Home » Periodic Functions

Tag: Periodic Functions

Prove some properties of even, periodic, integrable functions

Let f be an integrable, even, periodic function with period 2. Define

    \[ g(x) = \int_0^x f(t) \, dt, \qquad \text{and} \qquad A = g(1). \]

  1. Prove g is odd and g(x+2) - g(x) = g(2).
  2. Compute g(2) and g(5) using that A = g(1).
  3. Find the value of A that makes g periodic with period 2.

  1. Proof. Using the expansion/contraction property (Apostol, Thm 1.19 with k = -1) we compute,

        \begin{align*}  g(-x) &= \int_0^{-x} f(t) \, dt \\  & = - \int_0^x f(-t) \, dt \\  & = - \int_0^x f(t) \, dt & (f \text{ is even}) \\  & = - g(x). \end{align*}

    Thus, g is odd. Next,

        \begin{align*}  g(x+2) - g(x) &= \int_0^{x+2} f(t) \, dt - \int_0^x f(t) \, dt \\  &= \int_0^2 f(t) \,dt + \int_2^{x+2} f(t) \, dt - \int_0^x f(t) \, dt \\  &= g(2) + \int_0^x f(t+2) \, dt - \int_0^x f(t) \, dt \\  & = g(2) + \int_0^x f(t) \, dt - \int_0^x f(t) \, dt & (f \text{ periodic}) \\  & = g(2). \qquad \blacksquare \end{align*}

  2. First, we compute g(2),

        \begin{align*}  g(2) &= g(3) - g(1)  &(\text{part (a) with } x = 1) \\  &= \int_0^3 f(t) \, dt - \int_0^1 f(t) \, dt \\  &= \int_1^3 f(t) \, dt \\  & = \int_{-1}^1 f(t+2) \, dt &(\text{Translation prop, Thm 1.18})\\  & = 2 \int_0^1 f(t) \, dt \\  & = 2A, \end{align*}

    where the second to last line followed from the fact that f is even and this exercise, so \int_{-b}^b f(t)\, dt = 2 \int_0^b f(t) \, dt.

    Since g(1) = A by definition, the above also gives us g(3) = g(2) + g(1) = 3A. So, again using part (a) (this time with x = 3) we have

        \begin{align*}  g(5) &= g(2) + g(3) \\  &= 2A + 3A \\  &= 5A. \end{align*}

  3. Here we want to find a value of A such that g is periodic with period 2. For g to be periodic with period 2, we must have

        \[ g(x+2) = g(x) \quad \implies \quad g(x+2) - g(x) = 0. \]

    But, since g(0) = \int_0^0 f(t) \, dt = 0, and the above must be true for all x, we evaluate at x=0 and use part (b) that gave us g(2) = 2A to compute,

        \[ g(2) = 0 \quad \implies \quad 2A = 0 \quad \implies \quad A = 0. \]

Prove some formulas for odd periodic functions

Let f be an odd, periodic, integrable everywhere function with period 2. Then, define

    \[ g(x) = \int_0^x f(t) \, dt. \]

  1. Prove that g(2n) = 0 for all n \in \mathbb{Z}.
  2. Prove that g is an even, periodic function with period 2.

  1. Proof. We can establish this by a direct computation.

        \begin{align*}  g(2n) &= \int_0^{2n} f(t) \, dt \\  &= \int_{-2n}^0 f(t+2n) \, dt & (\text{translation property, Thm 1.18})\\  &= \int_{-2n}^0 f(t) \, dt & (f \text{ periodic, period 2}) \\  &= - \int_0^{-2n} f(t) \, dt \\  &=  \int_0^{2n} f(-t) \, dt & (\text{Expansion/contraction property}) \\  &= - \int_0^{2n} f(t) \, dt & (f \text{ is odd so } f(-t) = -f(t)) \end{align*}

    But then we have

        \[ \int_0^{2n} f(t) \, dt = - \int_0^{2n} f(t) \, dt \quad \implies \quad \int_0^{2n} f(t) \, dt = 0 \]

    Since g(2n) = \int_0^{2n} f(t) \, dt we then have the requested result,

        \[ g(2n) = \int_0^{2n} f(t) \, dt = 0. \qquad blacksquare \]

  2. First, to show that g is even we need to show g(-x) = g(x) for all x. We compute, using the fact that f is odd and the expansion/contraction property of the integral (Theorem 1.19 in Apostol) with k=-1.

        \begin{align*}  g(-x) &= \int_0^{-x} f(t) \, dt \\  &= - \int_0^x f(-t) \, dt & (\text{Exp/Cont property}) \\  &= \int_0^x f(t) \, dt & (f \text{ is odd}) \\  &= g(x) & (\text{Def of } g). \end{align*}

    Next, to show that g is periodic with period 2 we must show g(x+2) = g(x) for all x.

        \begin{align*}  g(x+2) &= \int_0^{x+2} f(t) \, dt \\  &= \int_0^2 f(t) \, dt + \int_2^{x+2} f(t) \, dt \\  &= 0 + \int_0^x f(t+2) \, dt &(\text{part (a) and translation property}) \\  &= \int_0^x f(t) \, dt & (f \text{ is periodic, period 2}) \\  & = g(x). \qquad \blacksquare \end{align*}

Prove an integral formula for periodic functions

Let f be an integralable function on [0,p], and let f be periodic with period p > 0. Prove

    \[ \int_0^p f(x) \, dx = \int_a^{a+p} f(x) \, dx \]

for all a \in \mathbb{R}.


Proof. For any a, p \in \mathbb{R} with p > 0 we know from a previous exercise that there exists a unique n \in \mathbb{Z} such that

    \[ n \leq \frac{a}{p} < n + 1 \quad \implies \quad np \leq a < (n+1)p \leq a + p. \]

Then, we start by splitting the integral into two pieces. (The goal here is to rearrange the integral so that it starts at np and ends at (n+1)p, and then use the fact that f is periodic to conclude that this integral is the same as the one from 0 to p.)

    \begin{align*}  \int_a^{a+p} f(x) \, dx &= \int_a^{(n+1)p} f(x) \, dx + \int_{(n+1)p}^{a+p} f(x) \, dx \\  &= \int_{a-np}^{(n+1)p - np} f(x+np) \, dx + \int_{(n+1)p - (n+1)p}^{a+p-(n+1)p} f(x+(n+1)p) \, dx \\  &= \int_{a-np}^p f(x) \, dx + \int_0^{a - np} f(x) \, dx & (\text{using periodicity})\\  &= \int_0^p f(x) \, dx & (\text{since } 0 \leq a-np < p ). \end{align*}

This completes the proof. \qquad \blacksquare