This question might be too elementary but it arises naturally as a part of a more complicated computation and I struggle to find the answer.

Let $V$ be an $n$-dimensional complex vector space. Consider a map $$Sym^m(V)\otimes Sym^k(V)\longrightarrow Sym^{m-1}(V)\otimes Sym^{k+1}(V)$$ given by $$x_1 \dots x_m \otimes y_1 \dots y_k \mapsto \sum x_1 \dots x_{i-1} x_{i+1} \dots x_m \otimes x_i y_1 \dots y_k.$$ For $m=k=1$ this map is surjective and for $m=2,\ k=0$ it is injective. In the first case the kernel is $\Lambda^2 V$ which is also a cokernel for the second case. For $n=1$ it is always an isomorphism.

  1. For what triples of $n, m, k$ this map is injective or surjective?
  2. Can the kernel / cokernel be presented functorially in terms of V?
  3. Is this map a part of some natural resolution?

1 Answer 1


[too long for a comment] Your map is the composition of two natural maps: the differential (aka polarization) $$ Sym^m(V)\longrightarrow Sym^{m-1}(V)\otimes V $$ (of it helps, regard $Sym(V)$ as an algebra of functions on $V^*$), and the symmetrisation $$ V\otimes Sym^{k-1}(V)\longrightarrow Sym^k(V)\, . $$ The former is always injective (there are no non-constant homogeneous polynomials for $m>0$), the latter is always surjective. In order to describe its kernel for $k>2$, I guess you have to consider the map $$ \Lambda^2(V)\otimes Sym^{k-2}(V)\longrightarrow V\otimes Sym^{k-1}(V)\, , $$ which is basically the odd-parity analog of your map, i.e., fist it sends $\Lambda^2(V)$ to $V\otimes V$, and then it symmetrizes $V$ with $Sym^{k-2}(V)$. [This was edited: my previous suggestion was slightly wrong]

As a matter of fact, if one of the two $V$'s appearing in your formula was a $V^*$ instead, everything would be geometrically more transparent.

Finally, if one of the $Sym$ was a $\Lambda$ instead, then your map would be part of the Koszul complex (the de Rham differential for polynomials): I would bet that in one of these "supercommutative" geometric environments (where all the parities get mixed up) there is also a parity-altered version of the Koszul complex, which may provide you some clue about the resolution.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.