A point of a set is an extreme point if , with and , forces . For a convex set it is enough to test midpoint decompositions: a nontrivial segment containing in its interior contains a smaller segment with midpoint . The nonempty convex set has no extreme points, since every is the midpoint of and .
For the real l-infinity sequence space, a vector in the closed unit ball is extreme precisely when every coordinate is or . If for some , choose . The distinct vectors remain in the unit ball and have midpoint , so is not extreme. Conversely suppose all coordinates have modulus one and , with . In each real interval , an endpoint can be a midpoint only if both entries equal that endpoint. Thus for every and . Therefore
For the real absolutely summable sequence space, any point with is nonextreme: choose a nonzero coordinate perturbation of norm less than . If and two coordinates are nonzero, take and put
Both perturbations preserve the signs of those two coordinates. One adds to one absolute value and subtracts it from the other, so . Hence such an is not extreme. The only remaining candidates are . If with , its th coordinate forces . That consumes the entire norm allowance, forcing all other coordinates to vanish. The argument for is identical. Thus
For the real l2 sequence space, interior points of the unit ball again permit sufficiently small opposite perturbations and are not extreme. If and , write , . The parallelogram law gives
The left side is at most one, so and . Consequently
These are the extreme points of real sequence-space unit balls: signs in , coordinate spikes in , and the whole sphere in . The restriction to real scalars matters for the first description; the complex supremum ball has arbitrary unit-modulus coordinates instead.

Articles by others on the same topic (0)

There are currently no matching articles.