Let x,y,h∈A be such that
[y,x]=h,[h,x]=2x,[h,y]=−2y(15)
(i) For m,n∈N, we have
(n)(m)(m−p)∑(m+n−p−1−h)(n−p)
xy = yx. (16)
p≥0
(ii) Let A′ be the Z-subalgebra of A generated by the x(m) and the y(m) for m∈N. Then (nh)∈A′ for all n∈N.
Formula (16) can be written in the equivalent form
(n)(m)(m−p)∑(m+n−p−1−h)(n−p)
(ad x)y = yx.(17m)
p≥1
This is trivial for m=0. We argue by induction on m. From (17m), we obtain
(m+ 1)(ad x(n))y(m+1)= (ad x(n))y(m).y+y(m).(ad x(n))y (18)
(m−p)(m+n−p−1−h)(n−p)(m)−
=∑yxy+y(n−1−h)x(n1)
p≥1
(§1, no. 1, Lemma 1). Now, applying the same lemma, and then Lemma 3,
we have
(m+n−p−1−h)(n−p)
(m+n−p−1−h)(n−p)−−
= (yx+(n−p−1−h)x(np1))
(m+n−p+1−h)(n−p)
(m+n−p−1−h)−−
+ (n−p−1−h)x(np1).
Inserting this into (18), we obtain
(m+ 1)(ad x(n))y(m+1)
−(m+n−p+1−h)−
=∑(m−p+1)y(mp+1)x(np)
p≥1
−(m+n−p−1−h)−−
+∑y(mp)(n−p−1−h)x(np1)
p≥1
+y(m)(n−1−h)x(n−1)
−(m+n−p+1−h)−
=∑(m−p+1)y(mp+1)x(np)
p≥1
(m−p)(m+n−p−1−h)−−
+∑y(n−p−1−h)x(np1).
p≥0 Changing p to p−1 in the second sum, and regrouping the terms, we obtain
(m+ 1)(ad x(n))y(m+1)=∑p≥1y(m−p+1)Apx(n−p) (19)
with
(m+n−p+1−h)(m+n−p−h)
Ap=(m−p+1)+(n−p−h).
Putting z=m+n−p−h, this can also be written as
Ap=p1!(m−p+1)(z+1)z(z−1)...(z−p+2)
+ 1 (z−m)z(z−1)...(z−p+2)
(p− 1)!
=p1! z(z−1)...(z−p+2)[(m−p+1)(z+1)+p(z−m)]
(z)((m+1)+n−p−1−h)
=(m+1)=(m+1)
Inserting this into (19), we obtain (17m+1), hence (i).
Assume that (ph)∈A′ for p<n. Then, for all P∈Q[T] of degree <n such that P(Z)⊂Z, we have P(h)∈A′ (no. 4, Cor. of Prop. 2). Hence, in view of (16) with m=n,
(h)(n−1−h)
(−1)n =
−n−1(2n−p−1−h)−′
=−x(n)y(n)+∑y(np)x(np)∈A ;
p=0
hence (ii) by induction on n.
Bourbaki sets the proof directly after the statement, and the corpus does not separate them, so what is above is the statement and its proof together.