The name is absent

Note that Gaussian random variables describe variations in the dimensions
of gates (or equivalently gate delays), i.e., d_u = d%_l + ψ_u where d® is nominal
dimension of the gate.

2.2.2 Compressive sensing _β

The compressive sensing concepts, that enable us to reconstruct a sparse vector
by partial measurement, are explained here (see [10,15,24]). A vector is called
s-sparse when it has only s non-zero elements. Assume X is an s-sparse N × 1
vector. Assume Y is described based on the following equation

Y = UX + e.                        (2.2)

Vector X is the unknown sparse vector; U is a known K×N measurement matrix
and e is measurement noise. Note that not only are the values of the non-zero
components of X are not known, neither which components are zero. The vector
Y is our observation (measurement). The goal is to estimate the sparse vector X
using the measurement vector Y. To retrieve the vector X_f one might choose a
vector that minimizes ∣∣V — tλX^∣∣2∙ Because of the measurement noise and small
number of measurements, this procedure usually leads to a non-sparse signal.
However, solving the following optimization problem finds an sparse solution

min ∣∣X∣∣₀ + α∣∣e∣∣₂                           (2.3)

such that Y = UX + e,

20

<<   26   27   28   29   30   31   32   >>

More intriguing information