Use the usual weighted interpretation, identifying functions equal almost everywhere; on raw pointwise functions the displayed integral is only a seminorm. Assume . If is orthogonal to , then for every ,
Conversely, suppose minimizes the error. For and every real , compare with :
If the linear coefficient were nonzero, a sufficiently small with the appropriate sign would make the expression negative. Thus for every , proving best approximation is equivalent to orthogonality of the residual.

Articles by others on the same topic (0)

There are currently no matching articles.