Romberg Integration

The Romberg integral converges as n increases.
The number of subintervals is calculated by doubling from 2, 4, 8, 16, ..., N=2ⁿ, doubling the number of partitions each time.
If the endpoints are not analytic, the value is calculated using the midpoint rule. Example: sin(x)/x, [0.π/2]
For periodic functions, the values are determined using a nonlinear transformation. Examples: (cos(64x))^2, [0, π]

\(\normalsize Romberg\ integration\\ (1)\ Trapezoidal\ rule\\ \hspace{10px} R_{\small 0}^{\small\ 0}= {\large\frac{h_0}{2}}\{f(a)+f(b)\}, \hspace{20px}h_0=b-a\\ \hspace{8px} R_{\small n}^{\small\ 0}= {\large\frac{R_{\small n-1}^{\small\ 0}}{2}}+h_n{\large \displaystyle \sum_{\tiny j=1}^{\small 2^{n-1}}}f(a+(2j-1)h_n)\\ \hspace{190px}h_n={\large\frac{b-a}{2^n}}\\ (2)\ Midpoint\ rule\\ \hspace{10px} R_{\small 0}^{\small\ 0}= h_0f(a+{\large\frac{h_0}{2}}),\hspace{20px}h_0=b-a\\ \hspace{10px} R_{\small n}^{\small\ 0}= h_n{\large \displaystyle \sum_{\tiny j=1}^{\small 2^n}}f(a+(j-\frac{1}{2})h_n), \hspace{15px}h_n={\large\frac{b-a}{2^n}}\\ (3)\ R_{\small n}^{\small\ k}= {\large\frac{4^{\small k} R_{\small n}^{\small\ k-1}-R_{\small n-1}^{\small\ k-1}}{4^{\small k}-1}}\\ (4)\ Relative\ Error\hspace{30px} \epsilon_{\small n}=\left|\frac{R_{\small n}^{\small\ n}-R_{\small n-1}^{\small\ n-1}}{R_{\small n}^{\small\ n}}\right|\\ Non-linear\ substitution\ in\ x\\ \hspace{25px} I={\large\int_a^{b}}f(x)dx\ ={\large\int_{\small-1}^{\small1}}f(x)\ \frac{b-a}{2}\ \frac{3(1-u^2)}{2}du\\ \hspace{100px}x=\frac{b-a}{2}t+\frac{b+a}{2},\hspace{20px}t=\frac{u}{2}(3-u^2)\)