If 2 is not a zero divisor inA, the elementsJ(ep)forp∈P∩Cform a basis of the module of anti-invariant elements ofA[P].
The weights w(p) for w∈W and p∈P∩C are pairwise distinct. It follows that the J(ep) for p∈P∩C are linearly independent.
On the other hand, let x=∑pxpep be an anti-invariant element of A[P]. If p0 belongs to a wall, it is invariant under a reflection s∈W and
x=p∑xpep=−s(x)=−p∑xpes(p).
It follows that 2xp0=0, so xp0=0. Since every element that does not belong to any wall can be written uniquely in the form w(p) with w∈W and p∈P∩C, we thus have
x=p∈P∩C∑w∈W∑xw(p)ew(p).(2)
Since w(x)=∑pxpew(p)=ε(w)∑pxpep, xw(p)=ε(w)xp and we deduce from (2) that
x=p∈P∩C∑xpJ(ep),
which completes the proof.
Consider now the element d of the algebra A[21P] defined by