Machine-translated by Google. Questions are authored in English.
Top

Discussion

The characteristic function of a vector \(\underline{X} \sim N_{p}(\underline{\mu}, \Sigma)\) is given by :

  • A.\(e^{\mu't - \frac{1}{2}t'\Sigma t}\)
  • B.\(e^{i\mu't - \frac{1}{2}t'\Sigma t}\)
  • C.\(e^{\mu't - \frac{i}{2}t'\Sigma t}\)
  • D.\(e^{i\mu't - \frac{i}{2}t'\Sigma t}\)

Answer: B

The characteristic function is \(e^{i\mu' t - \frac{1}{2}t'\Sigma t}\).

For \(X \sim N_{p}(\mu, \Sigma)\), \(t'X\) is univariate normal with mean \(t'\mu\) and variance \(t'\Sigma t\). The characteristic function of \(N(m, v)\) at 1 is \(e^{im - v/2}\), so

\[ E[e^{it'X}] = e^{i\mu' t - \frac{1}{2}t'\Sigma t} \]

Option A is the moment generating function evaluated wrongly (the m.g.f. is \(e^{\mu' t + \frac{1}{2}t'\Sigma t}\)). Options C and D put \(i\) on the quadratic term, which would make the characteristic function have modulus 1 everywhere, impossible for a non-degenerate normal.

No comment is present. Be the first to comment.
Loading…

Post your comment

Please log in or register to post a comment.