Let $k$ be a field, let $V$ and $W$ be $k$-vector spaces of dimensions $n$ and $m$ respectively, and let $f:V\to W$ be a $k$-linear transformation. Let $\Lambda(V)$ and $\Lambda(W)$ denote the exterior algebras of $V$ and $W$ respectively. So we have $$\Lambda(V) = \Lambda^0(V)\oplus\Lambda^1(V)\oplus\cdots\oplus\Lambda^n(V)$$ and $$\Lambda(W) = \Lambda^0(W)\oplus\Lambda^1(W)\oplus\cdots\oplus\Lambda^m(W).$$

The wikipedia page on exterior algebras states that there is a unique function $\Lambda(f):\Lambda(V)\to\Lambda(W)$ such that $\Lambda(f)|_{\Lambda^1(V)}:\Lambda^1(V)\to\Lambda^1(W)$ is defined by $\Lambda(f)(v)=f(v)$.

In fact, $\Lambda(f)$ preserves grading (i.e. it can be written as a sum of maps $\Lambda^k(f):=\Lambda(f)|_{\Lambda^k(V)}:\Lambda^k(V)\to\Lambda^k(W)$). If $1\leq k \leq n$, then $\Lambda(f)$ is given by $$\Lambda^k(f)(v_1\wedge\cdots\wedge v_k) = f(v_1)\wedge\cdots\wedge f(v_k).$$

I do not understand how this function acts on $\Lambda^0(V)=k$. I know that we have a map $$\Lambda^0(f):\Lambda^0(V)\to\Lambda^0(W)$$ which is really the same as $$\Lambda^0(f):k\to k.$$

My question has two parts: what is $\Lambda^0(f)$ and how is it determined from the universal mapping property for exterior algebras?

  • 1
    $\begingroup$ Isn't it just the identity map? You need it to be $k$-linear, so you need $\Lambda^0(f)(a) = \Lambda^0(f)(a\cdot 1) = a\Lambda^0(1)$, so it is completely determined by the image of $1$. And $\Lambda^k(f)(v) = \Lambda^k(f)(1\cdot v) = \Lambda^0(f)(1)\wedge \Lambda^k(f)(v)$, so $\Lambda^0(f)(1)=1$. But I could be doing something unwarranted here. $\endgroup$ – Arturo Magidin Mar 28 '11 at 4:41
  • 1
    $\begingroup$ You are quite correct, Arturo. The same thing happens with the tensor algebra, or the symmetric algebra... Also, the image of $1$ has to be $1$ for the map to be a morphism of algebras, so one can skip the computation :) $\endgroup$ – Mariano Suárez-Álvarez Mar 28 '11 at 4:46
  • $\begingroup$ @Arturo. I was unaware that $\Lambda^k(f)(a\cdot v) = \Lambda^0(f)(a)\wedge\Lambda^k(v)$. Does that derivation still work if $f:V\to W$ is the zero map? $\endgroup$ – NymSudo Mar 28 '11 at 4:58
  • $\begingroup$ @Nymsudo: I derived that from the graded structure: if $a\in\Lambda^k(V)$ and $b\in\Lambda^{\ell}(V)$, then $a\wedge b = (-1)^{k\ell}(a\wedge b)\in\Lambda^{k+\ell}$. The fact that your map has to respect the graded structure gives that equality. $\endgroup$ – Arturo Magidin Mar 28 '11 at 13:36
  • $\begingroup$ @Arturo: If you answer, I'll upvote to get this off the unanswered list. $\endgroup$ – Jonas Meyer Apr 2 '11 at 19:45

The map is the identity map. The map must be $k$-linear, so $$\Lambda^0(f)(a) = \Lambda^0(f)(a\cdot 1) = a\Lambda^0(f)(1)$$ hence the map is completely determined by the image of $1$. But since it is a morphism of algebras, the map must send $1$ to $1$, so $\Lambda^0(f)(1)=1$, hence $\Lambda^0(f)(a) = a$ for all $a\in k$.


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.