# Reference Request: Differential Graded Hopf Algebras

To prepare an upcoming seminar talk I’m trying to find introductory texts on differential graded Hopf algebras (over a field $$k$$).

My knowledge of Hopf algebras encompasses roughly the first 5 chapters of Sweedler’s book. I also have a rough idea of how to define dg-Hopf algebras, namely by the same diagrams as an ordinary Hopf algebra but with dg-vector spaces instead of just vector spaces (i.e. as “Hopf objects” in the monoidal category of dg-vector spaces). I’m thus looking for references that do the following:

• Providing an explicit definition of dg-Hopf algebras and showing their fundamental properties.
• Giving (basic) examples for dg-Hopf algebras.
• Explaining the differences and similarities between ordinary Hopf-algebras and dg-Hopf algebras.
• (Giving me a way to check my thoughts and suspicions.)

I was originally given [1] and [2] as references (the last of which is also given on the nlab) but [1] only really deals with ordinary Hopf algebras and [2] doesn’t seem to provide introductory material. I already found [3], which provides some background on dg-algebras, -coalgebras and -Lie algebras, but only little on dg-bialgebras and nothing on dg-Hopf algebras.

[1]: Loday, Cyclic Homology, Appendix A
[2]: Quillen, Rational Homotopy Theory, Appendix B
[3]: Félix, Halperin, Thomas, Rational Homotopy Theory, Chapters 3,21

I cannot give you a reference better than "wikipedia definition of Hopf algebra in a braided category" and it specialization to the braiding $$v\otimes w\mapsto (-1)^{|v||w|}w\otimes v$$. But I can give you the first elementary and nontrivial example of d.g. Hopf algebra that is not a Hopf algebra in the usual sense:
Fix a field of characteristic not 2 and $$H=k[x]/x^2$$. Then $$H$$ is not a Hopf algebra, because it is commutative and has nontrivial nilpotent elements. On the other hand, it is a d.g. Hopf algebra with $$|1|=0$$, $$|x|=1$$ and $$d(x)=1$$, simply defining $$\Delta(x)=x\otimes 1 +1\otimes x$$
The reason is that, if you define $$\Delta$$ as above, the way you extend it to $$k[x]$$ in the d.g. setting is considering the algebra structure in $$k[x]\otimes k[x]$$ given by $$(x^n\otimes x^m)\cdot (x^i\otimes x^j)=(-1)^{im}x^{n+1}\otimes x^{n+j}$$ With this structure, the element $$x^2$$ mapsto $$(x\otimes 1+1\otimes x) ^2 =x^2\otimes 1+x\otimes x -x\otimes x+1\otimes x^2=x^2\otimes 1+1\otimes x^2$$ (and not to $$x^2\otimes 1+2x\otimes x+1\otimes x^2$$) So, the "super sign" makes the job so that the ideal $$(x^2)$$ is also a coideal, and hence $$\Delta:k[x]/x^2\to k[x]/x^2\otimes k[x]/x^2$$ is a well-defined algebra map, when considering $$k[x]/x^2\otimes k[x]/x^2$$ as algebra using the Koszul signs.
• I don’t see how $d(x) = 1$ is supposed to work: Then $1 \in k[x]/x^2$ is a boundary but $1 \in k$ is not, so the counit $\varepsilon \colon k[x]/x^2 \to k$ cannot be a morphism of dg-algebras. I think it should be $d(x) = 0$, so that we get the dg-symmetric algebra on a one-dimensional dg-vector space concentrated in degree 1 (if I understand this correctly). – Jendrik Stelzner May 18 at 10:10
You are right. $$\epsilon\circ d=0$$ for any coderivation d, in the same way as d(1)=0 for a derivation d. Fortunately, with dx=0 the example survives