Given the quadrature rule $Q[f]=f(x_1)+f(x_2) \approx \int_{-1}^{1}f(x)dx = I[f]$ , find the nodes $x_1, x_2$

45 Views Asked by At

Given is the following quadrature rule : $$Q[f]=f(x_1)+f(x_2) \approx \int_{-1}^{1}f(x)dx = I[f]$$ and the degree of accuracy $q \geq2$. We're asked to find the nodes $x_1, x_2$.

May I ask you to give me feedback on my attempt ?


We evaluate $I[x^0]=2, I[x^1]=0, I[x^2]= \frac{2}{3}$. Thus: $$x^0: f(x_1)+f(x_2)= 1 + 1 \overset{!}{=} 2$$ $$x^1: f(x_1)+f(x_2)= x_1 + x_2 \overset{!}{=} 0$$ $$x^2: f(x_1)+f(x_2)= x_1^2 + x_2^2 \overset{!}{=} \frac{2}{3}$$

Thus: $x_2 = - x_1 \rightarrow x_1 = \pm \frac{1}{\sqrt{3}} \rightarrow x_2 = \pm \frac{1}{\sqrt{3}}$


Is what I did correct ? Did I forget something ? Thanks for your feedback !

1

There are 1 best solutions below

2
On BEST ANSWER

What you have done is fine, but for higher orders you might want to systematize the process via Newton's identities. In that case you have the sums of powers $$p_k=\sum_{i=1}^nx_i^k$$ And the elementary symmetric polynomials $e_k$ which are the sums of all the distinct products of the $n$ variables taken $k$ at a time. Thus $$\begin{align}e_0&=1\\ e_1&=\sum_{i=1}^nx_i\\ e_2&=\sum_{i=1}^{n-1}\sum_{j=i+1}^nx_ix_j\\ e_3&=\sum_{i=1}^{n-2}\sum_{j=i+1}^{n-1}\sum_{k=j+1}^nx_ix_jx_k\end{align}$$ And so on. Newton's identities read, for $n\ge k\ge1$, $$ke_k=\sum_{i=1}^k(-1)^{i-1}e_{k-i}p_i$$ Thus in your case they read $p_0=2$, $p_1=0$, $p_2=\frac23$, $e_0=1$, $1e_1=e_1=e_0p_1=(1)(0)=0$, and $2e_2=e_1p_1-e_0p_2=(0)(0)-(1)\left(\frac23\right)=-\frac23$. Then the equation that $x_1$ and $x_2$ must satisfy is $$\sum_{i=0}^n(-1)^ie_ix^{n-i}=x^2-(0)x+\left(-\frac13\right)=x^2-\frac13=0$$ So could you apply this procedure to find the corresponding Chebyshev quadrature rule with equal weights and $3$ nodes?