I encountered an example in a paper telling that $\underline{SM}(\mathbb{R}^{0|1},X)\cong \pi TX $, where $X$ is some fixed ordinary Riemannian manifold, $\pi TX $ is the supermanifold with base manifold $X$ and structural sheaf $\pi(\wedge^*(TX)^\vee) $ according to my advisor, seen as a functor $\pi TX:=SM(-,\pi TX): SM^{\mathrm{op}} \to Set$, $S\mapsto SM(S,\pi TX)$ where $SM$ stands for the category of supermanifolds. $\underline{SM}(\mathbb{R}^{0|1},X)$ is the functor which acts on objects by $\underline{SM}(\mathbb{R}^{0|1},X)(S)= SM(S\times \mathbb{R}^{0|1},X) $, with the obvious action on morphisms. "$\cong$" means that there is a natural isomorphism between the two functors.

Now, by the theorem that supermanifolds are "affine" we have $$\underline{SM}(\mathbb{R}^{0|1},X)(S) \cong SAlg_{\mathbb{R}}(C^\infty(X), C^\infty(S)\otimes C^\infty(\mathbb{R}^{0|1})) $$

where $SAlg_{\mathbb{R}}$ is the category of super $\mathbb{R}$-algebra in which the morphisms are parity-preserving $\mathbb{R}$-algebra homomorphisms. Let $\theta$ be the odd coordinate of $\mathbb{R}^{0|1}$ and we see that an element in $\underline{SM}(\mathbb{R}^{0|1},X)(S)$ is identified with a super $\mathbb{R}$-algebra morphism $\Phi^*:C^\infty(X)\to C^\infty(S)\otimes C^\infty(\mathbb{R}^{0|1})\cong C^\infty(S) \oplus C^\infty(S)\cdot \theta $.

Decompose $\Phi^*$ using the direct sum and we can write $\Phi^*=f+\phi\theta$ where $f:C^\infty(X)\to C^\infty(S)$ is a super $\mathbb{R}$-algebra morphism and $\phi:C^\infty(X)\to C^\infty(S)$ is a parity-reversing map such that

$$\phi(ab)=\phi(a)f(b)+f(a)\phi(b)= \phi(a)f(b)+(-1)^{p(a)}f(a)\phi(b)$$ for any $a,b\in C^\infty(X)$, where $p(a)$ stands for the parity of $a$ which is always zero.

The paper refers to Deligne & Morgan's Notes on Supersymmetry (following Joseph Bernstein) and says that this is the standard description of $\pi TX$ in terms of its $S$-points so we get $\underline{SM}(\mathbb{R}^{0|1},X)\cong \pi TX $, but Deligne & Morgan's description is not clear to me at all. My advisor told me that $\pi TX$ is supposed to be as above and left the rest to me as an exercise.

I have no difficulty understanding $\pi TX$ for sure, but I don't have a clue how a morphism $\varphi: S\to \pi TX$, which can be identified with a super $\mathbb{R}$-algebra morphism $\varphi^*:\pi(\wedge^*(TX)^\vee) \to C^\infty(S) $, could be translated to some $\Phi^*:C^\infty(X)\to C^\infty(S)\oplus C^\infty(S)\cdot \theta$ satisfying the axioms above.

For the simplest case, say $X=\mathbb{R}^1$, we have $\pi(\wedge^*(TX)^\vee) =C^\infty(\mathbb{R})\oplus C^\infty(\mathbb{R})\cdot dx$ where $dx$ is considered to be odd. From this $\varphi^*$ gives a parity-preserving $\mathbb{R}$-algebra $\bar f:C^\infty(X)\to C^\infty(S)$ which I guess is the candidate of $f$ (or not?). Put $s:=\varphi^*(dx)$ and we see that $\varphi^*$ is determined by $\bar f$ and $s$, as

$$ \varphi^* (a+bdx)=\bar f(a)+\bar f(b)s.$$

Unless $\phi=\bar f\cdot s$, which doesn't make any sense, I cannot see where $\phi$ could possibly come from. I think I do need some help about this. Thanks in advance.

  • $\begingroup$ I give a hint; maybe someone else will spell it out. The component $f$ of your $\Phi^*$ is (dual to) the composition $S \overset\varphi\to \pi T X \to X$. The remaining data is basically a vector field on $S$; more precisely, it is a field on $S$ valued the pulled-back tangent bundle $f^* \pi T_X$. Now recall that vector fields are the same as derivations. In this case, you find that sections of $f^* \pi T_X$ are precisely (odd) "derivations along $f$", i.e. functions $\phi$ satisfying the Leibniz-like relation that you wrote down. $\endgroup$ Mar 1 at 16:35
  • $\begingroup$ @TheoJohnson-Freyd Thanks for the hint; so my naive observation for candidate of $f$ is correct. But how does the tangent bundle $\pi TX$ pullback to a bundle on $S$? It seems that it should be compatible with post-composition a derivation $D:C^\infty(X)\to C^\infty(X)$ by $f$, getting $f\circ D: C^\infty(X)\to C^\infty(S)$, but for an arbitrary open subset $U\subset S$, what is $(f^*\pi TX)(U)$ supposed to be? To have a locally free sheaf we can't take simply the composition $C^\infty(X)\to C^\infty(S)\to C^\infty(U)$; if $X$ should be replaced by some open subset of $X$, what is our choice? $\endgroup$
    – Shana
    Mar 2 at 3:15
  • $\begingroup$ @TheoJohnson-Freyd Moreover, even with $f^*\pi TX$ defined as a sheaf, I can't see how a section might be related to $\varphi:S\to \pi TX$, with $\pi TX$ the supermanifold. I read similar things in Deligne & Morgan before, but nothing is clear to me. $\endgroup$
    – Shana
    Mar 2 at 3:22

1 Answer 1


My advisor told me the answer; I was just one step away from it.

A morphism $\varphi:S\to \pi TX$ is determined by the natural transformation between the structural sheaves, which is locally determined, so it suffices to assume that $X$ has coordinates $x_1,\cdots,x_n$. Now $C^\infty(\pi TX)=\pi(\wedge^*(TX)^\vee) $ is finitely generated by $d x_1,\cdots,dx_n$ over $C^\infty(X)$, the composition $C^\infty(X)\hookrightarrow \pi(\wedge^*(TX)^\vee)\to C^\infty(S)$ gives our $f$ and the remaining data of $\varphi^*:\pi(\wedge^*(TX)^\vee)\to C^\infty(S)$ is given by the images of $d x_1,\cdots,dx_n$.

Let $s^i:= \varphi^*(dx_i)$, then $\varphi^*$ on $dx_i$'s can be expressed by (with abuse of notation) $\varphi^*=\sum_i \partial_{x_i}\otimes_f s^i$ in light of $\left(\sum_i\partial_{x_i}\otimes _f s^i\right)(dx_j)= \sum_i dx_j(\partial_{x_i})\otimes_f s^i =s^j $.

Now, $\sum_i \partial_{x_i}\otimes_f s^i$ gives an element in $\text{Der}(C^\infty(X),C^\infty(X))\otimes_f C^\infty(S)$, which is by definition (see Wikipedia) a global section of the pullback bundle $f^*\pi TX$ on $S$. Finally, as $\text{Der}(C^\infty(X),C^\infty(X))\otimes_f C^\infty(S)\cong \text{Der}_f(C^\infty(M),C^\infty(N))$, where $\text{Der}_f$ denotes derivations with respect to $f$, via $W\otimes s\mapsto s(f^*\circ W)$ with inverse $V\mapsto \sum_i \partial_{x_i}\otimes _f V(x_i) $, the above relates $\varphi$ with $(f,\phi)$ bijectively.


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.