Next:Properties of Unified InaccuraciesUp:Generalized Information Measures
Previous:Properties of Unified Relative Go to:Table of Contents

Unified (r,s)-Inaccuracies

Similar to expression (4.1), we shall give below the three different forms of writing directly the unified$ (r,s)-$inaccuracy measures in three different forms:
$\displaystyle ^\alpha{\ensuremath{\boldsymbol{\mathscr{H}}}}^s_r(P\vert\vert Q)......1 \\ H(P\vert\vert Q)=-\sum_{i=1}^n{p_i\log q_i},& r=1,\ s=1\end{array}\right.$
    (4.5)

$ \alpha=1,2$ and 3, where

$ \displaystyle ^1M_r(P\vert\vert Q)=\sum_{i=1}^n{p_iq^{r-1}_i},$
$ \displaystyle ^2M_r(P\vert\vert Q)=\Big(\sum_{i=1}^n{p_iq^{(r-1)/r}_i}\Big)^r$
and
$ \displaystyle^3M_r(P\vert\vert Q)={\sum_{i=1}^n{p^r_i}\over\sum_{i=1}^n{p^r_iq^{1-r}_i}}.$
The expressions $ ^1H^1_r(P\vert\vert Q)$$ ^2H^1_r(P\vert\vert Q)$ and$ ^3H^1_r(P\vert\vert Q)$ are due to Nath (1968) [73], Van der Lubbe (1978) [115] and Nath (1975) [74] respectively. The expressions $ ^1H^s_r(P\vert\vert Q)$ and $ H^s_1(P\vert\vert Q)$ are due to Sharma and Gupta (1976) [89].

We can write 

$\displaystyle ^\alpha {\ensuremath{\boldsymbol{\mathscr{H}}}}^s_r(P\vert\vert Q...... {\ensuremath{\boldsymbol{\mathscr{H}}}}^1_r(P\vert\vert Q)\big),\ \alpha=1,2\ $   and$\displaystyle \ 3 $

where $ {\ensuremath{\boldsymbol{\mathscr{N}}}}_s$ is as given in (4.4).

From the unified espression (4.5) it is clear that the measures $ ^\alpha {\ensuremath{\boldsymbol{\mathscr{H}}}}^s_r(P\vert\vert Q)\ (\alpha=1,2\ $   and$ \3)$ are continuous with respect to the parameters $ r$ and $ s$. This allows us to write 

$\displaystyle ^\alpha {\ensuremath{\boldsymbol{\mathscr{H}}}}^s_r(P\vert\vert Q......symbol{\mathscr{H}}}}^s_r(P\vert\vert Q):r\neq 1,\ s\neq 1\Big\},\ \alpha=1,2\ $   and$\displaystyle \ 3, $

where "CE" stands for "continuous extension".

In particular, when $ P=Q$, we have 

$\displaystyle ^1{\ensuremath{\boldsymbol{\mathscr{H}}}}^s_r(P\vert\vert Q)=\,\,......athscr{H}}}}^s_r(P\vert\vert Q)={\ensuremath{\boldsymbol{\mathscr{H}}}}^s_r(P),$

and when$ Q=P^r$, we have 
$\displaystyle ^2{\ensuremath{\boldsymbol{\mathscr{H}}}}^s_r(P\vert\vert Q)={\ensuremath{\boldsymbol{\mathscr{H}}}}^s_r(P),$
where
$\displaystyle P^r=\Big( {p^r_1\over \sum_{i=1}^n{p^r_i}},...,{p^r_n\over\sum_{i=1}^n{p^r_i}}\Big) \in \Delta_n.$


21-06-2001
Inder Jeet Taneja
Departamento de Matemática - UFSC
88.040-900 Florianópolis, SC - Brazil