Numerical Methods For Engineers Formulas
Numerical Methods For Engineers Formulas
Absolute Error
𝐴𝐸 = 𝜀𝑎 = |𝑥 − 𝑥̂| Inverse of 2x2 Matrix
𝑎 𝑏 𝟏 𝑑 −𝑏
If 𝑨 = [ ] then 𝑨−𝟏 = [ ]
Absolute Relative Error 𝑐 𝑑 𝒂𝒅−𝒃𝒄 −𝑐 𝑎
|𝑥−𝑥̂| |𝑥̂𝑘+1 −𝑥̂𝑘 |
(true) 𝜀𝑟𝑒𝑙 = , (approx) 𝜀𝑟𝑒𝑙 = ,
|𝑥| |𝑥̂𝑘+1 | Determinant of an Upper/Lower triangular matrix
If 𝑨 is an 𝑛 × 𝑛 upper or lower triangular, then det (𝑨) is
Condition Number the product of diagonal elements.
𝑓 ′ (𝑥̂) 𝒏
𝜅 = |𝑥̂ | det(𝑨) = ∏ 𝑎𝑖𝑖
𝑓(𝑥̂)
𝒊=𝟏
Bisection Method
𝑓(𝑥𝑙 )𝑓(𝑥𝑢 ) < 0 Eigenvalue of A Square Matrix
𝑥𝑚 = (𝑥𝑙 + 𝑥𝑢 )/2 If 𝑨 is an 𝑛 × 𝑛 matrix, then the roots of the polynomial
from the determinant equality
Newton Raphson Method det(𝑨 − 𝜆𝑰) = 0
𝑓(𝑥𝑖 ) gives the eigenvalues of 𝑨.
𝑥𝑖+1 = 𝑥𝑖 −
𝑓′(𝑥𝑖 )
Condition Number of A Square Matrix
Numerical Derivative of First Order If 𝑨 is an 𝑛 × 𝑛 matrix, then the condition number of the
𝑓(𝑥𝑖+1 ) − 𝑓(𝑥𝑖 ) matrix is given by
𝑓(𝑥𝑖 ) = |𝜆𝑚𝑎𝑥 |
𝑥𝑖+1 − 𝑥𝑖 cond(𝑨) =
|𝜆𝑚𝑖𝑛 |
Derivative Rules
where 𝜆𝑚𝑎𝑥 and 𝜆𝑚𝑖𝑛 are the max and min valued
eigenvalues of 𝑨.
Jacobi Iterations
For 𝑨𝒙 = 𝒃, if D,U,L are the diagonal and upper lower
partitions of A, then
GS Iterations
For 𝑨𝒙 = 𝒃, if D,U,L are the diagonal and upper lower
Chain Rules partitions of A, then
Radius of Convergence
The iteration 𝒙𝑖+1 = 𝑩 𝒙𝑖 + 𝒄 converges if and only if
𝜌 (𝐵) < 1. 𝜌 (𝐵) is defined by 𝜌 (𝑩) = max (|𝜆𝑘 |), where
1≤𝑘≤𝑛
𝜆𝑘 are the eigenvalues of 𝑩.