Numrical Analysis Summary

Evaluating a polynomial

Direct form:O($n^2$)

$P(x)=a_0+a_{1}x+a_2{x}^2+...+a_n{x}^n$

Horner’s method:O(n)

$P(x)=a_0+x(a_1+x(a_2+x\ast (...(a_{n-1}+a_{n}x)))$

Example:求二进制幂O(n)->O($lo{g}_{2}n$)

$a^n=a^{p(2)}$$=a^{b_m{2}^m+...+b_i{2}^i+b_0}$
$a^{2p+b_i}=a^{2p}\ast a$ or $a^{2p}$

Errors

Assume that r is the root of function f(x)=0.
Forward Error:$|r-x_a|$
Backward Error:$f(x_a)$

Solving Equations

Newton’s Method
$x_0=initialguess$
$x_{i+1}=x_i-\frac{f(x_i)}{f^{\prime }(x_i)}$

NaN:not a number

Convex Optimization

Convex combination: ( x , y ) → { α x + β y ∣ α + β = 1 , α , β ≥ 0 } (x,y)\rightarrow\{\alpha x + \beta y|\alpha + \beta =1, \alpha,\beta \geq 0\} (x,y)→{αx+βy∣α+β=1,α,β≥0}

Convex set S: $(x1,...,xk)\in S$, then convex combination of the points x1, . . . , xk should $\in$ S.

A function f : $R^n\to R$ is convex if $\forall x,y$ and $\forall \alpha \in [0,1]$

$$f(\alpha x+(1-\alpha )y)\le \alpha f(x)+(1-\alpha )f(y)$$