Math Problem.

<p>Since this place seems like it full of people with excellent math skills, I thought maybe I could ask for help on a difficult math problem. I am completely stumped on how to do this.</p>

<p>Given random variables X1,X2,…,Xn. </p>

<p>X1,…,Xn are independent if and only if</p>

<p>fx1,x2,…,xn(x1,x2,…,xn) = fx1(x1)<em>fx2(x2)</em>…*fx3(x3)</p>

<p>if and only if </p>

<p>Fx1,x2,…,xn(x1,x2,…,xn) = Fx1(x1)<em>Fx2(x2)</em>…*Fx3(x3)</p>

<p>if and only if </p>

<p>Mx1,M2,…,Mn(x1,x2,…,xn) = Mx1(x1)<em>Mx2(x2)</em>…*Mx3(x3).</p>

<p>f is the pdf
F is the cdf
M is the mgf</p>

<p>help me prove this statement.</p>

<p>sad no on here could answer this very difficult math problem.</p>

<p>create a matrix with row vectors and each vector-variable X(n) for each column. take the determinant of that matrix. if it is non-zero, the contained vectors are linearly independent.</p>

<p>you just need to show that they form a basis for Rn:
X(n) x X(n+1) =/= 0 . that is… vector cross vector is not zero and therefore forms a basis in R2</p>

<p>What you posted has nothing to do with my question. My question is about independence of random variables.</p>

<p>It is trivial to prove that if f(x1,x2,…,xn) = f(x1)<em>f(x2)</em>…*f(xn) then X1, X2, …, Xn are independent. It suffices to see that</p>

<p>P( (X1,X2,…,Xn) \in A1 x A2 x … An) = \int_{A1 x A2 x…x An} f(x1,…,xn)
dxn…dx1</p>

<p>= \prod<em>{i=1}^{n} \int</em>{A<em>{i}} f(x</em>{i}) dx_{i} </p>

<p>= \prod<em>{i=1}^{n} P ( X</em>{i} \in A_{i}), \forall A1, …, An</p>

<p>The only if part is also trivial when a joint pdf exists. It suffices to see that, if X1, …Xn are independent, then</p>

<p>P( X1 <= x1, … Xn <= xn) =</p>

<p>= P(X1<=x1) * P(X2<=x2) *… P(Xn <=xn)</p>

<p>= F(x1) * F(x2) <em>…</em> F(xn).</p>

<p>Taking the derivate on both sides with respective to x1, x2,…xn, we get then f(x1,…,xn) = f(x1) * f(x2) <em>…</em> f(xn). </p>

<p>Incidentally, I also proved above that if X1,…Xn are independent, then the joint cdf is separable. The reverse again is trivial when a joint pdf exists (if a joint pdf does not exist, it is true, but requires higher math, namely namely measure theory ).</p>

<p>BTW, you can also use a similar reasoning to prove the statements for the joint characteristic function recalling that it is like the Fourier transform of the joint pdf f(x1,…,xn).</p>

<p>See, I wanted to proof if a joint pdf does not exsist as well for the cdf. Is it that complicated.</p>

<p>Thanks.</p>