In Rudin RCA theorem $2.20 $ Page no:$51$

Rudin say that if $n >N$, then we have $$\Lambda_N g =\Lambda_n g \leq \Lambda_n f \leq \Lambda_n h = \Lambda_N h$$

Here I'm confused that why $\Lambda_N g =\Lambda_n g $?

My attempt:

It is given that $g$ is constant on each box in $\{\Omega_n\}$

Now if $n > N$ , then $$\Lambda_n g := 2^{-nk} \sum\limits_{x \in P_n} g(x)$$

Now using the Property : For $\{\Omega_n\}$, if $Q\in \Omega_r$, then vol$(Q)=2^{-rk}$; and if $n>r$, the set $P_n$ has exactly $2^{(n-r)k}$ points in $Q$

we have $$\Lambda_n g := 2^{-nk} g(x)\sum\limits_{x \in P_n}.1=2^{-nk} g(x)2^{(n-N)k}=2^{-Nk}g(x)$$

$$\implies 2^{-Nk}g(x) \neq \Lambda_N g := 2^{-Nk} \sum\limits_{x \in P_n} g(x)$$

Therefore $$\Lambda_N g \neq \Lambda_n g$$


2 Answers 2


The issue with your approach is that you’ve factored $g(x)$ out of a sum which is indexed by $x$, but $g$ is not constant on all of $P_n$. However, $g$ is compactly supported and constant on every $Q \in \Omega_N$. This means that:

  1. The sums have finitely many nonzero terms, so you can group terms however you like; and,
  2. For every $Q \in \Omega_N$, the $2^{(n - N) k}$ points of $P_n$ which lie inside $Q$ all have the same value under $g$, and we can call it $c(Q)$.

Hence, the sum may be regrouped as \begin{align*} 2^{-nk} \sum_{x\in P_n} g(x) &= 2^{-nk} \sum_{Q\in\Omega_N} \sum_{x\in Q\cap P_n} g(x) \\ &= 2^{-nk} \sum_{Q\in\Omega_N} c(Q) \sum_{x\in Q\cap P_n} 1 \\ &= 2^{-nk} \sum_{Q\in\Omega_N} c(Q) \cdot 2^{(n - N)k} \\ &= 2^{-Nk} \sum_{Q\in\Omega_N} c(Q) \\ &= 2^{-Nk} \sum_{x\in P_N} g(x), \end{align*} where the first equality is justified as grouping together all the $x$’s which lie in the same $Q$; and the last equality is justified as each $Q \in \Omega_N$ can be identified by its bottom-left corner $x \in P_N$, and $c(Q) = g(x)$ in that case.

  • $\begingroup$ thanks u @Shoteyes Can you recommend a good book for measure theory ? Im weak in measure theory $\endgroup$
    – jasmine
    Jun 2, 2021 at 17:06
  • 1
    $\begingroup$ Honestly, I like Rudin’s RCA even though it’s not strictly a measure theory book, since it covers so much of real and complex analysis. The only qualm I have with it is that some things lack motivation, but drawing pictures yourself can help with that. $\endgroup$
    – shoteyes
    Jun 2, 2021 at 17:10

Note : $x \in P_N $ not $P_n$, so $P_n$ has exactly $2^{(N-N)k}$ points in $Q$. This implies

$$\Lambda_N g = 2^{-Nk} \sum\limits_{x \in P_N} g(x) =2^{-Nk} g(x) 2^{(N-N)k}=2^{-Nk}g(x)=\Lambda_n g$$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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