It's a classical fact that the commutative power series ring $\mathbb{Z}_p[[x_1,x_2]]$ is isomorphic to the completed group algebra $\mathbb{Z}_p[[\mathbb{Z}_p\times\mathbb{Z}_p]]$, the isomorphism sending generators $a_1,a_2$ of $\mathbb{Z}_p\times\mathbb{Z}_p$ to $1+x_1,1+x_2$.

In $\mathbb{Z}_p[[x_1,x_2]]$, we have the coset $1+(x_1,x_2)$ of the ideal $(x_1,x_2)$, which is a closed and open subgroup of its group of units, which certainly contain $1+x_1,1+x_2$, and hence contains the closed subgroup $\langle 1+x_1,1+x_2\rangle$ generated by $1+x_1,1+x_2$, which is isomorphic to $\mathbb{Z}_p\times\mathbb{Z}_p$, but is strictly smaller than $1+(x_1,x_2)$.

Is it possible to describe the image of the closed subgroup $\langle 1+x_1,1+x_2\rangle$ in $\mathbb{Z}_p[[x_1,x_2]]$?

Ie, suppose we are given a power series $f\in 1+(x_1,x_2)\subset\mathbb{Z}_p[[x_1,x_2]]$. How can we determine if $f$ lies in the closed subgroup $\langle 1+x_1,1+x_2\rangle$?

(I know this question isn't really well-defined. I'm just looking for certain criteria, ideally linked to algebraic properties of the power series, which can be used to identify the image of $\langle 1+x_1,1+x_2\rangle$).

It would also be nice if one could describe the quotient $(1+( x_1,x_2))/\langle 1+x_1,1+x_2\rangle$.

  • $\begingroup$ Are you perhaps using the same notation $\langle,\rangle$ for both ideals (in $1+\langle x_1,x_2\rangle$) and subgroups (in $\langle 1+x_1, 1+x_2\rangle$)? My apologies if I misunderstood. $\endgroup$ – Yaakov Baruch Aug 8 '17 at 9:15
  • $\begingroup$ @YaakovBaruch Yes, that's right. Perhaps that is confusing. I've now replaced the ideal $\langle x_1,x_2\rangle$ with "$(x_1,x_2)$". $\endgroup$ – stupid_question_bot Aug 8 '17 at 15:50

The usual power series for $\log(1+x)$ determines an injection from $1+(x_1,x_2)\subset\mathbb{Z}_p[[x_1,x_2]]$ into $\mathbb{Q}_p[[x_1,x_2]]$. A power series $f\in 1+(x_1,x_2)$ is in $\langle 1+x_1,1+x_2\rangle$ if and only if there are $\alpha_1$, $\alpha_2\in\mathbb{Z}_p$ such that \begin{align*} \log f(x)&=\alpha_1\log(1+x_1)+\alpha_2\log(1+x_2)\\ &=\alpha_1\sum_{n\geq 1}\frac{(-1)^{n-1}}{n}x_1^n+\alpha_2\sum_{n\geq 1}\frac{(-1)^{n-1}}{n}x_2^n. \end{align*} Note that $\alpha_1$ and $\alpha_2$ can be read off from $\log f(x)$: they are the coefficients of $x_1$ and $x_2$, respectively. So if $$ \log f(x)=\sum_{i,j\geq 0}a_{ij}x_1^{i}x_2^{j}, $$ a necessary and sufficient condition for $f(x)\in\langle 1+x_1,1+x_2\rangle$ is that $a_{ij}=0$ if $i$, $j\geq 1$, and $$ a_{n,0}=\frac{(-1)^{n-1}}{n}a_{1,0}, $$ $$ a_{0,n}=\frac{(-1)^{n-1}}{n}a_{0,1}. $$


The group algebra $\mathbb{Z}_p[[\mathbb{Z}_p\times \mathbb{Z}_p]]$ has the structure of a complete Hopf algebra, where $\mathbb{Z}_p\times \mathbb{Z}_p$ consists of precisely the group-like elements. Via the given isomorphism, we get a comultiplication on $\mathbb{Z}_p[[x_1,x_2]]$, where $1+x_1$ and $1+x_2$ are group-like. The formula for the comultiplication is $\Delta x_i=x_i\otimes 1+1\otimes x_i+x_i\otimes x_i$ for $i\in\{1,2\}$. A power series $f\in 1+(x_1,x_2)$ is in $\langle 1+x_1,1+x_2\rangle$ if and only if $f$ is group-like, i.e. if and only if there is an equality of formal power series $$ f(x_1+y_1+x_1y_1,x_2+y_2+x_2y_2) = f(x_1,x_2)f(y_1,y_2)\in\mathbb{Z}_p[[x_1,x_2,y_1,y_2]]. $$

  • $\begingroup$ Interesting... this is definitely reminiscent of formal group laws... $\endgroup$ – stupid_question_bot Aug 9 '17 at 20:24

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.