v
向量的1-范数
∥ x ∥ 1 = ∑ k = 1 n ∣ ξ k ∣ \|x\|_1=\sum_{k=1}^n|\xi_k| ∥x∥1=∑k=1n∣ξk∣
向量的2-范数
∥ x ∥ 2 = ( ∑ k = 1 n ∣ ξ k ∣ 2 ) 1 2 \|x\|_2=(\sum_{k=1}^n|\xi_k|^2)^\frac12 ∥x∥2=(∑k=1n∣ξk∣2)21
向量的 ∞ \infty ∞-范数
∥ x ∥ ∞ = max 1 ≤ k ≤ n ∣ ξ k ∣ \|x\|_\infty=\max_{1\leq k \leq n}|\xi_k| ∥x∥∞=max1≤k≤n∣ξk∣
列范数=每列模的和取最大
∥ A ∥ 1 = max 1 ≤ j ≤ n ∑ i = 1 n ∣ a i j ∣ \|A\|_1=\max_{1\leq j \leq n}\sum_{i=1}^n|a_{ij}| ∥A∥1=max1≤j≤n∑i=1n∣aij∣
行范数=每行模的和取最大
∥ A ∥ ∞ = max 1 ≤ i ≤ n ∑ j = 1 n ∣ a i j ∣ \|A\|_\infty=\max_{1\leq i \leq n}\sum_{j=1}^n|a_{ij}| ∥A∥∞=max1≤i≤n∑j=1n∣aij∣
2-范数导出的方阵A的算子范数
∥ A ∥ 2 = max ∥ x ∥ 2 = 1 ∥ A x ∥ 2 = ρ ( A H A ) \|A\|_2=\max_{\|x\|_2=1}\|Ax\|_2=\sqrt{\rho(A^HA)} ∥A∥2=max∥x∥2=1∥Ax∥2=ρ(AHA)
frobenius Form
∥ A ∥ F = ( ∑ i = 1 n ∑ j = 1 n ∣ a i j ∣ 2 ) 1 2 \|A\|_F=(\sum_{i=1}^n\sum_{j=1}^n|a_{ij}|^2)^\frac{1}{2} ∥A∥F=(∑i=1n∑j=1n∣aij∣2)21
谱半径=A中最大的特征值的模
ρ ( A ) = m a x { ∣ λ 1 ∣ , ∣ λ 2 ∣ , . . . , ∣ λ n ∣ } d e t ( λ E − A ) = ∣ λ E − A ∣ \rho(A)=max\{|\lambda_1|,|\lambda_2|,...,|\lambda_n|\} det(\lambda E-A)=|\lambda E-A| ρ(A)=max{∣λ1∣,∣λ2∣,...,∣λn∣}det(λE−A)=∣λE−A∣
正规矩阵 酉对角化
正规矩阵
A H A = A A H A^HA=AA^H AHA=AAH
酉矩阵
A H A = A A H = E A^HA=AA^H=E AHA=AAH=E
正交矩阵,当酉矩阵中
A ∈ R n × n A \in \mathbb{R}^{n \times n} A∈Rn×n
Hermite矩阵
A H = A A^H=A AH=A
实对称矩阵,当Hermite矩阵中
A ∈ R n × n A \in \mathbb{R}^{n \times n} A∈Rn×n
条件数
c
o
n
d
1
A
=
∥
A
−
1
∥
1
∥
A
∥
1
cond_1A=\|A^{-1}\|_1\|A\|_1
cond1A=∥A−1∥1∥A∥1
c
o
n
d
2
A
=
∥
A
−
1
∥
2
∥
A
∥
2
=
max
{
σ
(
A
T
A
)
}
min
{
σ
(
A
T
A
)
}
cond_2A=\|A^{-1}\|_2\|A\|_2=\sqrt{\frac{\max\{\sigma(A^TA)\}}{\min\{\sigma(A^TA)\}}}
cond2A=∥A−1∥2∥A∥2=min{σ(ATA)}max{σ(ATA)}
c
o
n
d
∞
A
=
∥
A
−
1
∥
∞
∥
A
∥
∞
cond_\infty A=\|A^{-1}\|_\infty\|A\|_\infty
cond∞A=∥A−1∥∞∥A∥∞
严格对角占优
∣ a i i ∣ > ∑ j = 1 , j ≠ i n ∣ a i j ∣ |a_{ii}|>\sum_{j=1,j\neq i}^n|a_{ij}| ∣aii∣>∑j=1,j=in∣aij∣
Jacobi迭代,D为A对角线上元素构成的矩阵
M 1 = D − 1 ( D − A ) M_1=D^{-1}(D-A) M1=D−1(D−A)
Gauss-Seidel迭代,D为A对角线上元素构成的矩阵
M 2 = ( D − L ) − 1 U M_2=(D-L)^{-1}U M2=(D−L)−1U
填空题
Lagrange插值多项式
L n ( x ) = ∑ k = 0 n y k l k ( x ) = ∑ k = 0 n ( ∏ i = 0 , i ≠ k n x − x i x k − x i ) y k L_n(x)=\sum_{k=0}^ny_kl_k(x)=\sum_{k=0}^n(\prod_{i=0,i\neq k}^n\frac{x-x_i}{x_k-x_i})y_k Ln(x)=∑k=0nyklk(x)=∑k=0n(∏i=0,i=knxk−xix−xi)yk
Newton插值多项式
N n ( x ) = f [ x 0 ] + ∑ k = 1 n f [ x 0 , x 1 , . . . , x k ] ω k ( x ) N_n(x)=f[x_0]+\sum_{k=1}^nf[x_0,x_1,...,x_k]\omega_k(x) Nn(x)=f[x0]+∑k=1nf[x0,x1,...,xk]ωk(x)
N 3 ( x ) = f ( x 0 ) + f [ x 0 , x 1 ] ( x − x 0 ) + f [ x 0 , x 1 , x 2 ] ( x − x 0 ) ( x − x 1 ) + f [ x 0 , x 1 , x 2 , x 3 ] ( x − x 0 ) ( x − x 1 ) ( x − x 2 ) N_3(x)=f(x_0) +f[x_0,x_1](x-x_0) +f[x_0,x_1,x_2](x-x_0)(x-x_1) +f[x_0,x_1,x_2,x_3](x-x_0)(x-x_1)(x-x_2) N3(x)=f(x0)+f[x0,x1](x−x0)+f[x0,x1,x2](x−x0)(x−x1)+f[x0,x1,x2,x3](x−x0)(x−x1)(x−x2)
迭代法收敛
差商
Hermite插值
一次多项式
S 1 ∗ = a 0 ∗ + a 1 ∗ x S_1^*=a_0^*+a_1^*x S1∗=a0∗+a1∗x
M = s p a n { 1 , x } M=span\{1,x\} M=span{1,x}
φ 0 ( x ) = 1 , φ 1 ( x ) = x \varphi_0(x)=1,\varphi_1(x)=x φ0(x)=1,φ1(x)=x
< φ 0 , φ 0 > = ∑ 1 × 1 <\varphi_0,\varphi_0>=\sum1 \times 1 <φ0,φ0>=∑1×1
< φ 0 , φ 1 > = < φ 1 , φ 0 > = ∑ 1 × x k <\varphi_0,\varphi_1>=<\varphi_1,\varphi_0>=\sum1 \times x_k <φ0,φ1>=<φ1,φ0>=∑1×xk
< φ 1 , φ 1 > = ∑ x k 2 <\varphi_1,\varphi_1>=\sum x_k^2 <φ1,φ1>=∑xk2
< f , φ 0 > = ∑ f ( x k ) <f,\varphi_0>=\sum f(x_k) <f,φ0>=∑f(xk)
< f , φ 1 > = ∑ x k f ( x k ) <f,\varphi_1>=\sum x_k f(x_k) <f,φ1>=∑xkf(xk)
[ < φ 0 , φ 0 > < φ 0 , φ 1 > < φ 1 , φ 0 > < φ 1 , φ 1 > ] [ a 0 a 1 ] = [ < f , φ 0 > < f , φ 1 > ] \begin{bmatrix} <\varphi_0,\varphi_0> &<\varphi_0,\varphi_1>\\ <\varphi_1,\varphi_0> &<\varphi_1,\varphi_1>\\ \end{bmatrix} \begin{bmatrix} a_0\\ a_1\\ \end{bmatrix} =\begin{bmatrix} <f,\varphi_0>\\ <f,\varphi_1>\\ \end{bmatrix} [<φ0,φ0><φ1,φ0><φ0,φ1><φ1,φ1>][a0a1]=[<f,φ0><f,φ1>]
Euler 稳定性
0 < h ≤ − 2 λ 0<h\leq -\frac{2}{\lambda} 0<h≤−λ2
Runge-kutta 稳定性
0 < h ≤ − 2.78 λ 0<h\leq -\frac{2.78}{\lambda} 0<h≤−λ2.78
三次样条
初等变换
Smith标准型
λ E − A \lambda E-A λE−A
Jordan标准型
初等因子组 y-1
有理标准型 自然法式
不变因子 d1(y) 求相伴矩阵
初值问题
求A的最小多项式 m(y)=dn(y)
∣ λ E − A ∣ |\lambda E-A| ∣λE−A∣
验证相乘是否为O
e A t e^{At} eAt
根据特征值的个数与重根,设A的次数
e A t = a 0 ( t ) E + a 1 ( t ) A + a 2 ( t ) A 2 e^{At}=a_0(t)E+a_1(t)A+a_2(t)A^2 eAt=a0(t)E+a1(t)A+a2(t)A2
e λ t e^{\lambda t} eλt
T ( λ t ) = a 0 ( t ) + a 1 ( t ) λ + a 2 ( t ) λ 2 T(\lambda t)=a_0(t)+a_1(t)\lambda+a_2(t)\lambda^2 T(λt)=a0(t)+a1(t)λ+a2(t)λ2
计算x(t),其中C为初值
x ( t ) = e A t C x(t)=e^{At}C x(t)=eAtC
det e A = e t r A \det e^A=e^{trA} deteA=etrA
曲线拟合
最佳平方逼近 Legendre多项式
S 3 ∗ ( x ) = 1 2 < f , p 0 > p 0 ( x ) + 3 2 < f , p 1 > p 1 ( x ) + 5 2 < f , p 2 > p 2 ( x ) + 7 2 < f , p 3 > p 3 ( x ) S_3^*(x)=\frac12<f,p_0>p_0(x)+\frac32<f,p_1>p_1(x)+\frac52<f,p_2>p_2(x)+\frac72<f,p_3>p_3(x) S3∗(x)=21<f,p0>p0(x)+23<f,p1>p1(x)+25<f,p2>p2(x)+27<f,p3>p3(x)
< g , p k > = ∫ − 1 1 g ( x ) ⋅ p k d x <g,p_k>=\int_{-1}^{1}g(x)\cdot p_kdx <g,pk>=∫−11g(x)⋅pkdx
p k = 1 , x , 3 x 2 − 1 2 , 5 x 3 − 3 x 2 p_k=1,x,\frac{3x^2-1}{2},\frac{5x^3-3x}{2} pk=1,x,23x2−1,25x3−3x
δ 2 = b − a 2 [ ∥ g ∥ 2 2 − ∑ k = 0 n 2 k + 1 2 ∣ < g , p k > ∣ 2 ] \delta^2=\frac{b-a}2[\|g\|_2^2-\sum_{k=0}^n\frac{2k+1}2|<g,p_k>|^2] δ2=2b−a[∥g∥22−∑k=0n22k+1∣<g,pk>∣2]
∥ g ∥ 2 2 = ∫ − 1 1 ∣ g ( t ) ∣ 2 d t \|g\|_2^2=\int_{-1}^1|g(t)|^2dt ∥g∥22=∫−11∣g(t)∣2dt
∑ k = 0 n 2 k + 1 2 ∣ < g , p k > ∣ 2 = 1 2 ∣ < g , p 0 > ∣ 2 + 3 2 ∣ < g , p 1 > ∣ 2 + 5 2 ∣ < g , p 2 > ∣ 2 + 7 2 ∣ < g , p 3 > ∣ 2 \sum_{k=0}^n\frac{2k+1}2|<g,p_k>|^2=\frac12|<g,p_0>|^2+\frac32|<g,p_1>|^2+\frac52|<g,p_2>|^2+\frac72|<g,p_3>|^2 ∑k=0n22k+1∣<g,pk>∣2=21∣<g,p0>∣2+23∣<g,p1>∣2+25∣<g,p2>∣2+27∣<g,p3>∣2
分部积分法
∫ a b u ( x ) v ′ ( x ) d x = u ( x ) v ( x ) ∣ a b − ∫ a b u ′ ( x ) v ( x ) d x \int_a^b u(x)v'(x)dx=u(x)v(x)|_a^b-\int_a^bu'(x)v(x)dx ∫abu(x)v′(x)dx=u(x)v(x)∣ab−∫abu′(x)v(x)dx
( u ( x ) v ( x ) ) ′ = u ( x ) v ′ ( x ) + u ′ ( x ) v ( x ) (u(x)v(x))'=u(x)v'(x)+u'(x)v(x) (u(x)v(x))′=u(x)v′(x)+u′(x)v(x)
Romberg算法 初始计算 迭代计算
I = ∫ a b f ( x ) d x I=\int_a^bf(x)dx I=∫abf(x)dx
T 2 0 = b − a 2 [ f ( a ) + f ( b ) ] T_{2^0}=\frac{b-a}2[f(a)+f(b)] T20=2b−a[f(a)+f(b)]
T 2 1 = T 2 0 2 + b − a 2 1 [ f ( a + b 2 ) ] T_{2^1}=\frac{T_{2^0}}2+\frac{b-a}{2^1}[f(\frac{a+b}2)] T21=2T20+21b−a[f(2a+b)]
T 2 2 = T 2 1 2 + b − a 2 2 [ f ( 1 4 ( a + b ) ) + f ( 3 4 ( a + b ) ) ] T_{2^2}=\frac{T_{2^1}}2+\frac{b-a}{2^2}[f(\frac14(a+b))+f(\frac34(a+b))] T22=2T21+22b−a[f(41(a+b))+f(43(a+b))]
T 2 3 = T 2 2 2 + b − a 2 3 [ f ( 1 8 ( a + b ) ) + f ( 3 8 ( a + b ) ) + f ( 5 8 ( a + b ) ) + f ( 7 8 ( a + b ) ) ] T_{2^3}=\frac{T_{2^2}}2+\frac{b-a}{2^3}[f(\frac18(a+b))+f(\frac38(a+b))+f(\frac58(a+b))+f(\frac78(a+b))] T23=2T22+23b−a[f(81(a+b))+f(83(a+b))+f(85(a+b))+f(87(a+b))]
S 2 0 = 4 T 2 1 − T 2 0 4 − 1 , S 2 1 = 4 T 2 2 − T 2 1 4 − 1 , S 2 2 = 4 T 2 3 − T 2 2 4 − 1 S_{2^0}=\frac{4T_{2^1}-T_{2^0}}{4-1}, S_{2^1}=\frac{4T_{2^2}-T_{2^1}}{4-1}, S_{2^2}=\frac{4T_{2^3}-T_{2^2}}{4-1} S20=4−14T21−T20,S21=4−14T22−T21,S22=4−14T23−T22
C 2 0 = 4 2 S 2 1 − S 2 0 4 2 − 1 , C 2 1 = 4 2 S 2 2 − S 2 1 4 2 − 1 C_{2^0}=\frac{4^2S_{2^1}-S_{2^0}}{4^2-1}, C_{2^1}=\frac{4^2S_{2^2}-S_{2^1}}{4^2-1} C20=42−142S21−S20,C21=42−142S22−S21
R 2 0 = 4 3 C 2 1 − C 2 0 4 3 − 1 R_{2^0}=\frac{4^3C_{2^1}-C_{2^0}}{4^3-1} R20=43−143C21−C20
Runge-Kutta法
{ y ′ = z z ′ = ( 1 + x 2 ) y + 1 y ( 0 ) = 1 , z ( 0 ) = 3 \begin{cases} y'=z \\ z'=(1+x^2)y+1 \\ y(0)=1,z(0)=3 \end{cases} ⎩⎪⎨⎪⎧y′=zz′=(1+x2)y+1y(0)=1,z(0)=3
{ y n + 1 = y n + h 6 ( k 1 + 2 k 2 + 2 k 3 + k 4 ) z n + 1 = z n + h 6 ( l 1 + 2 l 2 + 2 l 3 + l 4 ) k 1 = z n , l 1 = ( 1 + x n 2 ) y n + 1 k 2 = z n + h 2 l 1 , l 2 = ( 1 + ( x n + h 2 ) 2 ) ( y n + h 2 k 1 ) + 1 k 3 = z n + h 2 l 2 , l 3 = ( 1 + ( x n + h 2 ) 2 ) ( y n + h 2 k 2 ) + 1 k 4 = z n + h l 3 , l 4 = ( 1 + ( x n + h ) 2 ) ( y n + h k 3 ) + 1 y 0 = 1 , z 0 = 3 ( n = 0 , 1 , 2 , . . . , N − 1 ) \begin{cases} y_{n+1}=y_n+\frac{h}6(k_1+2k_2+2k_3+k_4)\\ z_{n+1}=z_n+\frac{h}6(l_1+2l_2+2l_3+l_4)\\ k_1=z_n,l_1=(1+x_n^2)y_n+1 \\ k_2=z_n+\frac{h}2l_1,l_2=(1+(x_n+\frac{h}2)^2)(y_n+\frac{h}2k_1)+1 \\ k_3=z_n+\frac{h}2l_2,l_3=(1+(x_n+\frac{h}2)^2)(y_n+\frac{h}2k_2)+1 \\ k_4=z_n+hl_3,l_4=(1+(x_n+h)^2)(y_n+hk_3)+1 \\ y_0=1,z_0=3 \end{cases} (n=0,1,2,...,N-1) ⎩⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎧yn+1=yn+6h(k1+2k2+2k3+k4)zn+1=zn+6h(l1+2l2+2l3+l4)k1=zn,l1=(1+xn2)yn+1k2=zn+2hl1,l2=(1+(xn+2h)2)(yn+2hk1)+1k3=zn+2hl2,l3=(1+(xn+2h)2)(yn+2hk2)+1k4=zn+hl3,l4=(1+(xn+h)2)(yn+hk3)+1y0=1,z0=3(n=0,1,2,...,N−1)
证明题
参考文献
工程数学基础教程