Exercise 4.3Z: Exponential and Laplace Distribution

From LNTwww

Exponential PDF (above) d
Laplace PDF (below)

We consider here the probability density functions  $\rm (PDF)$  of two continuous random variables:

  • The random variable   $X$  is exponentially distributed (see top plot):   For  $x<0$    ⇒   $f_X(x) = 0$,  and for positive $x$–values:
$$f_X(x) = \lambda \cdot {\rm e}^{-\lambda \hspace{0.05cm}\cdot \hspace{0.05cm}x}\hspace{0.05cm}. $$
  • On the other hand, for the Laplace distributed random variable  $Y$  in the whole range  $ - \infty < y < + \infty$  holds (lower sketch):
$$f_Y(y) = \lambda/2 \cdot {\rm e}^{-\lambda \hspace{0.05cm}\cdot \hspace{0.05cm}|\hspace{0.05cm}y\hspace{0.05cm}|}\hspace{0.05cm}.$$

To be calculated are the differential entropies  $h(X)$  and  $h(Y)$  depending on the PDF parameter  $\it \lambda$.  For example:

$$h(X) = -\hspace{-0.7cm} \int\limits_{x \hspace{0.05cm}\in \hspace{0.05cm}{\rm supp} \hspace{0.03cm}(\hspace{-0.03cm}f_X)} \hspace{-0.55cm} f_X(x) \cdot {\rm log} \hspace{0.1cm} \big [f_X(x) \big ] \hspace{0.1cm}{\rm d}x \hspace{0.05cm}.$$

If  $\log_2$  is used, add the pseudo-unit  "bit".


In subtasks  (2)  and  (4)  specify the differential entropy in the following form:

$$h(X) = {1}/{2} \cdot {\rm log} \hspace{0.1cm} ({\it \Gamma}_{{\hspace{-0.01cm} \rm L}}^{\hspace{0.08cm}(X)} \cdot \sigma^2), \hspace{0.8cm}h(Y) = {1}/{2} \cdot {\rm log} \hspace{0.1cm} ({\it \Gamma}_{{\hspace{-0.05cm} \rm L}}^{\hspace{0.08cm}(Y)} \cdot \sigma^2) \hspace{0.05cm}.$$

Determine by which factor  ${\it \Gamma}_{{\hspace{-0.05cm} \rm L}}^{\hspace{0.08cm}(X)}$  the exponential PDF is characterized and which factor  ${\it \Gamma}_{{\hspace{-0.01cm} \rm L}}^{\hspace{0.08cm}(Y)}$  results for the Laplace PDF.





Hints:

  • The exercise belongs to the chapter  Differential Entropy.
  • Useful hints for solving this task can be found in particular on the page  Differential entropy of some power-constrained random variables.
  • For the variance of the exponentially distributed random variable  $X$  holds, as derived in  Exercise 4.1Z:   $\sigma^2 = 1/\lambda^2$.
  • The variance of the Laplace distributed random variable  $Y$  is twice as large for the same  $\it \lambda$:   $\sigma^2 = 2/\lambda^2$.


Questions

1

Calculate the differential entropy of the exponential distribution for  $\lambda = 1$.

$h(X) \ = \ $

$\ \rm bit$

2

What is the characteristic   ${\it \Gamma}_{{\hspace{-0.01cm} \rm L}}^{\hspace{0.08cm}(X)}$  for the exponential distribution corresponding to the form  $h(X) = {1}/{2} \cdot {\rm log}_2 \hspace{0.1cm} ({\it \Gamma}_{\hspace{-0.05cm} \rm L}^{\hspace{0.08cm}(X)} \cdot \sigma^2)$ ?

${\it \Gamma}_{{\hspace{-0.01cm} \rm L}}^{\hspace{0.08cm}(X)} \ = \ $

3

Calculate the differential entropy of the Laplace distribution for  $\lambda = 1$.

$h(Y) \ = \ $

$\ \rm bit$

4

What is the characteristic  ${\it \Gamma}_{{\hspace{-0.01cm} \rm L}}^{\hspace{0.08cm}(Y)} $  for the Laplace distribution corresponding to the form  $h(Y) = {1}/{2} \cdot {\rm log}_2 \hspace{0.1cm} ({\it \Gamma}_{\hspace{-0.05cm} \rm L}^{\hspace{0.08cm}(Y)} \cdot \sigma^2)$?

${\it \Gamma}_{{\hspace{-0.01cm} \rm L}}^{\hspace{0.08cm}(Y)} \ = \ $


Solution

(1)  Although in this exercise the result should be given in  "bit",  we use the natural logarithm for derivation.

  • Then the differential entropy is:
$$h(X) = -\hspace{-0.7cm} \int\limits_{x \hspace{0.05cm}\in \hspace{0.05cm}{\rm supp} \hspace{0.03cm}(\hspace{-0.03cm}f_X)} \hspace{-0.35cm} f_X(x) \cdot {\rm ln} \hspace{0.1cm} \big [f_X(x)\big] \hspace{0.1cm}{\rm d}x \hspace{0.05cm}.$$
  • For the exponential distribution,  the integration limits are  $0$  and  $+∞$.  In this range, the PDF  $f_X(x)$  according to the specification sheet is used:
$$h(X) =- \int_{0}^{\infty} \hspace{-0.15cm} \lambda \cdot {\rm e}^{-\lambda \hspace{0.05cm}\cdot \hspace{0.05cm}x} \hspace{0.05cm} \cdot \hspace{0.05cm} \left [ {\rm ln} \hspace{0.1cm} (\lambda) + {\rm ln} \hspace{0.1cm} ({\rm e}^{-\lambda \hspace{0.05cm}\cdot \hspace{0.05cm}x})\right ]\hspace{0.1cm}{\rm d}x - \hspace{0.05cm} {\rm ln} \hspace{0.1cm} (\lambda) \cdot \int_{0}^{\infty} \hspace{-0.15cm} \lambda \cdot {\rm e}^{-\lambda \hspace{0.05cm}\cdot \hspace{0.05cm}x}\hspace{0.1cm}{\rm d}x \hspace{0.1cm} + \hspace{0.1cm} \lambda \cdot \int_{0}^{\infty} \hspace{-0.15cm} \lambda \cdot x \cdot {\rm e}^{-\lambda \hspace{0.05cm}\cdot \hspace{0.05cm}x}\hspace{0.1cm}{\rm d}x \hspace{0.05cm}.$$

We can see:

  • The first integrand is identical to the PDF  $f_X(x)$ considered here.  Thus, the integral over the entire integration domain yields  $1$.
  • The second integral corresponds exactly to the definition of the mean value  $m_1$  (moment of first order).  For the exponential PDF,  $m_1 = 1/λ$ holds.  From this follows:
$$h(X) = - \hspace{0.05cm} {\rm ln} \hspace{0.1cm} (\lambda) + 1 = - \hspace{0.05cm} {\rm ln} \hspace{0.1cm} (\lambda) + \hspace{0.05cm} {\rm ln} \hspace{0.1cm} ({\rm e}) = {\rm ln} \hspace{0.1cm} ({\rm e}/\lambda) \hspace{0.05cm}.$$
  • This result is to be given the additional unit  "nat".  Using  $\log_2$  instead of  $\ln$,  we obtain the differential entropy in  "bit":
$$h(X) = {\rm log}_2 \hspace{0.1cm} ({\rm e}/\lambda) \hspace{0.3cm} \Rightarrow \hspace{0.3cm} \lambda = 1{\rm :} \hspace{0.3cm} h(X) = {\rm log}_2 \hspace{0.1cm} ({\rm e}) = \frac{{\rm ln} \hspace{0.1cm} ({\rm e})}{{\rm ln} \hspace{0.1cm} (2)} \hspace{0.15cm}\underline{= 1.443\,{\rm bit}} \hspace{0.05cm}.$$


(2)  Considering the equation  $\sigma^2 = 1/\lambda^2$  valid for the exponential distribution, we can transform the result found in  (1)  as follows:

$$h(X) = {\rm log}_2 \hspace{0.1cm} ({\rm e}/\lambda) = {1}/{2}\cdot {\rm log}_2 \hspace{0.1cm} ({\rm e}^2/\lambda^2) = {1}/{2} \cdot {\rm log}_2 \hspace{0.1cm} ({\rm e}^2 \cdot \sigma^2) \hspace{0.05cm}.$$
  • A comparison with the required basic form  $h(X) = {1}/{2} \cdot {\rm log}_2 \hspace{0.1cm} ({\it \Gamma}_{\hspace{-0.05cm} \rm L}^{\hspace{0.08cm}(X)} \cdot \sigma^2)$  leads to the result:
$${\it \Gamma}_{{\hspace{-0.05cm} \rm L}}^{\hspace{0.08cm}(X)} = {\rm e}^2 \hspace{0.15cm}\underline{\approx 7.39} \hspace{0.05cm}.$$


(3)  For the Laplace distribution, we divide the integration domain into two subdomains:

  • $Y$  negative   ⇒   proportion  $h_{\rm neg}(Y)$,
  • $Y$  positive   ⇒   proportion  $h_{\rm pos}(Y)$.


The total differential entropy, taking into account  $h_{\rm neg}(Y) = h_{\rm pos}(Y)$  is given by

$$h(Y) = h_{\rm neg}(Y) + h_{\rm pos}(Y) = 2 \cdot h_{\rm pos}(Y) $$
$$\Rightarrow \hspace{0.3cm} h(Y) = - 2 \cdot \int_{0}^{\infty} \hspace{-0.15cm} \lambda/2 \cdot {\rm e}^{-\lambda \hspace{0.05cm}\cdot \hspace{0.05cm}y} \hspace{0.05cm} \cdot \hspace{0.05cm} \left [ {\rm ln} \hspace{0.1cm} (\lambda/2) + {\rm ln} \hspace{0.1cm} ({\rm e}^{-\lambda \hspace{0.05cm}\cdot \hspace{0.05cm}y})\right ]\hspace{0.1cm}{\rm d}y = - \hspace{0.05cm} {\rm ln} \hspace{0.1cm} (\lambda/2) \cdot \int_{0}^{\infty} \hspace{-0.15cm} \lambda \cdot {\rm e}^{-\lambda \hspace{0.05cm}\cdot \hspace{0.05cm}y}\hspace{0.1cm}{\rm d}y \hspace{0.1cm} + \hspace{0.1cm} \lambda \cdot \int_{0}^{\infty} \hspace{-0.15cm} \lambda \cdot y \cdot {\rm e}^{-\lambda \hspace{0.05cm}\cdot \hspace{0.05cm}y}\hspace{0.1cm}{\rm d}y \hspace{0.05cm}.$$

If we again consider that the first integral gives the value  $1$   (PDF area) and the second integral gives the mean value  $m_1 = 1/\lambda$  we obtain:

$$h(Y) = - \hspace{0.05cm} {\rm ln} \hspace{0.1cm} (\lambda/2) + 1 = - \hspace{0.05cm} {\rm ln} \hspace{0.1cm} (\lambda/2) + \hspace{0.05cm} {\rm ln} \hspace{0.1cm} ({\rm e}) = {\rm ln} \hspace{0.1cm} (2{\rm e}/\lambda) \hspace{0.05cm}.$$
  • Since the result is required in  "bit",  we still need to replace  "$\ln$"  by  "$\log_2$":
$$h(Y) = {\rm log}_2 \hspace{0.1cm} (2{\rm e}/\lambda) \hspace{0.3cm} \Rightarrow \hspace{0.3cm} \lambda = 1{\rm :} \hspace{0.3cm} h(Y) = {\rm log}_2 \hspace{0.1cm} (2{\rm e}) \hspace{0.15cm}\underline{= 2.443\,{\rm bit}} \hspace{0.05cm}.$$


(4)  For the Laplace distribution, the relation  $\sigma^2 = 2/\lambda^2$ holds.  Thus, we obtain:

$$h(X) = {\rm log}_2 \hspace{0.1cm} (\frac{2{\rm e}}{\lambda}) = {1}/{2} \cdot {\rm log}_2 \hspace{0.1cm} (\frac{4{\rm e}^2}{\lambda^2}) = {1}/{2} \cdot {\rm log}_2 \hspace{0.1cm} (2 {\rm e}^2 \cdot \sigma^2) \hspace{0.3cm} \Rightarrow \hspace{0.3cm} {\it \Gamma}_{{\hspace{-0.05cm} \rm L}}^{\hspace{0.08cm}(Y)} = 2 \cdot {\rm e}^2 \hspace{0.15cm}\underline{\approx 14.78} \hspace{0.05cm}.$$
  • Consequently, the  ${\it \Gamma}_{{\hspace{-0.05cm} \rm L}}$ value is twice as large for the Laplace distribution as for the exponential distribution.
  • Thus, the Laplace PDF is better than the exponential PDF in terms of differential entropy when power-limited signals are assumed.
  • Under the constraint of peak limitation,  both the exponential and Laplace PDF are completely unsuitable, as is the Gaussian PDF.  These all extend to infinity.