For a $k$-Hopf algebra $H$ and element $h \in H$ is called grouplike is $\Delta(h) = h \otimes h$ and $\epsilon(h)=1_k$ ($\epsilon$ is the counit). The identity $1_H$ is clearly grouplike, but in general non-trivial grouplike elements may fail to exist. See this question for Is there a name for a Hopf algebra whose only grouplike element if the identity?
-
2$\begingroup$ @Martin Brandeburg: In fact, a grouplike element cannot be zero by definition. $\endgroup$– Salvatore SicilianoCommented Jul 31, 2021 at 19:39
-
1$\begingroup$ Alright. Also, ncatlab.org/nlab/show/grouplike+element demands that $\varepsilon(h)=1$. $\endgroup$– Martin BrandenburgCommented Jul 31, 2021 at 19:42
-
2$\begingroup$ @MartinBrandenburg I added the required condition. $\endgroup$– Duchamp Gérard H. E.Commented Aug 3, 2021 at 23:38
-
2$\begingroup$ You have an incomplete sentence "See this question for ...". What do you have in mind ? $\endgroup$– Duchamp Gérard H. E.Commented Aug 3, 2021 at 23:39
1 Answer
There is a one-to-one correspondence between the grouplike elements and the simple, $1$-dim subcoalgebras. So if the only grouplike element is $1_H$, then there is a unique $1$-dim simple subcoalgebra (which is $k\cdot 1_H$). In that case, the HA is - by definition- called: connected HA.
Also, notice that a connected HA is the same thing as an irreducible HA.
(Although if we confine ourselves at the level of coalgebras, the connected ones are a subset of the irreducibles. Actually, at the level of coalgebras, connected=pointed+irreducible, but i think that this is another story).
Edit: Motivated by the OP's comment, regarding the use of the term connected under the presence of an algebra grading, i think it would be useful to mention the following: the term connected is used in the encyclopedia link provided (https://encyclopediaofmath.org/wiki/Hopf_algebra) in the sense of a connected graded algebra. The way i am using the term connected HA in this post is different and is in the sense of a connected coalgebra. These are two different notions of connectedness with non-trivial interactions between them. I think that in contemporary literature the second use tends to be more "standard". For more details, i believe it would be interesting to take a look at: arXiv:1601.06687v1 [math.RA]
Edit 2: In my understanding, the question essentially provides another way to view the notion of a connected coalgebra as dual to the notion of a connected algebra (in the ungraded setting):
Since a connected algebra (ring) -in the ungraded setting- is one which has no non-trivial idempotents (other than $0$ and $1$) in the same way a connected coalgebra is one which has no non-trivial grouplikes (other than $1$).
Here, we are essentialy viewing the notion of an idempotent ($g\cdot g=g$) element of an algebra as dual to the notion of a grouplike ($\Delta(g)=g\otimes g$) element of a coalgebra.
-
$\begingroup$ I thought that connected required the existence of an algebra grading on the HA: encyclopediaofmath.org/wiki/Hopf_algebra $\endgroup$ Commented Aug 1, 2021 at 14:24
-
$\begingroup$ @Spyros, the notion of connected HA does not necessarily require the HA to be graded. I am on my phone right now and it is difficult to list references, but here is a quick one: arXiv:2004.11996 [math.RA] $\endgroup$ Commented Aug 1, 2021 at 18:24
-
$\begingroup$ It is true that during the first decades of the developments of the HA theory, the definitions were often assuming that we were talking about graded objects. This stems from the historical development of the topic and is still reflected in some modern references (especially those coming from the algebraic topology and the cohomology point of view). However, i think that most contemporary textbooks (i.e. those developed after the 80's), including Sweedler's textbook, have dropped this requirement and view graded HAs as a separate topic on its own. $\endgroup$ Commented Aug 1, 2021 at 21:21
-
$\begingroup$ Great, thanks for the explanation. $\endgroup$ Commented Aug 2, 2021 at 17:48