I'm trying to show that for an ordinary manifold $X$ and a supermanifold $S$, supermanifold morphisms $\varphi:S\to TX$ are one-to-one to the pairs $(f,F) $ where $f:C^\infty(X)\to C^\infty(S)$ is a super $\mathbb{R}$-algebra homomorphism and $F:C^\infty(X)\to C^\infty(S)$ is an even derivation with respect to $f$, i.e. $F$ is a parity-preserving $\mathbb{R}$-linear map and $F(ab)=F(a)f(b)+f(a)F(b)$ for any $a,b\in C^\infty(X)$.

Intuitively, $f$ is the pullback of smooth maps from the base manifold $X$ to $S$ and $F$ is a global section to the pullback vector bundle $f^*TX$ on $S$. Hence if we let $p:TX\to X$ be the usual projection, then $f= \varphi^*\circ p^* $. The remaining data of $\varphi$ should give $F$, but I cannot see how.

If $S$ is also an ordinary manifold, then $F$ can be defined point-wisely. We can express $\varphi$ point-wisely by $\varphi(p)=(f(p),V_{f(p)})$ for each $p\in S$, where $f(p)\in X$ and $V_{f(p)}\in T_{f(p)}X$. For any $a\in C^\infty(X)$, $F(a)\in C^\infty(S)$ is then the function that sends $p\in S$ to $V_{f(p)}a\in \mathbb{R}$.

However, if $S$ is a supermanifold, then the point-wise definition no longer works. I tried to formulate the point-wise definition in a non-point-wise way but failed.

I guess that $F$ should be a composition $C^\infty(X)\xrightarrow{D}C^\infty(TX)\xrightarrow{\varphi^*}C^\infty(S)$ for some natural derivation $D:C^\infty(X)\to C^\infty(TX)$ with respect to $p^*$, but I cannot see any derivation that arises "naturally".

Thanks in advance for any help.


1 Answer 1


The morphism $$\def\T{{\rm T}} φ:S→\T X$$ can be identified with the homomorphism of algebras $$\def\Ci{{\rm C}^∞} \Ci(\T X)→\Ci(S).$$ The algebra $\Ci(\T X)$ can be identified with the $\Ci$-symmetric algebra $$\def\CiSym{\mathop{\rm\Ci Sym}\nolimits} \CiSym_{\Ci(X)}(\Gamma(\T^*X))$$ of the $\Ci(X)$-module $\Gamma(\T^* X)$ of smooth sections of the cotangent bundle of $X$. Here $\Ci$-symmetric algebras are defined using exactly the same universal property as symmetric algebras, but working in the category of $\Ci$-rings instead of the category of commutative real algebras.

By the universal property of $\Ci$-symmetric algebras, the homomorphism $$\CiSym(\Gamma(\T^*X))→\Ci(S)$$ can be identified with a homomorphism of $\Ci$-algebras (equivalently, commutative real algebras) $$f:\Ci(X)→\Ci(S)$$ together with a morphism of $\Ci(X)$-modules $$Ψ:\Gamma(\T^*X)→\Ci(S).$$ The latter morphism can be identified with a derivation $$F:\Ci(X)→\Ci(S)$$ with respect to $f$: we set $\def\d{{\rm d}} F(g)=Ψ(\d g)$, where $\d g∈Γ(\T^* X)$ is the differential of $g$.

This approach works equally well for ordinary manifolds, supermanifolds, ${\bf Z}$-graded manifolds, derived manifolds, etc.

  • $\begingroup$ Thanks. Is it true that this $\Psi$ is the composition $\Gamma(T^*X)\hookrightarrow C^\infty(TX)\xrightarrow{\varphi^*} C^\infty(S)$, where $\Gamma(T^*X)\hookrightarrow C^\infty(TX)$ sends a global section $s\in \Gamma(T^*X)$, which is a $C^\infty(X)$-linear map $s:\Gamma(TX)\to C^\infty(X)$, to the smooth map on $TX$ that sends $(p,v_p)\in TX$ to $s(V)(p)\in \mathbb{R}$ for an arbitrarily chosen $V\in \Gamma(TX)$ with $V_p=v_p$? To be honest, I know nothing about $C^\infty$-rings, and the introduction on nlab is not quite clear to me. $\endgroup$
    – Shana
    Apr 14 at 6:02
  • $\begingroup$ Also, if my construction of $\Psi$ is correct, then I get one side of the one-to-one correspondence. But how can I tell the other side, without taking the universal property mentioned above for granted? Or equivalently, how can I show that the universal property is satisfied by $C^\infty(TX)$? I'd appreciate that if you could please give me some detailed materials about this fact. $\endgroup$
    – Shana
    Apr 14 at 6:30
  • 1
    $\begingroup$ @Shana: Yes, the description of Ψ is correct. Concerning the universal property, it is the fiberwise version of Proposition 1.1 of Models for smooth infinitesimal analysis by Moerdijk and Reyes. The first sections of this book offer a good introduction to C^∞-rings. $\endgroup$ Apr 14 at 15:44
  • $\begingroup$ Thanks a lot. I'm crystal clear now. $\endgroup$
    – Shana
    Apr 16 at 11:55

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.