NUMERICAL METHODS

Complete Formula Guide & Solved Examples

Euler • Modified Euler • Runge-Kutta 4th Order

Newton Interpolation Derivatives • Newton-Cotes Rules

Trapezoidal • Simpson's 1/3 & 3/8 • Gauss-Legendre Quadrature

Easy to understand — desi style! ■■

■ Contents

# Topic Description

1 Euler's Method Solving ODEs step-by-step

2 Modified Euler's Method Predictor-Corrector approach

3 Runge-Kutta 4th Order (RK4) The most popular ODE solver

Newton's Forward Interpolation

4 Derivative Derivative at start of table

Newton's Backward Interpolation

5 Derivative Derivative at end of table

6 Newton-Cotes Quadrature General numerical integration

7 Trapezoidal Rule Simplest integration rule

8 Simpson's 1/3 Rule Parabolic approximation

9 Simpson's 3/8 Rule Cubic approximation

10 Gauss-Legendre 2-Point Rule Optimal 2-point quadrature

11 Gauss-Legendre 3-Point Rule Optimal 3-point quadrature

UNIT 1

Euler's Method
Simplest ODE Solver — Like walking step by step!

What is it?
We have a differential equation: dy/dx = f(x, y) with initial condition y(x0) = y0.

Euler's method says: "From current point, move a tiny step h in the x-direction, and estimate y using the
current slope."

Think of it like this — you're walking on a hilly road. At each step, you look at the current slope and take a
small step. You don't look ahead, just the current direction!

■ Euler's Formula
yn+1 = yn + h · f(xn, yn)

xn+1 = xn + h

Step-by-Step Process
Step 1: Start with given x0, y0 and step size h.

Step 2: Calculate slope k = f(xn, yn)

Step 3: Apply: yn+1 = yn + h·k

Step 4: Update xn+1 = xn + h

Step 5: Repeat until you reach the required x value.

■ Quick Tip: Remember h (step size) — smaller h = more accurate result but more calculations!

✏■ Solved Example
Q: Solve dy/dx = x + y, y(0) = 1. Find y(0.2) using Euler's method with h = 0.1

Given: f(x,y) = x + y, x■ = 0, y■ = 1, h = 0.1

Step 1: Find y(0.1)

k■ = f(0, 1) = 0 + 1 = 1
y■ = y■ + h·k■ = 1 + 0.1 × 1 = 1.1
x■ = 0 + 0.1 = 0.1

Step 2: Find y(0.2)

k■ = f(0.1, 1.1) = 0.1 + 1.1 = 1.2
y■ = y■ + h·k■ = 1.1 + 0.1 × 1.2 = 1.1 + 0.12 = 1.22
x■ = 0.1 + 0.1 = 0.2

■ y(0.2) ≈ 1.22 (Exact value = 1.2428, error = 0.0228)

UNIT 2

Modified Euler's Method

Predictor-Corrector — First guess, then correct!

What's the idea?

Euler's method only uses the slope at the START of the interval. Modified Euler's method is smarter — it
predicts a value first, then corrects it using the average of slopes at both ends.

Like this: You take one step forward (predict), look around, average the old and new slopes, then correct
your position. Much more accurate!

■ Modified Euler's Formula (Heun's Method)

Predictor: y*n+1 = yn + h · f(xn, yn)

Corrector: yn+1 = yn + (h/2) · [ f(xn, yn) + f(xn+1, y*n+1) ]

Step-by-Step Process
Step 1: Predict y* using simple Euler formula.

Step 2: Calculate slope at predicted point: f(xn+1, y*)

Step 3: Average the two slopes: (slope at start + slope at predicted end) / 2

Step 4: Correct y using this average slope.

Step 5: Repeat for next step.

■ Quick Tip: The correction step can be iterated multiple times for better accuracy. Usually 1-2 corrections
are enough.
✏■ Solved Example
Q: Solve dy/dx = x + y, y(0) = 1. Find y(0.1) using Modified Euler's method with h = 0.1

Given: f(x,y) = x + y, x■ = 0, y■ = 1, h = 0.1

PREDICTOR step:
y* = y■ + h · f(x■, y■)
y* = 1 + 0.1 × f(0, 1)
y* = 1 + 0.1 × (0 + 1) = 1 + 0.1 = 1.1

CORRECTOR step:
Slope at start = f(0, 1) = 0 + 1 = 1.0
Slope at end = f(0.1, 1.1) = 0.1 + 1.1 = 1.2
Average slope = (1.0 + 1.2) / 2 = 1.1

y■ = y■ + h × (average slope)
y■ = 1 + 0.1 × 1.1 = 1 + 0.11 = 1.11

■ y(0.1) ≈ 1.11 (Exact ≈ 1.1103 — super close! ■)

UNIT 3

Runge-Kutta 4th Order (RK4)

The Gold Standard ODE Solver — 4 slopes, 1 genius formula!

What's the idea?

RK4 calculates 4 different slopes at different points within the interval and combines them as a weighted
average. It's like checking the slope at the start, twice in the middle, and once at the end — giving excellent
accuracy!

RK4 is the most widely used method in practice. When someone says 'numerical ODE solver', 9 out of 10
times they mean RK4!

■ RK4 Formulas
k1 = h · f(xn, yn)

k2 = h · f(xn + h/2, yn + k1/2)

k3 = h · f(xn + h/2, yn + k2/2)

k4 = h · f(xn + h, yn + k3)

yn+1 = yn + (1/6)(k1 + 2k2 + 2k3 + k4)

Weights breakdown: k■ gets weight 1, k■ gets weight 2, k■ gets weight 2, k■ gets weight 1. Total = 6. So
we divide by 6.

Think of it as: Start slope + 2×(Mid slope 1) + 2×(Mid slope 2) + End slope, all divided by 6

■ Quick Tip: Always calculate k1 → k2 → k3 → k4 in ORDER. Each depends on the previous one!
✏■ Solved Example
Q: Solve dy/dx = xy, y(0) = 1. Find y(0.1) using RK4 with h = 0.1

Given: f(x,y) = xy, x■ = 0, y■ = 1, h = 0.1

Calculate k■:
k■ = h · f(x■, y■) = 0.1 × f(0, 1) = 0.1 × (0 × 1) = 0.1 × 0 = 0

Calculate k■:
k■ = h · f(x■ + h/2, y■ + k■/2)
= 0.1 × f(0 + 0.05, 1 + 0)
= 0.1 × f(0.05, 1) = 0.1 × (0.05 × 1) = 0.1 × 0.05 = 0.005

Calculate k■:
k■ = h · f(x■ + h/2, y■ + k■/2)
= 0.1 × f(0.05, 1 + 0.0025)
= 0.1 × f(0.05, 1.0025) = 0.1 × (0.05 × 1.0025) = 0.1 × 0.050125 = 0.0050125

Calculate k■:
k■ = h · f(x■ + h, y■ + k■)
= 0.1 × f(0.1, 1 + 0.0050125)
= 0.1 × f(0.1, 1.0050125) = 0.1 × (0.1 × 1.0050125) = 0.1 × 0.10050125 = 0.010050

Final calculation:
y■ = y■ + (1/6)(k■ + 2k■ + 2k■ + k■)
= 1 + (1/6)(0 + 2(0.005) + 2(0.0050125) + 0.010050)
= 1 + (1/6)(0 + 0.01 + 0.010025 + 0.010050)
= 1 + (1/6)(0.030075)
= 1 + 0.0050125
■ y(0.1) ≈ 1.00501 (Exact = e^0.005 ≈ 1.00501 — virtually perfect! ■)
UNIT 4

Newton's Forward Interpolation — Derivatives

Use at the BEGINNING of the table

Difference Table — Build this first!

Given tabulated values, we first build a Forward Difference Table:

∆y■ = y■ - y■ (1st difference)

∆²y■ = ∆y■ - ∆y■ (2nd difference)

∆³y■ = ∆²y■ - ∆²y■ ... and so on.

Use p = (x - x■) / h where x■ is the first value and h is the equal step size.

■ Newton's Forward Derivative Formulas

dy/dx = (1/h)[ ∆y■ + ((2p-1)/2)∆²y■ + ((3p²-6p+2)/6)∆³y■ + ... ]

d²y/dx² = (1/h²)[ ∆²y■ + (p-1)∆³y■ + ((6p²-18p+11)/12)∆■y■ + ... ]

At x = x■ (i.e., p = 0):

(dy/dx) at x■ = (1/h)[ ∆y■ - (1/2)∆²y■ + (1/3)∆³y■ - (1/4)∆■y■ + ... ]

■ Quick Tip: Use FORWARD formula when x is near the BEGINNING of the table. p will be small (near 0).
✏■ Solved Example
Q: Given: x = 1.0, 1.2, 1.4, 1.6 and y = 2.7183, 3.3201, 4.0552, 4.9530. Find dy/dx at x = 1.0

Step 1: Build Forward Difference Table (h = 0.2)

x y ∆y ∆²y ∆³y
1.0 2.7183
0.6018
1.2 3.3201 0.1333
0.7351 0.0294
1.4 4.0552 0.1627
0.8978
1.6 4.9530

Step 2: At x = x■ = 1.0, use p = 0 formula:

dy/dx = (1/h)[∆y■ - (1/2)∆²y■ + (1/3)∆³y■]
dy/dx = (1/0.2)[0.6018 - (1/2)(0.1333) + (1/3)(0.0294)]
dy/dx = 5 × [0.6018 - 0.0667 + 0.0098]
dy/dx = 5 × 0.5449
■ dy/dx at x = 1.0 ≈ 2.7245 (Note: e^1.0 = 2.7183 — very close! ■)
UNIT 5

Newton's Backward Interpolation — Derivatives

Use at the END of the table

Key Difference from Forward

Same idea — but now we use Backward Differences (∇) and measure p from the LAST point x■.

∇y■ = y■ - y■■■ (1st backward difference)

∇²y■ = ∇y■ - ∇y■■■ (2nd backward difference)

Use p = (x - x■) / h where x■ is the LAST x-value.

■ Newton's Backward Derivative Formulas

dy/dx = (1/h)[ ∇y■ + ((2p+1)/2)∇²y■ + ((3p²+6p+2)/6)∇³y■ + ... ]

At x = x■ (i.e., p = 0):

(dy/dx) at x■ = (1/h)[ ∇y■ + (1/2)∇²y■ + (1/3)∇³y■ + (1/4)∇■y■ + ... ]

■ Quick Tip: Use BACKWARD formula when x is near the END of the table. p will be 0 or small negative.
✏■ Solved Example
Q: Using the same table: x = 1.0, 1.2, 1.4, 1.6 and y = 2.7183, 3.3201, 4.0552, 4.9530. Find dy/dx at x =
1.6

Step 1: Build Backward Difference Table (h = 0.2)

x y ∇y ∇²y ∇³y
1.0 2.7183
0.6018
1.2 3.3201 0.1333
0.7351 0.0294
1.4 4.0552 0.1627
0.8978
1.6 4.9530

At x■ = 1.6, p = 0:
∇y■ = 0.8978, ∇²y■ = 0.1627, ∇³y■ = 0.0294

dy/dx = (1/h)[∇y■ + (1/2)∇²y■ + (1/3)∇³y■]

dy/dx = (1/0.2)[0.8978 + (1/2)(0.1627) + (1/3)(0.0294)]
dy/dx = 5 × [0.8978 + 0.0814 + 0.0098]
dy/dx = 5 × 0.9890
■ dy/dx at x = 1.6 ≈ 4.9450 (e^1.6 = 4.9530 — excellent! ■)
UNIT 6

Newton-Cotes Quadrature Rule

The Family of Integration Rules — Parent of Trapezoidal & Simpson's!

What is it?
Newton-Cotes formulas are a group of numerical integration rules based on polynomial interpolation. They
all assume equally spaced points.

The general idea: Instead of integrating f(x) directly (which may be complicated), we approximate f(x) with a
simple polynomial and integrate that.

The general Newton-Cotes formula of order n is:

■ General Newton-Cotes Formula

∫[a to b] f(x) dx ≈ h · Σ w■ · f(x■)

where h = (b-a)/n, and w■ are the Newton-Cotes weights

n=1: Trapezoidal Rule | n=2: Simpson's 1/3 | n=3: Simpson's 3/8

Closed formulas include endpoints. Open formulas exclude endpoints.

All Trapezoidal, Simpson's rules are CLOSED Newton-Cotes formulas.

■ Quick Tip: Higher order Newton-Cotes isn't always better! For n ≥ 8, they can actually become unstable
(Runge's phenomenon). Use composite rules instead.

✏■ Solved Example
Q: Evaluate ∫[0 to 1] x² dx using Newton-Cotes (n=1 subintervals) — Trapezoidal, to see the concept

The Newton-Cotes n=1 formula (Trapezoidal):

∫ f(x) dx ≈ h/2 × [f(x■) + f(x■)]

With a=0, b=1, n=1: h = (1-0)/1 = 1

f(0) = 0² = 0
f(1) = 1² = 1

∫[0 to 1] x² dx ≈ (1/2) × [0 + 1] = 0.5

Exact answer = [x³/3] from 0 to 1 = 1/3 = 0.333...

■ Approx = 0.5 (Error = 0.167 — Newton-Cotes n=2 will be much better!)
UNIT 7

Trapezoidal Rule
Approximate the curve using trapezoids — Simple and effective!

The Idea
Divide [a, b] into n equal strips of width h = (b-a)/n. In each strip, approximate the curve as a straight line →
trapezoid shape.

Area of each trapezoid = h/2 × (sum of two heights). Add them all up!

■ Composite Trapezoidal Rule

∫[a to b] f(x) dx ≈ (h/2) [ y■ + 2y■ + 2y■ + ... + 2y■■■ + y■ ]

where h = (b-a)/n and y■ = f(x■)

Pattern: First + Last + 2×(All Middle terms)

Error = -(b-a)h²/12 × f''(ξ) for some ξ in [a,b]

■ Quick Tip: Memory trick: '1-2-2-2-...-2-1' pattern multiplied by h/2. First and last have coefficient 1, all
middle have coefficient 2!

✏■ Solved Example
Q: Evaluate ∫[0 to 1] 1/(1+x) dx using Trapezoidal rule with n=4 strips

Setup: a=0, b=1, n=4, h=(1-0)/4 = 0.25

Build the table:

x■ = 0.00, y■ = 1/(1+0) = 1.0000
x■ = 0.25, y■ = 1/(1+0.25) = 0.8000
x■ = 0.50, y■ = 1/(1+0.50) = 0.6667
x■ = 0.75, y■ = 1/(1+0.75) = 0.5714
x■ = 1.00, y■ = 1/(1+1) = 0.5000

Apply formula:
∫ ≈ (h/2)[y■ + 2y■ + 2y■ + 2y■ + y■]
= (0.25/2)[1.0000 + 2(0.8000) + 2(0.6667) + 2(0.5714) + 0.5000]
= 0.125 × [1.0000 + 1.6000 + 1.3334 + 1.1428 + 0.5000]
= 0.125 × 5.5762
= 0.6970
■ ≈ 0.6970 (Exact = ln(2) = 0.6931 — error is only 0.004! ■)
UNIT 8

Simpson's 1/3 Rule

Fit parabolas instead of lines — More accurate than Trapezoidal!

The Idea
Instead of approximating with straight lines (trapezoids), we use parabolas through every 3 consecutive
points. More accurate!

IMPORTANT: n must be EVEN (2, 4, 6, 8, ...) for Simpson's 1/3 rule.

■ Composite Simpson's 1/3 Rule

∫[a to b] f(x) dx ≈ (h/3) [ y■ + 4y■ + 2y■ + 4y■ + 2y■ + ... + 4y■■■ + y■ ]

Pattern: 1 - 4 - 2 - 4 - 2 - ... - 4 - 1 (multiply by h/3)

Error = -(b-a)h■/180 × f''''(ξ) → Much smaller error than Trapezoidal!

■ Quick Tip: Memory trick: '1-4-2-4-2-...-4-1'. First=1, Last=1, Odd positions=4, Even positions=2. Multiply
by h/3.

✏■ Solved Example
Q: Evaluate ∫[0 to 1] 1/(1+x) dx using Simpson's 1/3 rule with n=4

Setup: a=0, b=1, n=4 (even ■), h=0.25

Same table as before:

y■ = 1.0000 (coeff = 1)
y■ = 0.8000 (coeff = 4, odd index)
y■ = 0.6667 (coeff = 2, even middle)
y■ = 0.5714 (coeff = 4, odd index)
y■ = 0.5000 (coeff = 1)

Apply formula:
∫ ≈ (h/3)[y■ + 4y■ + 2y■ + 4y■ + y■]
= (0.25/3)[1.0000 + 4(0.8000) + 2(0.6667) + 4(0.5714) + 0.5000]
= (0.0833)[1.0000 + 3.2000 + 1.3334 + 2.2856 + 0.5000]
= 0.0833 × 8.3190
= 0.6932
■ ≈ 0.6932 (Exact = 0.6931 — ALMOST PERFECT! ■ Simpson's wins!)
UNIT 9

Simpson's 3/8 Rule

Fit cubics through 4 points — Use when n is a multiple of 3!

When to use 3/8 rule?

Use Simpson's 3/8 rule when n is a multiple of 3 (3, 6, 9, ...). It uses a cubic (degree 3) polynomial through
every 4 consecutive points.

It's slightly more accurate than 1/3 rule per interval but requires 3 sub-intervals at a time.

■ Composite Simpson's 3/8 Rule

∫[a to b] f(x) dx ≈ (3h/8) [ y■ + 3y■ + 3y■ + 2y■ + 3y■ + 3y■ + 2y■ + ... + y■ ]

Pattern: 1 - 3 - 3 - 2 - 3 - 3 - 2 - ... - 3 - 3 - 1 (multiply by 3h/8)

n must be a multiple of 3!

■ Quick Tip: Pattern for 3/8: '1-3-3-2-3-3-2-...-3-3-1'. First=1, Last=1, Every 3rd interior=2, All others=3.

✏■ Solved Example
Q: Evaluate ∫[0 to 3] x² dx using Simpson's 3/8 rule with n=3

Setup: a=0, b=3, n=3 (multiple of 3 ■), h=(3-0)/3 = 1

Build the table:

x■ = 0, y■ = 0² = 0 (coeff = 1)
x■ = 1, y■ = 1² = 1 (coeff = 3)
x■ = 2, y■ = 2² = 4 (coeff = 3)
x■ = 3, y■ = 3² = 9 (coeff = 1)

Apply formula:
∫ ≈ (3h/8)[y■ + 3y■ + 3y■ + y■]
= (3×1/8)[0 + 3(1) + 3(4) + 9]
= (3/8)[0 + 3 + 12 + 9]
= (3/8) × 24
= 72/8 = 9
■ = 9 (Exact = [x³/3] from 0 to 3 = 27/3 = 9 — EXACT! ■)
UNIT 10

Gauss-Legendre 2-Point Quadrature

Optimal 2-point formula — No equal spacing needed!

What's different about Gaussian Quadrature?

Newton-Cotes methods use equally spaced points. Gaussian quadrature is smarter — it chooses the
best points (not equally spaced) to maximize accuracy.

For n points, Gaussian quadrature gives exact results for polynomials up to degree 2n-1. Two points →
exact for degree 3 polynomials. Amazing!

Key step: First change the limits from [a, b] to [-1, 1] using substitution: x = ((b+a)/2) + ((b-a)/2)·t

■ 2-Point Gauss-Legendre Formula

∫[-1 to 1] f(t) dt ≈ f(t■) + f(t■)

where t■ = -1/√3 ≈ -0.5774 and t■ = +1/√3 ≈ +0.5774

Both weights w■ = w■ = 1

For [a,b]: x = ((b-a)/2)·t + ((a+b)/2), multiply result by (b-a)/2

■ Quick Tip: The points t■ and t■ are the roots of the Legendre polynomial P■(t) = (3t²-1)/2. Weights are
both 1.
✏■ Solved Example
Q: Evaluate ∫[0 to 1] e^x dx using 2-point Gauss-Legendre quadrature

Step 1: Transform limits from [0,1] to [-1,1]

a = 0, b = 1
x = ((b-a)/2)·t + (a+b)/2 = (0.5)·t + 0.5
dx = 0.5 dt
So: ∫[0 to 1] e^x dx = 0.5 × ∫[-1 to 1] e^(0.5t+0.5) dt

Step 2: Apply 2-point formula

t■ = -0.5774, x■ = 0.5(-0.5774) + 0.5 = 0.2113
t■ = +0.5774, x■ = 0.5(+0.5774) + 0.5 = 0.7887

f(t■) = e^(x■) = e^0.2113 = 1.2353

f(t■) = e^(x■) = e^0.7887 = 2.2008

Step 3: Sum with weights (both = 1) and multiply by (b-a)/2 = 0.5

∫ ≈ 0.5 × [w■·f(t■) + w■·f(t■)]
= 0.5 × [1×1.2353 + 1×2.2008]
= 0.5 × 3.4361
= 1.7181
■ ≈ 1.7181 (Exact = e¹ - e■ = 1.7183 — only 2 points and nearly exact! ■)
UNIT 11

Gauss-Legendre 3-Point Quadrature

3 optimal points → exact for polynomials up to degree 5!

3-Point Formula
With 3 cleverly chosen points and weights, we get an exact result for any polynomial up to degree 2×3-1 =
5. That's incredible — 3 points doing the work of 6!

The 3 points are the roots of the Legendre polynomial P■(t) = (5t³ - 3t)/2.

■ 3-Point Gauss-Legendre Formula

∫[-1 to 1] f(t) dt ≈ w■·f(t■) + w■·f(t■) + w■·f(t■)

Points: t■ = -√(3/5) ≈ -0.7746, t■ = 0, t■ = +√(3/5) ≈ +0.7746

Weights: w■ = 5/9 ≈ 0.5556, w■ = 8/9 ≈ 0.8889, w■ = 5/9 ≈ 0.5556

For [a,b]: multiply result by (b-a)/2 and substitute x = ((b-a)/2)·t + ((a+b)/2)

■ Quick Tip: Remember weights: outer two = 5/9, middle one = 8/9. Total = 5/9 + 8/9 + 5/9 = 18/9 = 2
(always adds to 2 for [-1,1] interval).

Quick Comparison: Gauss-Legendre Points & Weights

n Points t■ Weights w■ Exact for degree

1 t=0 w=2 ≤1

2 ±0.5774 1, 1 ≤3

3 0, ±0.7746 8/9, 5/9, 5/9 ≤5

4 ±0.3399, ±0.8611 0.6521, 0.3479, ... ≤7

✏■ Solved Example
Q: Evaluate ∫[0 to 1] x³ dx using 3-point Gauss-Legendre quadrature

Step 1: Transform [0,1] → [-1,1]

x = 0.5t + 0.5, dx = 0.5 dt
∫[0 to 1] x³ dx = 0.5 × ∫[-1 to 1] (0.5t+0.5)³ dt

Step 2: Calculate at 3 Gauss points

t■ = -0.7746 → x■ = 0.5(-0.7746) + 0.5 = 0.1127 → x■³ = 0.001433
t■ = 0 → x■ = 0.5(0) + 0.5 = 0.5 → x■³ = 0.125000
t■ = +0.7746 → x■ = 0.5(0.7746) + 0.5 = 0.8873 → x■³ = 0.698050

Step 3: Apply weights and multiply by (b-a)/2 = 0.5

∫ ≈ 0.5 × [w■·f■ + w■·f■ + w■·f■]
= 0.5 × [(5/9)(0.001433) + (8/9)(0.125000) + (5/9)(0.698050)]
= 0.5 × [0.000796 + 0.111111 + 0.387806]
= 0.5 × 0.499713
= 0.24986
■ ≈ 0.25000 (Exact = [x■/4] from 0 to 1 = 0.25 — EXACT to 4 decimals! ■)
UNIT ★

Quick Reference Cheat Sheet

All formulas at a glance — Print this page!

Method Formula Key Condition

Euler's y_{n+1} = y_n + h·f(x_n, y_n) Simple, least accurate

y* = Euler predict; y_{n+1} = y_n + h/2·[f_n +

Mod. Euler Predictor-Corrector
f(x_{n+1},y*)]

Most accurate ODE

RK4 y_{n+1} = y_n + (k1+2k2+2k3+k4)/6
solver

Fwd Deriv
dy/dx = (1/h)[∆y■ - ∆²y■/2 + ∆³y■/3 - ...] Use at start of table
(p=0)

Bwd Deriv
dy/dx = (1/h)[∇y_n + ∇²y_n/2 + ∇³y_n/3 + ...] Use at end of table
(p=0)

Trapezoidal (h/2)[y■ + 2(y■+...+y_{n-1}) + y_n] Any n; least accurate

Simpson 1/3 (h/3)[y■ + 4y■ + 2y■ + 4y■ + ... + y_n] n must be EVEN

Simpson 3/8 (3h/8)[y■ + 3y■ + 3y■ + 2y■ + ... + y_n] n multiple of 3

GL 2-point (b-a)/2 × [f(t■)+f(t■)], t=±0.5774, w=1 Exact up to degree 3

GL 3-point (b-a)/2 × [5/9·f(t■) + 8/9·f(t■) + 5/9·f(t■)] Exact up to degree 5

■ Quick Tip: For EXAM: ODE problems → check if Euler/[Link]/RK4. Derivative problems → build
difference table first. Integration → check if n is even (Simpson 1/3) or multiple of 3 (3/8). No equal spacing
needed? → Use Gauss-Legendre!

Made with ❤■ for students of Numerical Methods | All the best for your exams! ■