The Euclidean space¶
In this book we will be generally concerned with the Euclidean space \(\RR^N\). This section summarizes important results for this space.
\(\RR^2\) (the 2-dimensional plane) and \(\RR^3\) the 3-dimensional space are the most familiar spaces to us.
\(\RR^N\) is a generalization in \(N\) dimensions.
An element \(x\) in \(\RR^N\) is written as
where each \(x_i\) is a real number.
Vector space operations on \(\RR^N\) are defined by:
\(\RR^N\) comes with the standard ordered basis \(B = \{e_1, e_2, \dots, e_N\}\):
An arbitrary vector \(x\in\RR^N\) can be written as
Inner product¶
Standard inner product (a.k.a. dot product) is defined as:
This makes \(\RR^N\) an inner product space.
The result is always a real number. Hence we have symmetry:
Norm¶
The length of the vector (a.k.a. Euclidean norm or \(\ell_2\) norm) is defined as:
This makes \(\RR^N\) a normed linear space.
The angle \(\theta\) between two vectors is given by:
Distance¶
Distance between two vectors is defined as:
This distance function is known as Euclidean metric.
This makes \(\RR^N\) a metric space.
\(\ell_p\) norms¶
In addition to standard Euclidean norm, we define a family of norms indexed by \(p \in [1, \infty]\) known as \(l_p\) norms over \(\RR^N\).
\(\ell_p\) norm is defined as:
\(\ell_2\) norm¶
As we can see from definition, \(\ell_2\) norm is same as Euclidean norm. So we have:
\(\ell_1\) norm¶
From above definition we have
We use norms as a measure of strength of a signal or size of an error. Different norms signify different aspects of the signal.
Quasi-norms¶
In some cases it is useful to extend the notion of \(\ell_p\) norms to the case where \(0 < p < 1\).
In such cases norm as defined in (2) doesn’t satisfy triangle inequality, hence it is not a proper norm function. We call such functions as quasi-norms.
\(\ell_0\)-“norm”¶
Of specific mention is \(\ell_0\)-“norm”. It isn’t even a quasi-norm. Note the use of quotes around the word norm to distinguish \(\ell_0\)-“norm” from usual norms.
\(\ell_0\)-“norm” is defined as:
where \(\supp(x) = \{ i : x_i \neq 0\}\) denotes the support of \(x\).
Note that \(\| x \|_0\) defined above doesn’t follow the definition in (2).
Yet we can show that:
which justifies the notation.