2
$\begingroup$

According to the wikipedia article on the lognormal distribution, the lognormal distribution is "the maximum entropy probability distribution for a random variate $X$ for which the mean and variance of $\log(X)$ is fixed".

Is there a not too complicated account of what this means and how this is derived?

$\endgroup$
3
  • $\begingroup$ You didn't correctly copy the sentence from wikipedia. Insert a $\ln(X)$ between 'is' and 'fixed'. $\endgroup$ – Glen_b Feb 26 '13 at 2:47
  • $\begingroup$ I've inserted a $\ln(X)$ before "is fixed", thanks for the hint. $\endgroup$ – Pugl Feb 26 '13 at 7:44
  • $\begingroup$ The Wikipedia article references a clear paper by Park & Bera in which the result is derived. $\endgroup$ – whuber Feb 26 '13 at 15:10
3
$\begingroup$

Note that the $\log(X)\sim \mbox{Normal}(\mu,\sigma^2)$, therefore they are actually making a claim regarding the normal distribution. For a normal distribution we have that

The normal distribution Normal$(\mu,\sigma^2)$ has maximum entropy among all real-valued distributions with specified mean $\mu$ and standard deviation $\sigma$.

Check this wikipedia entry for more details (including the proof of this result): http://en.wikipedia.org/wiki/Maximum_entropy_probability_distribution#Given_mean_and_standard_deviation:_the_normal_distribution

$\endgroup$
2
  • 1
    $\begingroup$ Thanks a lot for this hint, it was quite helpful. Also I had seen the paper linked on wikipedia, it did not look trivial though. I have one question regarding this property: We do have biological data where the data is a set of measured lenghts, and these lengths have a lognormal distribution. Knowing that in biology cost-minimization is crucial, I wondered if the Maximum Entropy property could be linked to cost-reduction? Sorry if the question is vague! P. $\endgroup$ – Pugl Feb 28 '13 at 20:00
  • $\begingroup$ This is a new question, you should then ask it as a new question, not just in comments! $\endgroup$ – kjetil b halvorsen Jun 5 '17 at 23:38

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.