Mathematics

10 Topics Covered
• Limits, continuity and L'Hopital's rule
• Differentiation — all rules and standard derivatives
• Integration — all techniques, definite and indefinite
• Maxima and minima — optimisation problems
• Partial derivatives, gradient, Jacobian matrix
• Matrices, determinants, systems of equations
• Eigenvalues, eigenvectors, matrix rank
• 1st and 2nd order ODEs
• Laplace transforms and inverse Laplace
• Numerical methods — Newton-Raphson, Simpson's rule
 1. Limits, Continuity & L'Hopital's Rule Foundation of all calculus

1.1 What is a limit?

The limit describes the value a function approaches as the input gets closer to some point — even if the function is
undefined at that exact point. Written as: lim(x to a) f(x) = L

Concept Definition Exists when

Left-hand limit x approaches a from the LEFT (x to a-) lim(x to a-) f(x) = L_left

Right-hand limit x approaches a from the RIGHT (x to a+) lim(x to a+) f(x) = L_right

Two-sided limit x approaches a from either side Left-hand limit = Right-hand limit = L

Continuity at x=a f behaves smoothly at x=a with no jumps or holes (1) f(a) exists (2) limit exists (3) lim = f(a)

Basic limit laws (assuming lim f and lim g exist as x approaches a):
lim(f + g) = lim(f) + lim(g)
lim(f - g) = lim(f) - lim(g)
lim(f * g) = lim(f) * lim(g)
lim(f / g) = lim(f) / lim(g) [provided lim(g) is not 0]
lim(c) = c (constant rule)
lim(x^n) = a^n (direct substitution when no 0/0 issue)

1.2 L'Hopital's rule — resolving indeterminate forms

When direct substitution gives 0/0 or infinity/infinity, the limit is indeterminate. L'Hopital's rule resolves it by
differentiating numerator and denominator separately (NOT the quotient rule).
L'Hopital's rule:
If lim f(x)/g(x) gives 0/0 or inf/inf as x approaches a, then:
lim f(x)/g(x) = lim f'(x)/g'(x)

Differentiate TOP and BOTTOM separately.

Apply repeatedly if the result is still indeterminate.

Common indeterminate forms:

0/0 inf/inf 0 * inf inf - inf 0^0 1^inf inf^0

Example: Find lim(x to 0) sin(x) / x [famous limit — result = 1]

Direct substitution: sin(0)/0 = 0/0 — indeterminate — apply L'Hopital
d/dx[sin(x)] = cos(x)
d/dx[x] = 1
lim(x to 0) cos(x)/1 = cos(0)/1 = 1/1 = 1
Answer: lim(x to 0) sin(x)/x = 1

Example: Find lim(x to inf) (3x^2 + 2x) / (5x^2 - 1)

Direct substitution: inf/inf — apply L'Hopital twice
1st application: (6x + 2) / (10x) — still inf/inf
2nd application: 6 / 10 = 3/5
Answer: 3/5 (same as dividing leading coefficients: 3x^2 / 5x^2 = 3/5)
 2. Differentiation — All Rules & Techniques Derivation

2.1 Definition and physical meaning

The derivative f'(x) is the instantaneous rate of change of f at x — equivalently, the slope of the tangent line to the
curve at that point. Formally defined as a limit of the difference quotient.
Definition (first principles):
f'(x) = lim(h to 0) [f(x+h) - f(x)] / h

Notation: f'(x) = dy/dx = d/dx[f(x)] = Df(x)

Physical meanings:
If f = position -> f' = velocity -> f'' = acceleration
If f = total cost -> f' = marginal cost
If f = voltage -> f' = rate of change of voltage

2.2 All differentiation rules

Rule Formula Worked example

Power rule d/dx[x^n] = n * x^(n-1) d/dx[x^5] = 5x^4

Constant d/dx[c] = 0 d/dx[7] = 0

Constant multiple d/dx[c*f] = c * f' d/dx[3x^2] = 6x

Sum / difference d/dx[f +/- g] = f' +/- g' d/dx[x^2 + sin x] = 2x + cos x

Product rule d/dx[f*g] = f'g + fg' d/dx[x^2 * sin x] = 2x*sin x + x^2*cos x

Quotient rule d/dx[f/g] = (f'g - fg') / g^2 d/dx[sin x / x] = (x*cos x - sin x) / x^2

Chain rule d/dx[f(g(x))] = f'(g(x)) * g'(x) d/dx[sin(x^2)] = cos(x^2) * 2x

2.3 Standard derivatives table — memorise all of these

f(x) f'(x) f(x) f'(x)

xn n × x(n-1) sin x cos x

ex ex cos x -sin x

ax ax × ln(a) tan x sec2(x)

ln x 1/x sin-1 x 1 / √(1-x2)

Log 1 / (x × ln tan-1 x 1 / (1+x2)

{a(x)} a)

Example: Chain rule: differentiate f(x) = e^(3x^2 + 2x)

Outer function: e^u -> derivative = e^u
Inner function: u = 3x^2 + 2x -> u' = 6x + 2
Chain rule: f'(x) = e^(3x^2+2x) * (6x + 2)

Example: Product rule: d/dx [x^3 * ln(x)]

f = x^3, f' = 3x^2
g = ln x, g' = 1/x
d/dx[f*g] = f'g + fg' = 3x^2 * ln(x) + x^3 * (1/x) = 3x^2*ln(x) + x^2
2.4 Higher-order derivatives and implicit differentiation

Higher-order derivatives:
f''(x) = d^2y/dx^2 (second derivative — measures concavity and acceleration)
f'''(x) = d^3y/dx^3 (third derivative)

f''(a) > 0 -> concave UP at a (like a smile / valley)

f''(a) < 0 -> concave DOWN at a (like a frown / hill)

Implicit differentiation (when y cannot be isolated):

Differentiate BOTH sides w.r.t. x.
Every time you differentiate y, multiply by dy/dx.

Example: x^2 + y^2 = 25 (equation of a circle)

Differentiating both sides: 2x + 2y * (dy/dx) = 0
Solving: dy/dx = -2x / (2y) = -x/y
 3. Integration — Definite, Indefinite & Techniques

3.1 Fundamental theorem and standard integrals

Indefinite integral (antiderivative):

integral f(x) dx = F(x) + C where F'(x) = f(x)
ALWAYS include + C (constant of integration) for indefinite integrals!

Definite integral (net area under curve from a to b):

integral[a to b] f(x) dx = F(b) - F(a) (no + C needed)

Fundamental Theorem of Calculus:

d/dx [integral(a to x) f(t) dt] = f(x)
(differentiation and integration are exact inverses)

integral f(x) dx Result + C integral f(x) dx Result + C

x^n dx (n not -1) x^(n+1) / (n+1) sin x dx -cos x

1/x dx ln|x| cos x dx sin x

e^x dx e^x sec^2(x) dx tan x

a^x dx a^x / ln(a) 1/sqrt(1-x^2) dx sin^(-1) x

1/(1+x^2) dx tan^(-1) x ln x dx x*ln(x) - x

3.2 Integration by substitution (u-substitution)

Use when the integrand contains a composite function and its derivative. Substitute u = inner function to convert to a
simpler integral.
Procedure:
1. Identify u = g(x) (choose the inner function)
2. Compute du = g'(x) dx -> dx = du / g'(x)
3. Substitute: replace g(x) with u and dx with du/g'(x)
4. Integrate the simpler expression in terms of u
5. Substitute back: replace u with g(x)
Hint for choosing u: look for a function whose derivative also appears in the integrand.

Example: Integral of 2x * e^(x^2) dx

Let u = x^2 -> du = 2x dx -> 2x dx = du
Integral 2x * e^(x^2) dx = integral e^u du = e^u + C = e^(x^2) + C

Example: Integral of cos(3x+1) dx

Let u = 3x+1 -> du = 3 dx -> dx = du/3
Integral cos(u) * du/3 = (1/3) sin(u) + C = (1/3) sin(3x+1) + C

3.3 Integration by parts — LIATE rule

Used when the integrand is a product of two different types of functions. Converts one integral into a simpler one.
Choose u using the LIATE priority order.
Formula: integral u dv = u*v - integral v du

LIATE rule — choose u as whichever comes FIRST in this list:

L = Logarithmic functions: ln x, log x <- highest priority for u
I = Inverse trigonometric: sin^-1 x, tan^-1 x
A = Algebraic (polynomial): x^2, 3x+1, x
T = Trigonometric: sin x, cos x, tan x
E = Exponential: e^x, a^x <- lowest priority for u
(The other function becomes dv — you need to integrate it to get v)
Example: Integral of x * e^x dx [x is Algebraic (A), e^x is Exponential (E) -> u=x]
u = x -> du = dx
dv = e^x dx -> v = e^x
Integral x*e^x dx = x*e^x - integral e^x dx = x*e^x - e^x + C = e^x(x-1) + C

Example: Integral of ln(x) dx [ln x is Logarithmic (L) -> u=ln(x), dv=dx]

u = ln x -> du = (1/x) dx
dv = dx -> v = x
Integral ln x dx = x*ln(x) - integral x*(1/x) dx = x*ln(x) - integral 1 dx
= x*ln(x) - x + C

Exam tip: Tabular integration (rapid repeated by-parts): when u is a polynomial (e.g. x^3*e^x), set up a table — repeatedly
differentiate u column, repeatedly integrate dv column, multiply diagonals with alternating +/- signs. Much faster than applying
the formula repeatedly.
 4. Maxima, Minima & Optimisation 'Finding extreme point'

4.1 Critical points — where extrema occur

A critical point is where f'(x) = 0 or f'(x) is undefined. At a critical point, the function may have a local maximum, local
minimum, or neither (saddle/inflection point).
Finding extrema — systematic procedure:

STEP 1: Find f'(x) and solve f'(x) = 0 -> critical points c_1, c_2, ...

STEP 2: Classify each critical point using the Second Derivative Test:
f''(c) > 0 -> LOCAL MINIMUM (concave UP at c — valley shape)
f''(c) < 0 -> LOCAL MAXIMUM (concave DOWN at c — hill shape)
f''(c) = 0 -> INCONCLUSIVE (use First Derivative Test or higher derivatives)

STEP 3: Evaluate f at all critical points AND at endpoints to find GLOBAL extrema.

Useful signs:
f'(x) > 0 on interval -> f is INCREASING (going uphill)
f'(x) < 0 on interval -> f is DECREASING (going downhill)

Example: Find all extrema of f(x) = 2x^3 - 3x^2 - 12x + 4

Step 1: f'(x) = 6x^2 - 6x - 12 = 6(x^2 - x - 2) = 6(x-2)(x+1)
f'(x) = 0 -> x = 2 or x = -1

Step 2: f''(x) = 12x - 6

At x = 2: f''(2) = 24 - 6 = 18 > 0 -> LOCAL MINIMUM
At x = -1: f''(-1) = -12 - 6 = -18 < 0 -> LOCAL MAXIMUM

Step 3: f(2) = 2(8) - 3(4) - 12(2) + 4 = 16 - 12 - 24 + 4 = -16 <- local min

f(-1) = 2(-1) - 3(1) - 12(-1) + 4 = -2 - 3 + 12 + 4 = 11 <- local max

4.2 First derivative test — alternative classification

Check the SIGN of f'(x) on each side of the critical point c:

f' changes: + -> 0 -> - (positive to negative) -> LOCAL MAXIMUM

f' changes: - -> 0 -> + (negative to positive) -> LOCAL MINIMUM
f' same sign on both sides -> INFLECTION POINT (no extremum)

Inflection points: where f''(x) = 0 AND f'' changes sign (concavity changes).
At inflection: curve changes from concave up to concave down (or vice versa).

Exam tip: Optimisation word problem strategy: (1) define one variable, (2) write an expression for the quantity to optimise (area,
cost, time), (3) use any constraint to reduce to one variable, (4) differentiate and set = 0, (5) use second derivative test to
confirm max or min.
 5. Partial Derivatives, Gradient & Jacobian Matrix Jacobian matrix

5.1 Partial derivatives

For a function of several variables f(x, y), a partial derivative measures how f changes when only one variable
changes while all others are held constant.
Notation: df/dx = f_x (partial derivative w.r.t. x)
df/dy = f_y (partial derivative w.r.t. y)

To compute df/dx: treat ALL variables except x as CONSTANTS, differentiate w.r.t. x.

To compute df/dy: treat ALL variables except y as CONSTANTS, differentiate w.r.t. y.

Mixed partial derivatives (order usually doesn't matter for smooth functions):
d^2f/dxdy = d/dx[df/dy] (differentiate first w.r.t. y, then w.r.t. x)
Clairaut's theorem: d^2f/dxdy = d^2f/dydx (for continuous second partials)

Example: Find df/dx and df/dy for f(x,y) = x^3*y^2 + 2xy + y^3
df/dx (treat y as constant):
= 3x^2 * y^2 + 2y + 0 (y^3 has no x -> derivative = 0)

df/dy (treat x as constant):

= 2x^3 * y + 2x + 3y^2 (x^3 is constant; d/dy[y^2]=2y; d/dy[y^3]=3y^2)

5.2 Gradient vector — direction of steepest increase

Gradient: grad(f) = (df/dx) i-hat + (df/dy) j-hat + (df/dz) k-hat

Physical meaning:
grad(f) points in direction of STEEPEST ASCENT (fastest increase)
|grad(f)| = magnitude = rate of maximum increase
grad(f) = 0 at critical points (candidates for max, min, saddle points)
grad(f) is perpendicular to level curves / surfaces

Directional derivative (rate of change in direction of unit vector u-hat):

D_u f = grad(f) . u-hat = |grad(f)| * cos(theta)
Maximum when u-hat is parallel to grad(f) (steepest direction)
Zero when u-hat is perpendicular to grad(f) (along level curve)

5.3 Jacobian matrix — the derivative of a vector function

The Jacobian generalises the derivative to vector-valued functions of multiple variables. It contains all first-order
partial derivatives in a matrix. Confirmed directly in the previous exam.
For a vector function F: R^n -> R^m
F(x_1, x_2, ..., x_n) = [f_1, f_2, ..., f_m]^T

Jacobian matrix J (m x n matrix of all partial derivatives):

J = [ df_i / dx_j ] (row i = function f_i, column j = variable x_j)

[ df_1/dx_1 df_1/dx_2 ... df_1/dx_n ]

J = [ df_2/dx_1 df_2/dx_2 ... df_2/dx_n ]
[ ... ... ... ... ]
[ df_m/dx_1 df_m/dx_2 ... df_m/dx_n ]

Jacobian determinant (when m = n, square matrix):

det(J) = volume scaling factor of the transformation
det(J) = 0 -> transformation is singular (collapses a dimension)

Used in: coordinate changes, robotics velocity analysis, Newton's method for systems
Example: Find the Jacobian of F(x,y) = [x^2*y, x + y^2] (2 functions, 2 variables)
f_1 = x^2*y, f_2 = x + y^2

df_1/dx = 2xy, df_1/dy = x^2

df_2/dx = 1, df_2/dy = 2y

[ 2xy x^2 ]
J = [ 1 2y ]

det(J) = 2xy * 2y - x^2 * 1 = 4xy^2 - x^2 = x(4y^2 - x)

Link to robotics: The velocity Jacobian in robot kinematics is the SAME concept.
Ẋ = J(theta) * theta_dot relates joint velocities to end-effector velocity.
det(J)=0 means a singularity where the robot loses a degree of motion.
6. Matrices, Determinants & Systems of Equations Matrix operations

6.1 Matrix operations

Addition/subtraction (A +/- B): same dimensions required

(A +/- B)_ij = A_ij +/- B_ij

Scalar multiplication: (cA)_ij = c * A_ij

Matrix multiplication (A is m x n, B is n x p -> AB is m x p):

(AB)_ij = sum_k A_ik * B_kj (row i of A . column j of B)
NOT commutative: AB is NOT equal to BA in general!

Transpose: (A^T)_ij = A_ji (swap rows and columns)

Properties: (AB)^T = B^T * A^T (A^T)^T = A

6.2 Determinants

2x2 determinant:
| a b |
| c d | = ad - bc

3x3 determinant (cofactor expansion along row 1):

| a11 a12 a13 |
| a21 a22 a23 | = a11(a22*a33 - a23*a32) - a12(a21*a33 - a23*a31) + a13(a21*a32 - a22*a31)
| a31 a32 a33 |

Key properties:
det(AB) = det(A) * det(B)
det(A^T) = det(A)
det(cA) = c^n * det(A) for n x n matrix
Any row/column all zeros -> det = 0
Two identical rows -> det = 0
det = 0 -> matrix is SINGULAR (no inverse, no unique solution)

Example: 3x3 determinant for A = [[1,2,3],[0,1,4],[5,6,0]]

det(A) = 1*(1*0 - 4*6) - 2*(0*0 - 4*5) + 3*(0*6 - 1*5)
= 1*(0-24) - 2*(0-20) + 3*(0-5) = (-24) - (-40) + (-15) = -24 + 40 - 15 = 1

6.3 Matrix inverse and solving linear systems

2x2 inverse (if det is not zero):

A = [a b; c d] -> A^(-1) = (1/det(A)) * [d, -b; -c, a]
(swap main diagonal, negate off-diagonal, divide by determinant)

Solving Ax = b by Gauss-Jordan elimination:

1. Form augmented matrix [A | b]
2. Apply row operations to reduce A to identity matrix I
3. Right side becomes the solution: x = A^(-1)b

Cramer's rule (for 2x2 and 3x3 by hand):

x_j = det(A_j) / det(A)
where A_j = matrix A with column j REPLACED by the vector b

Example: Solve: 2x + y = 5 and x + 3y = 10 using Cramer's rule

A = [2 1; 1 3], b = [5; 10]
det(A) = 2*3 - 1*1 = 6 - 1 = 5

A_x = [5 1; 10 3]: det(A_x) = 5*3 - 1*10 = 15 - 10 = 5 x = det(A_x)/det(A) = 5/5 = 1

A_y = [2 5; 1 10]: det(A_y) = 2*10 - 5*1 = 20 - 5 = 15 y = det(A_y)/det(A) = 15/5 = 3

Verification: 2(1)+3=5 checkmark 1+3(3)=10 checkmark
 7. Eigenvalues, Eigenvectors & Matrix Rank Characteristic equation and Cayley-Hamilton
theorem

7.1 Eigenvalues and eigenvectors

For a square matrix A, an eigenvector is a non-zero vector v that only gets scaled (not rotated) when multiplied by A.
The scaling factor is the corresponding eigenvalue lambda.
Definition: A * v = lambda * v
(multiplying eigenvector v by A = same as scaling v by lambda)

Equivalently: (A - lambda*I) * v = 0
For non-trivial solution: det(A - lambda*I) = 0
This is the CHARACTERISTIC EQUATION.

Procedure:
STEP 1: Solve det(A - lambda*I) = 0
This gives the characteristic polynomial in lambda.
Its roots lambda_1, lambda_2, ... are the eigenvalues.

STEP 2: For each eigenvalue lambda_i:

Substitute into (A - lambda_i * I) * v = 0
Solve for v (eigenvector).
Expect one free variable — eigenvectors are determined up to a scalar multiple.

Example: Find eigenvalues and eigenvectors of A = [[4,1],[2,3]]

Step 1: det(A - lambda*I) = det([[4-lambda, 1],[2, 3-lambda]]) = 0
(4-lambda)(3-lambda) - 2*1 = 0
12 - 4*lambda - 3*lambda + lambda^2 - 2 = 0
lambda^2 - 7*lambda + 10 = 0
(lambda - 5)(lambda - 2) = 0 -> lambda_1 = 5, lambda_2 = 2

Step 2 for lambda_1 = 5:

(A - 5I)v = [[-1,1],[2,-2]]v = 0
-v_1 + v_2 = 0 -> v_2 = v_1
Eigenvector: v_1 = [1, 1]^T (or any scalar multiple: [2,2], [-3,-3], etc.)

Step 2 for lambda_2 = 2:

(A - 2I)v = [[2,1],[2,1]]v = 0
2v_1 + v_2 = 0 -> v_2 = -2*v_1
Eigenvector: v_2 = [1, -2]^T

7.2 Cayley-Hamilton theorem and matrix rank

Cayley-Hamilton theorem:
Every square matrix satisfies its own characteristic equation.
Example: if characteristic equation is lambda^2 - 7*lambda + 10 = 0
Then: A^2 - 7A + 10I = 0 (substitute A for lambda)
Use: compute A^(-1) or high powers of A efficiently without brute-force multiplication.

Matrix rank:
Rank = number of linearly independent rows = non-zero rows in row echelon form
For m x n matrix: rank is at most min(m, n)
Full rank (square): rank = n -> det is not 0 -> matrix is invertible

Nullity and null space:

Null space = all vectors x such that Ax = 0 (the solution set of the homogeneous system)
Nullity = n - rank(A) (dimension of the null space)
Rank + Nullity = n (number of columns) -- Rank-Nullity Theorem
 8. Ordinary Differential Equations (ODEs) 1st and 2nd order

8.1 Introduction and classification

A differential equation contains a function and its derivatives. The solution is a function — not just a number. ODEs
are used throughout engineering: control systems, robot dynamics, circuit analysis, vibration.
Term Meaning Example

Order Highest derivative present y' = first order; y'' = second order

y and all its derivatives appear to the first power only —

Linear ODE dy/dx + P(x)y = Q(x)
no y*y', no (y')^2

Homogeneous Right-hand side = 0 ay'' + by' + cy = 0

Non-homogeneous Right-hand side has a function of x ay'' + by' + cy = f(x)

Contains arbitrary constants C_1, C_2 (= family of

General solution y = C_1*e^(2x) + C_2*e^(3x)
solutions)

Particular solution Constants determined by applying initial conditions y(0)=1, y'(0)=0 -> specific C values

8.2 First order ODEs — three methods

METHOD 1 — Separable: f(y) dy = g(x) dx

Separate x and y to opposite sides, then integrate both sides.
Example: dy/dx = x*y
-> dy/y = x dx
-> ln|y| = x^2/2 + C
-> y = A*e^(x^2/2) where A = e^C (arbitrary constant)

METHOD 2 — Linear first order: dy/dx + P(x)*y = Q(x)

Integrating factor: mu(x) = e^(integral P(x) dx)
Multiply both sides by mu: d/dx [mu * y] = mu * Q(x)
Integrate both sides: mu * y = integral mu*Q(x) dx + C
Solve for y.

METHOD 3 — Exact: M dx + N dy = 0
Exact if dM/dy = dN/dx
Solution F(x,y) = C where: dF/dx = M, dF/dy = N

Example: Linear 1st order: dy/dx + (2/x)y = x^2 — find general solution
P(x) = 2/x, Q(x) = x^2
Integrating factor: mu = e^(integral 2/x dx) = e^(2*ln x) = x^2

Multiply both sides by x^2:

d/dx[x^2 * y] = x^2 * x^2 = x^4

Integrate: x^2 * y = x^5/5 + C

Solve for y: y = x^3/5 + C/x^2

8.3 Second order linear ODE with constant coefficients

General form: a*y'' + b*y' + c*y = f(x)

STEP 1: Solve homogeneous equation (right side = 0): a*r^2 + b*r + c = 0

(substitute y = e^(rx) -> characteristic equation in r)

CASE 1: Two distinct real roots r_1 not equal to r_2

y_h = C_1*e^(r_1*x) + C_2*e^(r_2*x)
 CASE 2: Repeated real root r_1 = r_2 = r
y_h = (C_1 + C_2*x) * e^(r*x)

CASE 3: Complex conjugate roots r = alpha +/- beta*j

y_h = e^(alpha*x) * [C_1*cos(beta*x) + C_2*sin(beta*x)]

STEP 2: Find particular solution y_p (method of undetermined coefficients)

STEP 3: General solution: y = y_h + y_p
STEP 4: Apply initial conditions to find C_1, C_2

Example: Solve: y'' - 5y' + 6y = 0, with y(0)=2, y'(0)=5

Characteristic equation: r^2 - 5r + 6 = 0 -> (r-2)(r-3) = 0
r_1 = 2, r_2 = 3 -> Case 1 (distinct real roots)
y_h = C_1*e^(2x) + C_2*e^(3x)

Apply initial conditions:

y(0) = 2: C_1 + C_2 = 2 ... equation (i)
y'(0) = 5: 2*C_1 + 3*C_2 = 5 ... equation (ii)
From (i): C_1 = 2 - C_2
Substitute into (ii): 2(2-C_2) + 3*C_2 = 5 -> 4 + C_2 = 5 -> C_2 = 1, C_1 = 1

General solution: y = e^(2x) + e^(3x)

 9. Laplace Transforms — Solving ODEs Algebraically the table and method

9.1 Definition and purpose

The Laplace transform converts a function of time f(t) into a function of complex frequency F(s). This transforms a
differential equation in t into an algebraic equation in s — which is much easier to solve.
Definition:
L{f(t)} = F(s) = integral[0 to inf] f(t) * e^(-s*t) dt

Workflow (the power of Laplace):

ODE in t -> [Laplace transform] ->
Algebraic equation in s
| solve for F(s)
Solution y(t) <- [Inverse Laplace] <- F(s)

Linearity (most important property):

L{a*f(t) + b*g(t)} = a*F(s) + b*G(s)

Key derivative properties:

L{f'(t)} = s*F(s) - f(0)
L{f''(t)} = s^2*F(s) - s*f(0) - f'(0)
(Initial conditions y(0), y'(0) enter AUTOMATICALLY — no need to find C_1, C_2 separately!)

9.2 Laplace transform table — memorise all of these

f(t) L{f(t)} = F(s) f(t) L{f(t)} = F(s)

1 1/s sin(at) a / (s^2 + a^2)

t 1/s^2 cos(at) s / (s^2 + a^2)

t^n n! / s^(n+1) sinh(at) a / (s^2 - a^2)

e^(at) 1 / (s - a) cosh(at) s / (s^2 - a^2)

t*e^(at) 1 / (s-a)^2 t*sin(at) 2as / (s^2+a^2)^2

delta(t) impulse 1 u(t) unit step 1/s

Shifting theorems:
First shifting: L{e^(at)*f(t)} = F(s-a) (shift in s-domain)

Partial fractions — needed to invert F(s):

Decompose F(s) into simple fractions recognisable from the table.
1/(s+a)(s+b) = A/(s+a) + B/(s+b) find A and B by substituting s=-a and s=-b
1/(s+a)^2 = A/(s+a) + B/(s+a)^2 use cover-up method for A, compare for B

9.3 Complete ODE solution using Laplace — worked example

Example: Solve y'' + 3y' + 2y = 0, with y(0)=1 and y'(0)=0 using Laplace transforms

Step 1: Take Laplace of every term Step 4: Partial fraction decomposition

L{y''} + 3*L{y'} + 2*L{y} = 0 (s+3) / [(s+1)(s+2)] = A/(s+1) + B/(s+2)
[s^2*Y - s*1 - 0] + 3*[s*Y - 1] + 2*Y = 0 s+3 = A*(s+2) + B*(s+1)
(substituting y(0)=1, y'(0)=0) s=-1: 2 = A*(1) -> A = 2
s=-2: 1 = B*(-1) -> B = -1
Step 2: Collect terms in Y(s) Y(s) = 2/(s+1) - 1/(s+2)
s^2*Y - s + 3s*Y - 3 + 2Y = 0
Y*(s^2 + 3s + 2) = s + 3 Step 5: Inverse Laplace using table
y(t) = 2*e^(-t) - e^(-2t)
Step 3: Solve for Y(s)
Y = (s+3) / (s^2+3s+2) = (s+3) / [(s+1)(s+2)]

Verification: y(0) = 2-1 = 1 checkmark y'(0) = -2+2 = 0 checkmark

 10. Numerical Methods — Root Finding & Integration Newton-Raphson, Bisection,
Simpson's rule

10.1 Bisection method — simple and guaranteed

If f(a) and f(b) have opposite signs, a root lies between a and b (Intermediate Value Theorem). Bisection repeatedly
halves the interval until the root is pinpointed to any desired accuracy.
Procedure:
1. Choose a, b such that f(a)*f(b) < 0 (opposite signs -> root between them)
2. Compute midpoint: c = (a + b) / 2
3. If |f(c)| < tolerance -> c is the root. STOP.
If f(a)*f(c) < 0 -> root in [a, c] -> set b = c
If f(c)*f(b) < 0 -> root in [c, b] -> set a = c
4. Repeat from step 2

Convergence rate: LINEAR (slow — one correct bit per iteration)

Error after n iterations: |error| <= (b-a) / 2^n
Advantage: ALWAYS converges if starting interval brackets the root

Example: Find root of f(x) = x^3 - x - 2 in [1, 2] — first 3 iterations

f(1) = 1-1-2 = -2 (negative), f(2) = 8-2-2 = 4 (positive) -> root in [1, 2]

Iter 1: c = (1+2)/2 = 1.5

f(1.5) = 3.375-1.5-2 = -0.125 (negative)
f(1.5)<0, f(2)>0 -> root in [1.5, 2]

Iter 2: c = (1.5+2)/2 = 1.75

f(1.75) = 5.359-1.75-2 = 1.609 (positive)
f(1.5)<0, f(1.75)>0 -> root in [1.5, 1.75]

Iter 3: c = (1.5+1.75)/2 = 1.625

f(1.625) = 4.291-1.625-2 = 0.666 (positive)
-> root in [1.5, 1.625]. True root is approximately 1.5214

10.2 Newton-Raphson method — fast quadratic convergence

Uses the tangent line at the current guess to extrapolate to a better guess. Quadratic convergence means the number
of correct digits roughly doubles with each iteration — much faster than bisection.
Iteration formula:
x_(n+1) = x_n - f(x_n) / f'(x_n)

Derivation: tangent line at (x_n, f(x_n)) is: y - f(x_n) = f'(x_n)*(x - x_n)

Set y = 0 (find x-intercept of tangent) -> x = x_n - f(x_n)/f'(x_n)

Convergence: QUADRATIC near the root (doubles correct digits each step)
Failure cases:
x_0 is far from the root -> may diverge or find wrong root
f'(x_n) near 0 (flat tangent) -> huge step, may diverge

Example: Find root of f(x) = x^3 - x - 2 starting from x_0 = 2 — two iterations
f(x) = x^3 - x - 2, f'(x) = 3x^2 - 1

Iter 1: x_1 = 2 - f(2)/f'(2) = 2 - (8-2-2)/(12-1) = 2 - 4/11 = 2 - 0.364 = 1.636

Iter 2: f(1.636) = 4.377-1.636-2 = 0.741
f'(1.636) = 3*(1.636)^2 - 1 = 3*2.676-1 = 7.029-1 = 6.029
x_2 = 1.636 - 0.741/6.029 = 1.636 - 0.123 = 1.513

After just 2 iterations: 1.513 (true root = 1.5214 — already very close!)
Compare: bisection needed 3 iterations just to narrow to [1.5, 1.625]
10.3 Trapezoidal and Simpson's 1/3 rule — numerical integration

Trapezoidal rule (linear approximation):

integral[a to b] f(x) dx ~= (h/2) * [f(x_0) + 2f(x_1) + 2f(x_2) + ... + 2f(x_(n-1)) + f(x_n)]
h = (b-a)/n (strip width)
Error order: O(h^2)

Simpson's 1/3 rule (parabolic approximation — more accurate):

integral[a to b] f(x) dx ~= (h/3) * [f(x_0) + 4f(x_1) + 2f(x_2) + 4f(x_3) + ... + 4f(x_(n-1))
+ f(x_n)]
Coefficients pattern: 1, 4, 2, 4, 2, ..., 2, 4, 1 (always starts and ends with 1)
REQUIREMENT: n must be EVEN (odd number of function evaluations)
Error order: O(h^4) — much more accurate than trapezoidal for smooth functions

Example: Approximate integral[0 to 1] e^x dx using Simpson's 1/3 rule, n=4 strips
h = (1-0)/4 = 0.25
x_0=0.00: f(0.00) = e^0.00 = 1.0000
x_1=0.25: f(0.25) = e^0.25 = 1.2840
x_2=0.50: f(0.50) = e^0.50 = 1.6487
x_3=0.75: f(0.75) = e^0.75 = 2.1170
x_4=1.00: f(1.00) = e^1.00 = 2.7183

Simpson: (h/3)*[f_0 + 4f_1 + 2f_2 + 4f_3 + f_4]

= (0.25/3)*[1.0000 + 4(1.2840) + 2(1.6487) + 4(2.1170) + 2.7183]
= (0.0833)*[1.0000 + 5.1360 + 3.2974 + 8.4680 + 2.7183]
= (0.0833) * 20.6197 = 1.7179

True value: e^1 - e^0 = 2.7183 - 1 = 1.7183 Error: only 0.0004!

 Quick Reference — All Key Formulas

Topic Formula / Key fact Notes

L'Hopital's rule lim f/g = lim f'/g' when 0/0 or inf/inf Differentiate top and bottom SEPARATELY

Derivative definition f'(x) = lim(h->0) [f(x+h)-f(x)]/h Instantaneous rate of change

Power rule d/dx[x^n] = n*x^(n-1) Most-used single rule

Chain rule d/dx[f(g(x))] = f'(g(x))*g'(x) Outer derivative times inner derivative

Product rule d/dx[f*g] = f'g + fg' 'First times derivative of second plus...'

Quotient rule d/dx[f/g] = (f'g - fg') / g^2 Denominator squared on the bottom

Integral by substitution Let u=g(x), then du=g'(x)dx; integrate in u Replace composite function with u

Integration by parts integral u dv = u*v - integral v du LIATE: L>I>A>T>E for choosing u

Maxima/minima step 1 Solve f'(x) = 0 to find critical points Also check where f'(x) undefined

Second derivative test f''(c)>0 = min; f''(c)<0 = max; =0 inconclusive Min = concave up; Max = concave down

Differentiate w.r.t. x, treat all other vars as

Partial derivative df/dx Standard differentiation rules apply
constants

Gradient vector grad(f) = (df/dx)i + (df/dy)j + (df/dz)k Points in direction of steepest increase

Jacobian matrix J_ij = df_i/dx_j (all partial derivatives) det(J)=0 means singular transformation

Jacobian determinant (2x2) det(J) = J_11*J_22 - J_12*J_21 Same as 2x2 matrix determinant

2x2 determinant ad - bc Memorise — used everywhere

3x3 determinant Cofactor expansion along row 1 (3 cofactors) Alternate signs: +, -, +

Characteristic equation det(A - lambda*I) = 0 Roots are the eigenvalues lambda

Eigenvalue-eigenvector A*v = lambda*v -> (A-lambda*I)v = 0 Solve for v after finding lambda

1st order linear ODE dy/dx + P(x)y = Q(x) -> mu = e^(integral P dx) Multiply both sides by integrating factor mu

2nd order homogeneous ar^2 + br + c = 0 -> characteristic roots 3 cases: real distinct, repeated, complex

r = alpha +/- beta*j -> e^(ax)(C_1*cos(bx)

Complex roots (Case 3) Most common case in practice
+C_2*sin(bx))

Laplace: e^(at) L{e^(at)} = 1/(s-a) Most important entry in the table

Laplace: sin(at) L{sin(at)} = a/(s^2+a^2) Note: sin uses a in numerator

Laplace: cos(at) L{cos(at)} = s/(s^2+a^2) Note: cos uses s in numerator

Laplace of y'' s^2*Y - s*y(0) - y'(0) Initial conditions enter automatically

Newton-Raphson iteration x_(n+1) = x_n - f(x_n)/f'(x_n) Quadratic convergence near root

Simpson's 1/3 rule (h/3)*[f_0 + 4f_1 + 2f_2 + 4f_3 + ... + f_n] n must be EVEN; coefficients: 1,4,2,4,...,4,1