Operator norm

The norm of a linear map defined as the maximal expansion of unit vectors.
Operator norm

Let (V,V)(V,\|\cdot\|_V) and (W,W)(W,\|\cdot\|_W) be normed vector spaces over F\mathbb{F}, and let T:VWT:V\to W be linear. The operator norm (or induced norm) of TT is

T:=sup{TxW:xV, xV=1}. \|T\| := \sup\{\|T x\|_W : x\in V,\ \|x\|_V=1\}.

Equivalently,

T=supx0TxWxV. \|T\| = \sup_{x\ne 0}\frac{\|Tx\|_W}{\|x\|_V}.

The operator norm measures the largest factor by which TT can stretch vectors. It is fundamental in analysis because boundedness/continuity of linear maps between normed spaces is equivalently expressed by finiteness of T\|T\|.

Examples:

  • If T:RRT:\mathbb{R}\to\mathbb{R} is T(x)=axT(x)=ax, then T=a\|T\|=|a| (with the usual absolute value norm).
  • If T:RkRkT:\mathbb{R}^k\to\mathbb{R}^k is the identity map, then T=1\|T\|=1.
  • If T:R2R2T:\mathbb{R}^2\to\mathbb{R}^2 is rotation by angle θ\theta, then T=1\|T\|=1 (it preserves Euclidean length).