0% found this document useful (0 votes)
65 views4 pages

Universidade Da Beira Interior: Algebra Linear Num Erica

The document discusses several topics in numerical analysis including: 1) Definitions of absolute and relative error in approximations. 2) Taylor polynomial approximations and numerical differentiation formulas. 3) Methods for finding zeros of functions like bisection, false position, Newton-Raphson, and secant methods. 4) Techniques for solving systems of linear equations like Cholesky decomposition, Jacobi, Gauss-Seidel, and conditions for convergence. 5) Interpolation using Lagrange and Newton polynomials and error estimates. 6) Numerical integration by changing variables to transform integrals into standard forms.

Uploaded by

Júlio Santos
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
65 views4 pages

Universidade Da Beira Interior: Algebra Linear Num Erica

The document discusses several topics in numerical analysis including: 1) Definitions of absolute and relative error in approximations. 2) Taylor polynomial approximations and numerical differentiation formulas. 3) Methods for finding zeros of functions like bisection, false position, Newton-Raphson, and secant methods. 4) Techniques for solving systems of linear equations like Cholesky decomposition, Jacobi, Gauss-Seidel, and conditions for convergence. 5) Interpolation using Lagrange and Newton polynomials and error estimates. 6) Numerical integration by changing variables to transform integrals into standard forms.

Uploaded by

Júlio Santos
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

Universidade da Beira Interior

Algebra Linear Numérica


Formulário
6 de Abril de 2020
ˆ Definição de erros Seja x̄ a aproximação do valor exacto x então

|x − x̄|
Ea (x̄) = |x − x̄| e Er (x̄) =
|x|

ˆ Polinómio de Taylor

(x − x0 )2 00 (x − x0 )3 (3) (x − x0 )n (n)
Pn (x) = f (x0 )+(x−x0 )f 0 (x0 )+ f (x0 )+ f (x0 )+· · ·+ f (x0 )
2 3! n!
(x − x0 )(n+1) (n+1)
Rn (x) = f (ξ) ξ ∈ (x, x0 )
(n + 1)!
ˆ Diferenciação numérica
f 0 (xi ) = f 00 (xi ) =

1
h
[f (xi+1 ) − f (xi )] − h2 f 00 (ξ) 1
h2
[f (xi ) − 2f (xi+1 ) + f (xi+2 )] − hf 000 (ξ)
h2 (4)
1
h
[f (xi ) − f (xi−1 )] + h2 f 00 (ξ) 1
h2
[f (xi−1 ) − 2f (xi ) + f (xi+1 )] − 12
f (ξ)
2
1
2h
[−3f (xi ) + 4f (xi+1 ) − f (xi+2 )] + h3 f 000 (ξ) 1
h2
[f (xi−2 ) 000
− 2f (xi−1 ) + f (xi )] + hf (ξ)
2
1
2h
[−f (xi−1 ) + f (xi+1 )] − h6 f 000 (ξ)
2
1
2h
[f (xi−2 ) − 4f (xi−1 ) + 3f (xi )] + h3 f 000 (ξ)

xi − xi−1 = h > 0, i = 1, 2, . . .

ˆ Condicionamento
cond(A) = kAk.kA−1 k

ˆ Localização das raı́zes


Teorema de Bolzano: Seja f (x) uma função contı́nua em [a, b], se f (a)f (b) < 0 então existe
c ∈ [a, b] tal que f (c) = 0.
Teorema de Rolle: Seja f (x) uma função contı́nua e derivável em [a, b], se f (a) = f (b) então
existe c ∈ [a, b] tal que f 0 (c) = 0.

ˆ Zeros de funções f (x) = 0


ak + b k b−a
– Bissecção xk = , Ea (xk ) ≤
, k = 0, 1, . . .
2 2k+1
b k − ak
– Corda falsa xk = bk − f (bk ) , k = 0, 1, . . .
f (bk ) − f (ak )
f (xk )
– Newton-Raphson xk+1 = xk − 0 , k = 0, 1, . . .
f (xk )

1
f (xk )
– Secante xk+1 = xk − (xk − xk−1 ), k = 1, 2, . . .
f (xk ) − f (xk−1 )
Condições de convergência para os métodos de Newton e da secante
- f ∈ C 2 ([a, b])
- f (a)f (b) < 0
- f 0 (x) 6= 0, x ∈ [a, b]
- f 00 (x) não muda de sinal em [a, b]
f (a) f (b)
- f 0 (a) ≤ b − a e f 0 (b) ≤ b − a ou ∃x0 ∈ [a, b] : f (x0 ) × f 00 (x0 ) > 0

– Ponto fixo x = G(x) xk+1 = G(xk ), |G0 (x)| < 1, ∀x ∈ [a, b], k = 0, 1, . . .

ˆ Sistemas de equações lineares Ax = b

– Cholesky A = U T U
v
u j−1
√ a1j u X
u11 = a11 , u1j = , ujj = ajj −
t u2kj , j = 2, . . . , n
u11 k=1

i−1
!
1 X
uij = aij − uki ukj , i = 2, . . . , n, j = i + 1, . . . , n uij = 0, i > j
uii k=1

– Normas
p
kxk1 = |x1 |+· · ·+|xn |, kxk2 = |x1 |2 + · · · + |xn |2 , kxk∞ = max {|x1 |, . . . , |xn |}
i=1,...,n

v
n n
u n 2
X X uX
kAk1 = max |aij |; kAk∞ = max |aij |; kAkE = t aij
j=1,...,n i=1,...,n
i=1 j=1 i,j=1

– Jacobi A = D + L + U
n
(k+1) bi X aij (k)
xi = − x , i = 1, . . . , n k = 0, 1, . . .
aii j=1
aii j
j6=i

x(k+1) = −D−1 (L + U )x(k) + D−1 b M = −D−1 (L + U )

– Gauss-Seidel A = D + L + U
i−1 n
(k+1) bi X aij (k+1) X aij (k)
xi = − xj − xj , i = 1, . . . n k = 0, 1, . . .
aii j=1 aii j=i+1
a ii

x(k+1) = −(D + L)−1 U x(k) + (D + L)−1 b M = −(D + L)−1 U


kM k kM kk
– Erro Ea (x(k) ) ≤ kx(k) − x(k−1) k ou Ea (x(k) ) ≤ kx(1) − x(0) k
1 − kM k 1 − kM k
– Condições de convergência: A matriz A do sistema ser estritamente diagonal dominante
ou a matriz de iteração M ter kM k < 1.

2
ˆ Interpolação polinomial Pn (xi ) = f (xi ), i = 0, 1, . . . , n

– Polinómio de Lagrange
n n
X Y x − xi
Pn (x) = f (xk )Lk (x), Lk (x) =
k=0 i=0,i6=k
x k − xi

– Polinómio de Newton
Diferenças Divididas
f (xj ) − f (xi )
* ordem 1 : fi,j = f [xi , xj ] =
xj − xi
fi+1,...,i+j − fi,...,i+j−1
* ordem j : fi,...,i+j = f [xi , . . . , xi+j ] =
xi+j − xi
n
X
Pn (x) = f (x0 ) + f0,...,k (x − x0 ) . . . (x − xk−1 ),
k=1

– Erro para os polinómios de Lagrange e de Newton


Se f (x) é conhecida:

max |f (n+1) (x)|


x∈(x0 ,xn )
|f (x) − Pn (x)| ≤ |(x − x0 ) . . . (x − xn )|, x ∈ (x0 , xn )
(n + 1)!

Se f (x) é desconhecida:
|f (x) − Pn (x)| ≤ M |(x − x0 ) . . . (x − xn )| onde M é o maior módulo das diferenças
divididas de ordem n + 1 e x ∈ (x0 , xn ).
Rb R1
ˆ Integração numérica I = b−a b+a b−a

a
f (x) dx = −1
g(w) dw, com g(w) = 2
f 2
+ 2
w

– Regras de Newton-Cotes de intervalo fechado


h h3
* Trapézio I = [f (a) + f (b)] − f 00 (ξ), h = b − a
2 12
h a+b  h5 (b − a)
* Simpson I = f (a) + 4f ( ) + f (b) − f (4) (ξ), h =
3 2 90 2
(b − a)
– Regras compostas de intervalo fechado h = , xi = a + hi, ξ ∈ (a, b)
n
n−1 i b−a
hh X
* Trapézio I = f (a) + 2 f (xj ) + f (b) − h2 f 00 (ξ),
2 j=1
12
(n/2)−1 (n/2) i b−a
hh X X
* Simpson I = f (a) + 2 f (x2j ) + 4 f (x2j−1 ) + f (b) − h4 f (4) (ξ)
3 j=1 j=1
180
– Regras de Newton-Cotes de intervalo aberto
3h 3h3 00 b−a
* I= [f (x1 ) + f (x2 )] + f (ξ), h =
2 4 3
4h 14h5 (4) (b − a)
* I= [2f (x1 ) − f (x2 ) + 2f (x3 )] + f (ξ), h=
3 45 4

3
ˆ Equações diferenciais ordinárias

y 0 = f (t, y)
– Problemas de valores iniciais: , t ∈ [a, b]
y(t ) = y
0 0

Rt
Picard : yi+1 (t) = y0 + t0
f (x, yi (x)) dx
Euler explı́cito: yi+1 = yi + hf (ti , yi )
Euler implı́cito: yi+1 = yi + hf (ti+1 , yi+1 )
Trapézios: yi+1 = yi + h2 [f (ti , yi ) + f (ti+1 , yi+1 )]
h2 0 hn (n−1)
Taylor de ordem n: yi+1 = yi + hf (ti , yi ) + 2
f (ti , yi ) + ··· + n!
f (ti , yi ); f 0 = ft0 + fy0 × f (t, y)
Runge-Kutta 2: yi+1 = yi + h2 (k1 + k2 ), k1 = f (ti , yi ), k2 = f (ti + h, yi + hk1 )
Runge-Kutta 4: yi+1 = yi + h6 (k1 + 2k2 + 2k3 + k4 ), k1 = f (ti , yi ), k2 = f (ti + h2 , yi + h2 k1 )
k3 = f (ti + h2 , yi + h2 k2 ), k4 = f (ti + h, yi + hk3 )

You might also like