Metamath Proof Explorer


Theorem minveco

Description: Minimizing vector theorem, or the Hilbert projection theorem. There is exactly one vector in a complete subspace W that minimizes the distance to an arbitrary vector A in a parent inner product space. Theorem 3.3-1 of Kreyszig p. 144, specialized to subspaces instead of convex subsets. (Contributed by NM, 11-Apr-2008) (Proof shortened by Mario Carneiro, 9-May-2014) (New usage is discouraged.)

Ref Expression
Hypotheses minveco.x
|- X = ( BaseSet ` U )
minveco.m
|- M = ( -v ` U )
minveco.n
|- N = ( normCV ` U )
minveco.y
|- Y = ( BaseSet ` W )
minveco.u
|- ( ph -> U e. CPreHilOLD )
minveco.w
|- ( ph -> W e. ( ( SubSp ` U ) i^i CBan ) )
minveco.a
|- ( ph -> A e. X )
Assertion minveco
|- ( ph -> E! x e. Y A. y e. Y ( N ` ( A M x ) ) <_ ( N ` ( A M y ) ) )

Proof

Step Hyp Ref Expression
1 minveco.x
 |-  X = ( BaseSet ` U )
2 minveco.m
 |-  M = ( -v ` U )
3 minveco.n
 |-  N = ( normCV ` U )
4 minveco.y
 |-  Y = ( BaseSet ` W )
5 minveco.u
 |-  ( ph -> U e. CPreHilOLD )
6 minveco.w
 |-  ( ph -> W e. ( ( SubSp ` U ) i^i CBan ) )
7 minveco.a
 |-  ( ph -> A e. X )
8 eqid
 |-  ( IndMet ` U ) = ( IndMet ` U )
9 eqid
 |-  ( MetOpen ` ( IndMet ` U ) ) = ( MetOpen ` ( IndMet ` U ) )
10 oveq2
 |-  ( j = y -> ( A M j ) = ( A M y ) )
11 10 fveq2d
 |-  ( j = y -> ( N ` ( A M j ) ) = ( N ` ( A M y ) ) )
12 11 cbvmptv
 |-  ( j e. Y |-> ( N ` ( A M j ) ) ) = ( y e. Y |-> ( N ` ( A M y ) ) )
13 12 rneqi
 |-  ran ( j e. Y |-> ( N ` ( A M j ) ) ) = ran ( y e. Y |-> ( N ` ( A M y ) ) )
14 eqid
 |-  inf ( ran ( j e. Y |-> ( N ` ( A M j ) ) ) , RR , < ) = inf ( ran ( j e. Y |-> ( N ` ( A M j ) ) ) , RR , < )
15 1 2 3 4 5 6 7 8 9 13 14 minvecolem7
 |-  ( ph -> E! x e. Y A. y e. Y ( N ` ( A M x ) ) <_ ( N ` ( A M y ) ) )