Endomorphisms of R ( (X) ) Are Essentially
Endomorphisms of R ( (X) ) Are Essentially
Endomorphisms of R ( (X) ) Are Essentially
Mathematics
83
84 P. EAKIN AND A. SATHAYE
Let IC(R) denote the set of all aER such that there is an
R -homomorphism σ: ^[[X!]]-*!? with σ(Xι)= a.
Using the "essential continuity" we establish that IC(R) is an ideal
of R contained in the Jacobson radical of R and containing the
nil-radical of R. (Theorem E)
Once IC(R) is shown to be an ideal it is easy to show that IC(R) is
nothing but the union of all ideals 31 of R such that R is complete in the
Sl-adic topology. This fact is indeed the reason for the suffix " c " in
IC(R). This fact also answers some questions raised by Gilmer; see
remarks at the end.
l
γ(Σ/ι ( Z') = Σσ*(Λ l )Z where hx G R[[X]].
and/,EYi?[[Y]].
THEOREM D. Let
Ix = {a E R I there exists an R-automorphism σ: R[[X\]]-+ R[[Xι\]
with σ(Xί) = X1 + a}
I2 = {a E R I there exists an R-homomorphism σ: R[[X]]-> R[[Y]]
where X, Y are finite sets of indeterminates over R such that σ(Xt) = a + f
for some Xf E X and fE(YR[[Y]])}.
Then Ic = Iλ = I2.
T ( Z ) = Γ X + JCY.
REMARKS. (1) The fact that Ic is an ideal shows that Theorem 3.4
of [1] is true with no restriction on the element " r " . Thus the
conjecture which follows that theorem is false.
(2) In his review of [5] (MR47 # 8532) Gilmer suggests a program
for simplifying some of the proofs. This would rest on whether a ring R
is a complete Hausdorff space in its (au , αn)-adic topology, if it is a
complete Hausdorff space in its (α,-)-adic topology for each /. However,
it is easy to give an example where this does not hold. For Gilmer's
example in [1] is a ring R and an element a such that R is complete, but
not Hausdorff in its (α)-adic topology. On the other hand, by Theorem
D there is an automorphism of JR [[X]] which takes X to X 4 a. Since
JR[[X]] is a complete Hausdorff space in its X-adic topology, it is also a
complete Hausdorff space in its ( X + α)-adic topology. However, since
R is not Hausdorff in its (α)-adic topology, neither is R [[X]]. So, since
α E ( X , X + α)jR[[X]] we see that R[[X]] is not Hausdorff in its
(X, X 4- α)-adic topology.
(3) Ic may be properly contained in the Jacobson radical of R and
it may properly contain the nil-radical of R. For example if Rf =
Z/4[X], M = (2,X)R' and R = R$[Y]]. Then the nil-radical of R is
21?, Ic in this case is (2, Y) and the Jacobson radical is (2, Y, X).
(4) It would be nice to have an intrinsic characterization of the
ideal Ic since it allows us to utilize the form of Nakayama's lemma for
complete local rings, namely
The proof would be the same as in the complete local ring case [8,
Th. 7, p. 259].
K-ENDOMORPHISMS OF I?[[X]] ARE ESSENTIALLY CONTINUOUS 87
REFERENCES
8. O. Zariski and P. Samuel, Commutative Algebra, volume II, Van Nostrand, Princeton, 1960.
UNIVERSITY OF KENTUCKY
Pacific Journal of Mathematics
Vol. 66, No. 1 November, 1976