Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I am studying numerical approximation and verifying $S_{2n} = \frac{1}{3}\left(T_n +2 M_n\right)$. ($S_n$ refers to Simpson's Rule approximation, $T_n$ refers to Trapezoid Rule approximation and $M_n$ refers to Midpoint Rule approximation) I listed all the formulas and manipulated the equations, but I have no idea how $2n$ numbers come out at the left side while only $n$ numbers at the right side.

share|cite|improve this question
Perhaps you should try a very simple example, like $x^2$ from 0 to 4 or something (supposing that you know Simpson's formula aside from this relationship too) – mixedmath Feb 26 '12 at 18:35
@mixedmath Yes I know and I also did some exercises about those rules. What I want is a proof of this equation. – Shannon Feb 26 '12 at 20:22
up vote 4 down vote accepted

For the Trapezoidal Rule, you actually use $n+1$ points. For example, in the simple case where you are integrating $f(x)$ from $0$ to $1$, and you want $T_4$, you evaluate $f$ at the points $0/4$, $1/4$, $2/4$, $3/4$, and $4/4$. It is $n+1$ points because we use the endpoints.

For the Midpoint Rule, you use $n$ points, but these are not the same points as for the Trapezoidal Rule. They are the midpoints of our intervals. So in the example discussed above, for $M_4$ you would be evaluating $f$ at $1/8$, $3/8$, $5/8$, and $7/8$.

The Simpson Rule $S_{2n}$ uses evaluation of $f$ at $2n+1$ points. If for example $n=4$, then you are dividing the interval into $8$ subintervals. With $n=4$ and the interval $[0,1]$, you would be using the points $0/8$, $1/8$, $2/8$, $3/8$, $4/8$, $5/8$, $6/8$, $7/8$, and $8/8$.

Note that $1/8$, $3/8$, $5/8$ and $7/8$ are the points that were used for $M_4$. The points $0/8$, $2/8$, $4/8$, $6/8$, and $8/8$ are just $0/4$, $1/4$, $2/4$, $3/4$, and $4/4$, exactly the points that were used for $T_4$.

A more abstract summary: $T_n$ uses $n+1$ points, and $M_n$ uses $n$ points. But the $n$ points used by $M_n$ are completely different from the points used for $T_n$. So altogether, $T_n$ and $M_n$ carry information about function evaluation at $2n+1$ points, which is exactly what $S_{2n}$ does.

I have not written out a proof of the formula, only tried to deal with your discomfort with the $2n$ on one side and $n$'s on the other. The formula is not hard to verify. Let's do it explicitly for $n=4$. Write down, say for the interval $[0,1]$, what $T_4$ is. We have $$T_4=\frac{1}{8}(f(0)+2f(1/4)+2f(1/2)+2f(3/4)+f(1)).$$ Now write down $M_4$: $$M_4=\frac{1}{4}(f(1/8)+f(3/8)+f(5/8)+f(7/8)).$$ Now calculate $T_4+2M_4$. It is convenient for the addition to make sure that $M_4$ has denominator $8$, so write $2M_4$ as $\frac{1}{8}(4f(1/8)+4f(3/8)+4f(5/8)+4f(7/8))$, and add. Divide by $3$ and you will get the expression you would get in $S_{8}$. The same method works in general.

share|cite|improve this answer
I see. Thank you so much. – Shannon Feb 27 '12 at 1:55

For given $n\in{\mathbb N}$ each of $S_{2n}$, $T_n$ and $M_n$ are one number, and each of these three numbers claim to be an approximation to the number $$\int_a^b f(x)\ dx\ .$$ They are arrived at as follows: The interval $[a,b]$ is divided into $n$ equal parts $[x_{n-1},x_n]$ of length ${b-a\over n}$. When $f$ is interpolated by a piecewise linear function $g$ such that $g(x_i)=f(x_i)$ $\ (0\leq i\leq n)$ then $$T_n:=\int_a^b g(x)\ dx={b-a\over n}\bigl({1\over2}(f(a)+f(b))+\sum_{k=1}^{n-1} f(x_k)\bigr)\ .$$ For $M_n$ the function $f$ is replaced by a function $h$ which is constant on each subinterval $[x_{k-1},x_k]$, and the value of $h$ on this interval is defined to be $f\Bigl({x_{k-1}+x_k\over2}\Bigr)$. This leads to $$M_n=\int_a^b h(x)\ dx={b-a\over n}\sum_{k=1}^nf\Bigl({x_{k-1}+x_k\over2}\Bigr)\ .$$ Note that the $n$ places where $f(x)$ has to be computed to obtain $M_n$ are different from the $n+1$ places where $f(x)$ has to be computed to obtain $T_n$.

Finally, the Simpson value $S_{2n}$ is defined as $$S_{2n}:={1\over3}(T_n+2M_n)\ .$$ For $S_{2n}$ you have to evaluate $f$ at $2n+1$ different $x$'s; in return you get a much better approximation than with $M_n$ and $T_n$. One arrives at this formula by replacing $f$ in each subinterval $[x_{k-1},x_k]$ by a quadratic function having the same value as $f$ in the points $x_{k-1}$, ${x_{k-1}+x_k\over2}$ and $x_k$.

share|cite|improve this answer
Should f(b)in your Trapezoild equation be f(b)/2? – Shannon Feb 26 '12 at 20:24
@Shannon: $f(a)$ and $f(b)$ have equal weight ${1\over 2}\cdot{b-a\over n}$, as stated. – Christian Blatter Feb 26 '12 at 21:07

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.