Notes:Mdm of a discrete distribution lemma

From Maths
Revision as of 21:15, 21 January 2018 by Alec (Talk | contribs) (SAVING WORK)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
\newcommand{\P}[2][]{\mathbb{P}#1{\left[{#2}\right]} } \newcommand{\Pcond}[3][]{\mathbb{P}#1{\left[{#2}\!\ \middle\vert\!\ {#3}\right]} } \newcommand{\Plcond}[3][]{\Pcond[#1]{#2}{#3} } \newcommand{\Prcond}[3][]{\Pcond[#1]{#2}{#3} }
\newcommand{\E}[1]{ {\mathbb{E}{\left[{#1}\right]} } } \newcommand{\Mdm}[1]{\text{Mdm}{\left({#1}\right) } } \newcommand{\Var}[1]{\text{Var}{\left({#1}\right) } } \newcommand{\ncr}[2]{ \vphantom{C}^{#1}\!C_{#2} } \newcommand{\Floor}[1]{ \text{Floor}{\left({#1}\right)} }

Page notes

Let X be a distribution defined on \mathbb{Z} or a subset, which we will denote A, that is \P{X\eq k} is defined for all k\in A and there is no set of which A is a strict subset with \P{X\eq k} defined for all k within it.

  • This definition might be too general.

Statement

Below we show

  • for n\ge\Floor{\lambda}+1 then:
    • \sum^n_{k\eq 0}\vert k-\lambda\vert f(k)\eq\sum^{\Floor{\lambda} }_{k\eq 0}(\lambda-k)f(k)+\sum^n_{k\eq\Floor{\lambda}+1}(k-\lambda)f(k)
  • if n<\Floor{\lambda}+1 then we need to put like \text{Min}(\Floor{\lambda},n) as the end value for the first sum, the modifications are trivial.

We need to modify this slightly for a sum between k\eq\alpha to k\eq\beta say - but this is certainly a good result for now.

(ORIGINAL STUFF)

Let's just go ahead and state it:

Original statement

Let:

  • S:\eq\sum_{k\eq 0}^\infty \vert k-\lambda\vert f(k)[Note 1]
    • We claim:
      • S\eq \sum_{k\eq 0}^{\Floor{\lambda} }(\lambda-k)f(k)+\sum^\infty_{k\eq\Floor{\lambda}+1}(k-\lambda)f(k)

Proof

Let:

  • S:\eq\sum_{k\eq 0}^\infty \vert k-\lambda\vert f(k)
    • To evaluate this we must break the range \{0,1,2,\ldots,n-1,n,n+1,\ldots\} into a region for which
      1. \vert k-\lambda\vert\eq k-\lambda - meaning that k-\lambda\ge 0\ \implies k\ge \lambda and
      2. \vert k-\lambda\vert\eq \lambda- k meaning that k-\lambda\le 0\ \implies k\le \lambda
    • When calculating such Mdms we often use the floor function to break the domain, breaking it around \Floor{\lambda}

(Not sure where to take this)

Paper-style proof

Here's what I did on paper:

  1. I used the "floor corollary"[Note 2] which states:
    • (It may have multiple forms, they all basically say the same thing)
      1. PRIMARY FORM: \forall x\in\mathbb{R}_{\ge 0}\big[\Floor{x}\le x<\Floor{x}+1\big]
      2. ... the other forms will be slightly more detailed, for example if x\in\mathbb{R}_{\ge 0} -\mathbb{N}_{\ge 0} then \Floor{x}<x<\Floor{x}+1
  2. We look at the k\eq 0 forward case, and we see that if we sum over k for k from 0 to \Floor{\lambda} that:
    • 0\le k\le\Floor{\lambda}\le\lambda<\Floor{\lambda}+1
      • Specifically 0\le k\le\Floor{\lambda}
    • So for 0\le k\le \Floor{\lambda}\le \lambda we see that k\le \lambda\ \implies\ k-\lambda\le 0, so:
      • for 0\le k\le\Floor{\lambda}\le\lambda we have \lambda-k\ge 0 and thus \vert k-\lambda\vert\eq\vert\lambda-k\vert\eq \lambda-k as \lambda-k\ge 0 in this range
        • Specifically: 0\le k\le\Floor{\lambda} means \vert k-\lambda\vert\eq\lambda-k
    • So \sum^{\Floor{\lambda} }_{k\eq 0}\vert k-\lambda\vert f(k)\eq\sum^{\Floor{\lambda} }_{k\eq 0}(\lambda-k)f(k) - the first part of our sum
  3. Now suppose we are summing to n for n>\Floor{\lambda} then:
    • As \lambda<\Floor{\lambda}+1\le n we see for the range k\in\{\Floor{\lambda}+1,\ \Floor{\lambda}+2,\ \ldots,\ n\} we have \lambda<\Floor{\lambda}+1\le k\le n - specifically \lambda<k
      • So 0<k-\lambda which we shall write: k-\lambda>0
    • Now if k-\lambda>0 then \vert k-\lambda\eq k-\lambda when the \lambda<k condition holds
      • Thus for \Floor{\lambda}+1\le k\le n we have \vert k-\lambda\vert\eq k-\lambda
    • So \sum^n_{k\eq\Floor{\lambda}+1}\vert k-\lambda\vert f(k)\eq \sum^n_{k\eq\Floor{\lambda}+1}(k-\lambda)f(k)
  4. Lastly we note that \{0,\ldots,\Floor{\lambda} \}\cup\{\Floor{\lambda}+1,\ldots,n\} is the domain we intended to sum over

To get the stated result note that \sum^\infty_{k\eq 0}a_k:\eq \lim_{n\rightarrow\infty}(\sum_{k\eq 0}^n a_k) (by definition of limit of a series


To conclude:

  • for n\ge\Floor{\lambda}+1 then:
    • \sum^n_{k\eq 0}\vert k-\lambda\vert f(k)\eq\sum^{\Floor{\lambda} }_{k\eq 0}(\lambda-k)f(k)+\sum^n_{k\eq\Floor{\lambda}+1}(k-\lambda)f(k)
  • if n<\Floor{\lambda}+1 then we need to put like \text{Min}(\Floor{\lambda},n) as the end value for the first sum, the modifications are trivial.

Notes

  1. <cite_references_link_accessibility_label> For f(k):\eq\P{X\eq k} and \lambda:\eq\E{X} we get:
    • \sum^\infty_{k\eq 0}\vert k-\E{X}\vert\P{X\eq k}
    Which if course is the Mdm of X - provided X is defined on \mathbb{N}_0 and no set of which \mathbb{N}_0 is a proper subset
  2. <cite_references_link_accessibility_label> which should be on the floor function page but isn't there at time of writing (sig doesn't work? So: 21st of Jan 2018 ~ 9pm)