FORMULA SHEET FOR NUMERICAL METHODS FOR ENGINEERS

MATH 472

Part 1: Error Analysis

Error Definitions:

True error: Et = true value − approximation

True percent relative error:

true value − approximation
εt = × 100%
true value

Approximate percent relative error:

present approximation − previous approximation
εa = × 100%
present approximation

Stopping criterion:
Terminate computation when εa < εs where εs s is the desired percent relative error.

Taylor Series:

Taylor series expansion:

f ′′ (xi ) 2 f ′′′ (xi ) 3 f (n) (xi ) n

f (x(i+1) ) = f (xi ) + f ′ (xi )h + h + h + ··· + h + Rn .
2! 3! n!

f (n+1) (xi ) n+1

where Rn = h
(n + 1)!

1
 Part 2: Roots of Equations

Method Formulation
xu + xl
Bisection xr = if f (xl ) × f (xr ) < 0, set xu = xr
2
if f (xl ) × f (xr ) > 0, set xl = xr
f (xu )(xl − xu )
False position xr = xu − if f (xl ) × f (xr ) < 0, set xu = xr
f (xl ) − f (xu )
if f (xl ) × f (xr ) > 0, set xl = xr
f (xi )
Newton Raphson xi+1 = xi −
f (xi )
f (xi )(xi−1 − xi )
Secant xi+1 = xi −
f (xi−1 ) − f (xi )

Part 3: Numerical Differentiation and Integration

A. Numerical Differentiation

Method Formulation Errors

Forward Finite First derivative
f ′′ (xi )
divided difference f ′ (xi ) = f (xi+1h)−f (xi ) Et = 2! h, O(h)
Backward Finite First derivative
′ f ′′ (xi )
divided difference f (xi ) = f (xi )−fh (xi−1 ) Et = 2! h, O(h)
Centered Finite First derivative
f ′′′ (xi ) 2
divided difference f ′ (xi ) = f (xi+1 )−f
2h
(xi−1 )
Et = 3! h , O(h2 )
Centered Finite Second derivative
f (4) (xi ) 2
divided difference f ′′ (xi ) = f (xi+1 )−2fh(x2 i )+f (xi−1 ) Et = 12 h , O(h2 )
Three point First derivative
f (3) (ξ) 2
End point f ′ (x0 ) = 1
2h [−3f (x0 ) + 4f (x0 + h) − f (x0 + 2h)] Et = 3 h
Three point First derivative
(3)
Midpoint f ′ (x0 ) = [f (x0 + h) − f (x0 − h)]
1
2h Et = − f 6
(ξ) 2
h
Five point First derivative
1 f (5) (ξ) 4
Midpoint f ′ (x0 ) = [f (x0 − 2h) − 8f (x0 − h) Et = 30 h
12h
+ 8f (x0 + h) − f (x0 + 2h)]
Five point First derivative
′ 1 f (5) (ξ) 4
Endpoint f (x0 ) = [−25f (x0 ) + 48f (x0 + h) Et = 5 h
12h
− 36f (x0 + 2h) + 16f (x0 + 3h) − 3f (x0 + 4h)]

2
B. Numerical Integration

b−a
First: Closed Newton-Cots Formulas: h =
n

Trapezoidal Rule n = 1, h = b − a
∫b 2
f ′′ (ξ)
n=1 a
f (x) dx = h2 [(x0 ) + f (x1 )] Et = − h 12
Simpson’s Rule n = 2, h = b−a
∫b 2
5
f (4) (ξ)
n=2 a
f (x) dx = h3 [f (x0 ) + 4f (x1 ) + f (x2 )] Et = − h 90
Simpson’s n = 3, h = b−a
∫b 3
5
f (4) (ξ)
Three-eighths a
f (x) dx = 3h
8 [f (x0 ) + 3f (x1 ) + 3f (x2 ) + f (x3 )] Et = − 3h 80

b−a
Second: Open Newton-Cots Formulas: h =
n+2

Midpoint Rule n = 0, h = b−a

∫b 2 3
f ′′ (ξ)
n=0 a
f (x) dx = 2hf (x0 ) Et = − h 3
n = 1, h = b−a
∫b 3
3h3 f ′′ (ξ)
n=1 a
f (x) dx = 3h2 [f (x0 ) + f (x1 )] Et = 4
n = 2, h = b−a
∫b 4
5
f (4) (ξ)
n=2 a 3 [2f (x0 ) − f (x1 ) + 2f (x2 )]
f (x) dx = 4h Et = − 14h 45
n = 3, h = b−a
∫b 5h
5
95h5 f (4) (ξ)
n=3 a
f (x) dx = 45 [11f (x0 ) + f (x1 ) + f (x2 ) + 11f (x3 )] Et = 144

b−a
Third: Composite Numerical Integration: h = , n is even
n
b−a
Composite h= n
∫ b ∑2
n
h (b−a)h4 f (4) (ξ)
Simpson’s rule f (x) dx = [f (x0 ) + 4 (f (x2j−1 )) Et = 180
a 3 j=1
2 −1
n
∑
+2 (f (x2j )) + f (xn )]
j=1

3
 Part 4: Curve Fitting

Method Formulation Errors

∏
3
Newton’s divided difference Third order R3 ∼
= f [x4 , x3 , x2 , x1 , x0 ] (x − xi )
i=0
interpolating polynomial f3 (x) = b0 + b1 (x − x0 )
+ b2 (x − x0 )(x − x1 )
+ b3 (x − x0 )(x − x1 )(x − x2 )
where b0 = f (x0 )
b1 = f [x1 , x0 ]
b2 = f [x2 , x1 , x0 ]
b3 = f [x3 , x2 , x1 , x0 ]
∏
n
Lagrange interpolating Rn ∼
= f [xn+1 , xn , xn−1 , . . . x0 ] (x − xi )
i=0

∑
i=n
polynomial fn (x) = Li (x)f (xi )
i=0
where
∏
n
(x − xj )
Li (x) =
(xi − xj )
j=0,j̸=i
√
Sr
Linear Regression y = a0 + a1 x S xy =
n−2
where St − Sr
2
r =
∑
i=n
St
Sr = (yi − a0 − a1 xi )2 where
∑
i=0
∑ ∑ ∑
n x y − xi yi St = (yi − y)2
a1 = ∑i i2 ∑
n xi − ( xi )2
a0 = y − a1 x
Polynomial Regression √
2 Sr
2nd order poly. fit y = a0 + a1 x + a2 x S xy =
n − (m + 1)
where
 ∑ ∑ 2    ∑  St − Sr
r2 =
∑n ∑ x2i ∑ xi3 a0 ∑ yi St
 xi  a1  =  xi yi 
∑ 2 ∑ xi3 x
∑ 4 i ∑ 2 where
xi xi xi a2 xi yi ∑
St = (yi − y)2

4
 Part 5: Linear Algebraic Equations

Method Formulation Criteria of Convergence

xji − xj−1
Gauss-Seidel |εa,i | = i
× 100% < εs
xji
b1 − a12 x2 − a13 x3
for 3 × 3 equations x1 =
a11
b2 − a21 x1 − a23 x3
x2 =
a22
b3 − a31 x1 − a32 x2
x3 =
a33

Part 6: Ordinary Differential Equations

Method Formulation
Euler’s First-Order RK yi+1 = yi + k1 h
k1 = f (xi , yi )
1 1
Heun’s Second Order RK yi+1 = yi + ( k1 + k2 )h
2 2
k1 = f (xi , yi )
k2 = f (xi + h, yi + k1 h)
Midpoint Second Order RK yi+1 = yi + k2 h
k1 = f (xi , yi )
1 1
k2 = f (xi + h, yi + k1 h)
2 2
1 2
Ralston’s Second Order RK yi+1 = yi + ( k1 + k2 )h
3 3
k1 = f (xi , yi )
3 3
k2 = f (xi + h, yi + k1 h)
4 4
1
Classical Fourth Order RK yi+1 = yi + (k1 + 2k2 + 2k3 + k4 )h
6
k1 = f (xi , yi )
1 1
k2 = f (xi + h, yi + k1 h)
2 2
1 1
k3 = f (xi + h, yi + k2 h)
2 2
k2 = f (xi + h, yi + k3 h)