Problem
AA-1. Prove Proposition
.
Problem
AA-2. Represent the 1-unary algebra
explicitly as a subdirect product of SI algebras, where
has diagram
Problem
AA-3.
Let
be a distributive lattice and let
.
Define
by
and likewise
by
. As you know, these are lattice homomorphisms.
(a) Show that
. (Make a one-line proof based on the absorption law for lattices.)
(b) What embedding does (a) give?
(c) Show that the only SI distributive lattice is
2
. (Thus this
fact is very elementary. The subdirection representation theorem then
says that every distributive lattice is a subdirect product of copies
of
2
, a deeper fact that depends on the Axiom of Choice.)
Problem
AA-4. Say how to represent the group
as a subdirect product of subdirectly irreducible groups, using
as few factors as possible, by referring to the diagram of its
normal subgroups.
Problem AA-5. (a) Which finite abelian groups are SI? (Use any facts you know about finite abelian groups and their subgroup diagrams. An SI abelian group has a smallest proper subgroup.)
(b) Find all SI abelian groups, finite and infinite. (They can be described as subgroups of the circle group--the multiplicative group of all complex numbers of absolute value 1.)
Problem
AA-6. (a) Show that an SI 1-unary algebra has no ``fork'',
i.e., distinct elements
with
.
(Method: Let
denote the subalgebra generated by
, and similarly for
. For a subalgebra
of
let
mean the congruence relation obtained by
collapsing
to a point. Show that
con
if
give
a fork. You may use the fact that
con
is obtained by
first identifying
with
for each
and then seeing what equivalence relation that generates.)
(b) Using (a), try to find all finite SI 1-unary algebras whose diagram is connected.
(A useful observation: In an
-cycle, you get exactly the same
congruences as for the abelian group
Z
, so the congruence
lattice of an
-cycle is isomorphic to
Subgroup
Z
.)
Problem AA-7. Show that the finite SI 1-unary algebras are
(i) The algebra consisting of two fixed points,
(ii) the ``cyclic'' 1-unary algebras
of prime
power order (with
),
(iii) the algebras
where
and
,
for
.
(iv) the two-component algebras where one component is a fixed point and the other is of kind (ii).
(In (ii), it is handy to make this observation, which you may justify
very briefly: The congruence relations on an
-cycle regarded as a
1-unary algebra are exactly the same as those on the cycle regarded
as the group or ring
Z
. In all parts, you may justify
briefly why these are SI; it is most important to explain why
any finite SI must be of one of these forms.)
Problem
AA-8. Consider
the ring
Z
Z
, the ``direct sum'' of countably many copies of
the ring
Z
, or in other words, the subring
of
Z
Z
consisting of the
sequences that have only finitely many nonzero entries.
(a) Index the direct sum using
.
Show that the ideals of
correspond to subsets of
.
(b) Show that if
, then at least
one of
and
is isomorphic to
. (Method:
would be the internal direct sum of corresponding ideals
, so that
and
.)
(c) Show that
is not the direct product of directly
indecomposable algebras. (Use a cardinality argument.)
Problem
AA-9. (a) Show that direct-product decompositions of a
commutative ring with 1 into two factors correspond to
idempotents (elements
with
).
(b) Let
be the ring of all
-indexed sequences of
zeros and ones that are ``eventually constant'', with sequences
added and multiplied using the operations of
as a
ring. Find all direct-product decompositions of
.
(c) In (b), does
have a direct decomposition into directly
indecomposable factors? (Why or why not?)
(d) What about the Boolean algebra
Pow
for
countably infinite
?
Problem
AA-10. Suppose that
is a finite algebra. An
interesting question is whether
Var
contains finite SI
algebras larger than
, or even contains an infinite SI
algebra. If
is a lattice, for example, there are no larger
SI's; if
is a nonabelian
-group, the answer is that
there are arbitrarily large finite SI's and also infinite ones.
An easy case:
(a) Show that Shallon's algebra is SI, and in fact is simple.
(Method: Think about
con
for different possible distinct
elements
.)
More generally, Let
be the graph algebra based on
a graph like Shallon's but with
nodes, so that
has
elements and Shallon's algebra is
. Show
that
is SI.
(b) Show that
Var
. (Suggestion:
Write
.
Inside
, let
be the subalgebra generated
by elements whose entries are
's (zero or more),
then one
, and then the rest
's. Let
on
be the equivalence relation obtained by identifying
all elements of
that have an entry of 0 and letting
other blocks be singletons. Show that
is a congruence
relation on
. Then
.)
(c) Can you find an infinite SI in
Var
?