ferreat Posted July 11, 2013 Share Posted July 11, 2013 Hello,I have a doubt about the distribution of random variables that maximize the differential entropy in a set of inequalities. It is well known that the Normal distribution maximizes the differential entropy. I have the following set of inequalities:T1 < I(V;Y1|U)T2 < I(U;Y2)T3 < I(X1,X2;Y3|V)T4 < I(X1,X2;Y3)where, Y1=X1+N1, Y2=a*X1+N2, Y3=b*X1+X2+N3. N1,N2,N3 are Gaussian ~ N(0,1). The lower case a and b are positive real numbers a < b. U, V, X1 and X2 are random variables. I want to maximize that set of inequalities. I know the following:(i) From T4, h(Y3) maximum is when Y3 is Gaussian then X1 and X2 are Gaussian.(ii) From T2 we maximize it by having h(Y2) or h(a*X1+N2) maximum. From this by the Entropy Power Inequality (EPI) we bound -h(a*X1+N2|U) and have X1|U Gaussian.(iii) From T1 we maximize it by having h(Y1|U) or h(X1+N1|U) maximum which we can do as -h(a*X1+N2|U) in the part ii can be bounded having Y1 Gaussian (satisfying the maximum entropy theorem).The Question:From T3, can I assume that jointly Gaussian distribution will maximize h(Y3|V) or h(b*X1+X2+N3) having the assumptions i,ii,iii ?My aim is to show that jointly Gaussian distribution of U, V, X1 and X2 maximizes the set of inequalities. I hope anyone can help me out with this. Link to comment Share on other sites More sharing options...
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now