −···+ + ···.
4 3 5 7 9 2n + 1
This is known as the Leibniz formula for π .
 Warning: The convergency is very slow.
 Srinivasa Ramanujan (1887–1920) Indian mathematician.
∞
1 X (−1)n (6n)!(13591409 + 545140134n)
 = 12 3 6403203n+3/2
.
π n=0
(3n)!(n!)
1 13591409
 First term gives ≈ 12 · .
π 6403203/2
6403203/2
 π≈ = 3.141592653589793 · · · .
12 · 13591409
34 / 34

16