Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have some questions on formal group laws and their logarithms:

  1. Let $R$ be a graded commutative ring and $F \in R[[X, Y]]$ be a formal group law over $R$ that admits a logarithm. Can you tell me, why this logarithm is unique if $\mathbb{Q} \subset R$ and is unique for $R$ of characteristic $p$ if in addition we demand that the logarithm shall be a power series of the form $x+ a_1x+a_2x^2 +... $ with $a_{p^i-1}=0$ for all $ i \geq 1$? \
  2. If $R$ has characteristic $2$ and and $F$ is a graded formal group law fulfilling $F(x, x)=0$ why does there always exist a logarithm for $F$?
share|cite|improve this question
Aaron has answered your question - but for a relatively self-contained read on formal group laws, Doug Ravenel has an Appendix to his complex cobordism book available online – Juan S May 20 '12 at 23:12
In case you're interested in the general classification of formal groups, you might be interested to check out these notes I made a while back: – Aaron Mazel-Gee May 22 '12 at 15:33
up vote 4 down vote accepted

1a. If $\mathbb{Q} \subset R$, then you can construct a logarithm for any $F$ by explicit formula $l(x) = \int (1/F_y(x,0))dx$. Here, $$F_y(x,0) = \left. \frac{\partial F(x,y)}{\partial y} \right|_{y=0},$$ and by the fraction I mean the power series obtained from the relation $1/(1-t)=1+t+t^2+\cdots$. (Note that $F_y(x,0)=1+...$.) You need denominators to integrate, of course: $\int x^n dx = x^{n+1}/(n+1)$. Given the existence of a logarithm, you can check that it is unique by using the same technique as in my answer below: two distinct logarithms would give a nontrivial automorphism of $(\mathbb{G}_a)_R$, but any $\mathbb{Q}$-algebra has no (additive) torsion elements, so if $a_i(x+y)^i=a_i(x^i+y^i)$ (for $i>1$) then $a_i=0$.

1b. This should follow from the same sort of argument. These are good practice to work out yourself rather than watching somebody else do, so I'll leave it for you. Hint: Think about my counterexample below.

2) If you write $F(x,y)=\sum a_{ij}x^iy^j$, then the condition $F(x,x)=0$ is equivalent to saying that $\sum_{i+j=n}a_{ij}=0$. On the other hand, note that commutativity implies that $a_{ij}=a_{ji}$, so when $n$ is odd this is automatically satisfied and when $n=2m$ is even it implies that $a_{m,m}=0$. I'm not sure what you mean by a "graded formal group law", but hopefully it will imply that the above formula can be applied here (i.e., only even powers of $x$ show up in $F_y(x,0)$, so that we only need odd denominators, which necessarily exist: if $2=0$ then $2k+1=1$!).


below are answers to old questions; the question has been edited

  1. This is actually not true. If $l_1,l_2:F \rightarrow \mathbb{G}_a$ are two logarithms, then $l_1 \circ l_2^{-1}$ (compositional inverse) is an automorphism of $\mathbb{G}_a$. So the statement is equivalent to saying that these must be trivial. Suppose $l(z)=\sum_{i\geq 1}a_i z^i$ is such an automorphism. Then $l(x+y)=l(x)+l(y)$, so $\sum a_i(x+y)^i = \sum a_i (x^i+y^i)$. So if for example $\mbox{char }R=p$, then we can take $a_{p^j} \not= 0$.

  2. Sorry, your second question seems to have a typo...? If and when you edit it, comment on this question so I'll know to come back and answer.

share|cite|improve this answer
Thank you very much! In fact I have already made a mistake in my first question, which I now have fixed. – nick May 20 '12 at 13:24

Here is another answer to 1a, per a document of Neil Strickland.

Let $f(x),g(x) \in R[x]$ be strict isomorphisms $F_0 \to F_A$, and let $g^{-1}(x)$ denote the reverse of $g$. Define $k(x)=f(g^{-1}(x))$. Then $k(x+y) = fg^{-1}(x+y) = f(g^{-1}(x)+_{F_0} g^{-1}(y)) = fg^{-1}(x)+fg^{-1}(y) = k(x)+k(y)$.

Expand this series out and use the fact that binomial coefficients are invertible in $\mathbb{Q}$ and hence $R$. This will give $k(x)=\lambda x$, but since $f$ and $g$ are strict isomorphisms (i.e. $f'(0)=g'(0)=1)$ we can conclude that $\lambda=1$ and so $f$ and $g$ are isomorphisms.

share|cite|improve this answer
I think this is just a compactified version of my argument (although I omitted any mention of "strictness"). – Aaron Mazel-Gee May 21 '12 at 21:52
@Aaron: Dear Aaron, you may be correct! – Juan S May 27 '12 at 6:27

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.