The annihilator of a subspace
Let W be a subspace of V. Its annihilator, written W^0, is the set of all functionals that kill every vector of W: W^0 = { f in V* : f(w) = 0 for all w in W }. It is itself a subspace — of V*, not of V. Think of it as 'all the measurements that can't see W'.
The dimension law
Here is the clean theorem: dim W + dim W^0 = dim V. The bigger your subspace, the fewer functionals can blindly annihilate it. If W is a line in R^3 (dim 1), its annihilator is a plane of functionals (dim 2). This is the annihilator duality theorem, and it is the dual-space twin of the rank-nullity theorem.
- Pick a basis w1, ..., wk of W and extend it to a basis w1, ..., wk, u1, ..., um of all of V.
- Take the dual basis. The functionals dual to the u's vanish on every w — so they lie in W^0.
- Those m functionals are a basis of W^0, so dim W^0 = m = dim V - dim W.
Worked example and a glimpse ahead
V = R^3. W = span{ (1,2,0) } (a line, dim 1).
W^0 = { rows [a b c] : a*1 + b*2 + c*0 = 0 } = { a + 2b = 0 }.
Solve: a = -2b, c free. Basis of W^0:
f1 = [-2 1 0] (b=1, c=0)
f2 = [ 0 0 1] (b=0, c=1)
dim W^0 = 2, dim W = 1, sum = 3 = dim V. ✓
Geometrically: f1=0 and f2=0 are two planes whose
intersection is exactly the original line W.Two related facts you'll meet soon: a single nonzero functional cuts out a hyperplane (a separating functional that puts a subspace on one side), and the annihilator W^0 is naturally the dual of the quotient V/W. Annihilators are the dictionary that translates 'subspace of V' into 'subspace of V*'.