Opening the paper…
Figure from the original question paper The random variable $y$ is being transformed according to the measurement mapping: $$z = Cy + \epsilon,$$ where: - $z \in \mathbb{R}^d$, - $C \in \mathbb{R}^{d \times n}$, - $\epsilon \sim \mathcal{N}(0, \sigma^2 I_d)$ is independent Gaussian (measurement) noise. Which of the following represents the conditional probability distribution $p(z \mid y)$? Which of the following statements about singular value decomposition (SVD) is/are true?