Prove that the lower Riemann integral of $f$ on $[a, b]$ is $\sup\{\int_a^b\psi: \psi \text{ is a step function on } [a, b],\text{ such that } \psi\leq f \text{ on }[a, b]\}$.

The definitions that I am working with are:

For a partition $P = \{x_0, ..., x_n\}$ of $[a,b]$ with $a = x_0 < x_1 < ... < x_n = b$. Then $\int_a^b\psi = \sum_{j=1}^{n}\psi_{j}(x_j - x_{j-1})$ where $\psi_i$ is the constant value of $\psi$ on the interval $\psi$ on $[x_{j-1}, x_j]$.

The Riemann integral of $f$ is defined to be (using the partition $P$ above) $\int_a^b f = L(f,[a,b]) = U(f,[a,b])$ where $L(f,[a,b]) = \sum_{j=1}^{n}(x_j-x_{j-1})\inf_{[x_{j-1},x_j]}f~~, U(f,[a,b]) = \sum_{j=1}^{n}(x_j-x_{j-1})\sup_{[x_{j-1},x_j]}f ~$are the lower and upper Riemann sums respectively.

I was trying to show that the supremum over the step functions has to be equal to the lower sum/ the constant value $\psi_j = \inf f$ on any of our sub intervals, but going from here to proving that the sums are equal I was running to to troubles keeping my epsilons straight.

Basically, I wanted to show that for our step functions we need that the step function value needs to be equal to the $\inf f$ on any of the sub intervals and from there we'd hopefully get the desired equality.

Any help on if this is the correct direction to go/ what a clean proof looks like is appreciated. Thanks =)


You want to show two things:

  1. For every step function $\psi$ with $\psi \le f$ we have $\int \psi \le \underline{\int} f$. And to do this, simply define a partition so that $\psi$ is constant on each interval $(x_i, x_{i+1})$ (we don't need to worry about the endpoints of the intervals but do check this).

  2. For every $\varepsilon$, we can find a step function $\psi$ such that $\underline{\int} f \le \int \psi + \varepsilon$. And to do this, you can do exactly as you say: define $\psi_i = \inf_{[x_i, x_{i+1}]} f(x)$ and define $\psi$ piecewise using the constants $\psi_i$.

For 1. you want to check that $\int \psi \le L(f, \mathcal{P})$ for any sufficiently small partition (refining the partition given by $\psi$). It follows then that $\int \psi \le \sup_{\mathcal{P}} L(f, \mathcal{P}) = \underline{\int} f$.

For 2. you can check that $\int \psi = L(f, \mathcal{P})$ by definition and by definition of $\underline{\int} f = \sup_{\mathcal P} L(f, \mathcal{P})$ there exists a partition $\mathcal{P}$ such that $\underline{\int} f \le L(f, \mathcal{P}) + \varepsilon$ (that's the definition of $\sup$).

This is a common strategy: if you want to show $A = B$ you can show that $A \le B$ and $B \le A + \varepsilon$ for every $\varepsilon > 0$.


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.