91Ó°ÊÓ

A random vector is a collection of random variables often treated as a single entity because they are defined in a joint probability space. Think of it as an array of random values that can have correlations between one another. For example, in our exercise, the random vector \( Z \) contains four components: \( (Z_1, Z_2, Z_3, Z_4) \). Each of these components is random, meaning that they can vary and follow a specific probability distribution.
One important characteristic of the random vector \( Z \) is its covariance matrix, \( \sigma^2 I \). This indicates that each component \( Z_i \) has the same variance \( \sigma^2 \) and they are uncorrelated with one another. Knowing this makes further analysis, like calculating covariances, easier because we deal with simpler mathematical properties.

Linear combinations are fundamental in understanding random vectors and transformations. They involve adding and subtracting vectors with specific coefficients. In our problem, \( U \) and \( V \) are linear combinations of the components of \( Z \). Specifically, \( U = Z_1 + Z_2 + Z_3 + Z_4 \) and \( V = (Z_1 + Z_2) - (Z_3 + Z_4) \).
By creating linear combinations, we can simplify complex random vectors into more manageable forms. This method allows us to express these combinations in matrix form, facilitating their manipulation in mathematical equations. Understanding these combinations is crucial as they serve as the building blocks for matrix operations used in deriving properties like covariance.

Matrix multiplication is a core mathematical operation that allows combining matrices to transform vectors or express linear transformations. It is akin to blending two sets of numbers to produce a new set. For our random vector exercise, matrix multiplication was essential in determining the covariance of \( U \) and \( V \).
To express \( U \) and \( V \) in matrix form, we formed a matrix \( A \) representing their linear combinations and multiplied it with the covariance matrix of \( Z \). The formula used was \( \text{Cov}(U, V) = A \text{Cov}(Z) A^T \). Here, \( A \) is multiplied by the identity matrix \( \sigma^2 I \), followed by \( A^T \), the transpose of \( A \). This multiplication process simplified to give us a new covariance matrix where \( U \) and \( V \) were uncorrelated.

A transformation matrix is a tool that applies a specific linear transformation to a vector, changing its space or configuration. In our scenario, the transformation matrix \( A = \begin{pmatrix} 1 & 1 & 1 & 1 \ 1 & 1 & -1 & -1 \end{pmatrix} \) was used to convert the components of the random vector \( Z \) into the variables \( U \) and \( V \).
This matrix enabled us to express \( U \) and \( V \) as linear combinations of the original \( Z \) components. By applying matrix multiplication, we transformed the properties of \( Z \) into the properties of \( U \) and \( V \), notably, their variance and covariance. The transformation matrix, therefore, played a vital role by allowing us to capture important statistical properties between \( U \) and \( V \) at a new level of abstraction.