I am interested in elementary embeddings from a model of set theory into itself. One way of producing such elementary embeddings is when the model is generated by indiscernibles; this idea is very closely related to the existence of sharps. Jech and Kanamori discuss $0^{\#}$ and $0^\dagger$ in detail but don't tell me much about other sharps. More advanced resources are difficult to understand without a lot of background knowledge.


Let $\theta$ be an inaccessible cardinal, and suppose that some set $A$ of measurable cardinals below $\theta$ is a stationary subset of $\theta$. For each $\kappa \in A$, let $\mu_\kappa$ be a normal measure on $\kappa$, and let $\mathcal{U} = \{ \langle \kappa, \mu_\kappa \rangle : \kappa \in A \}$. Let $L[\mathcal{U}]_\theta$ denote those elements of $L[\mathcal{U}]$ of rank less than $\theta$.

Question statement

Do there exist large cardinal assumptions which imply the existence of a closed unbounded set of ordinal indiscernibles for $L[\mathcal{U}]_{\theta}$ such that every order-preserving map of these indiscernibles extends to an elementary embedding $j:L[\mathcal{U}]_{\theta} \to L[\mathcal{U}]_{\theta} \, \, ?$


The large cardinal assumptions may be on $\theta$, the elements of $A$, or some other large cardinal. The values of $\theta$, $A$, and the $\mu_\kappa$ may be chosen in whatever way you like subject to the hypotheses above -- I just want this to work in some example, not in every example.

In The Core Model, Dodd mentions double mice, a generalization of $0^\dagger$. Maybe some version of these can be used to answer the question affirmatively, but I know nothing about them.

  • $\begingroup$ It seems to me that a measurable cardinal above $\theta$ would give you indiscernibles of the sort you want, by the same argument that gives you $0^\#$ from one measurable cardinal. And my assumption of a measurable cardinal above $\theta$ is probably overkill; a Ramsey cardinal above $\theta$ should suffice, just as for $0^\#$. $\endgroup$ Commented Oct 9, 2011 at 23:46
  • $\begingroup$ Thanks, Andreas. What is the intuition behind why this should work? Of course we can get indiscernibles from that assumption, but why should they be a club, and why should they generate the model? If I replaced $\mathcal{U}$ with an arbitrary subset of $V_\theta$, would it still work? $\endgroup$ Commented Oct 12, 2011 at 15:20
  • $\begingroup$ My intuition here is only that I see no obstruction to transposing the usual proof for $0^\#$ into the new context. Note that you can't expect just any old class of indiscernibles to be a club, or any old Ehrenfeucht-Mostowski theory of indiscernibles to be satisfiable by a club. Even in the theory of $0^\#$, Silver needed to get the right E-M theory by taking a class of indiscernibles whose $\omega$-th element is as small as possible. $\endgroup$ Commented Oct 24, 2011 at 21:29

1 Answer 1


My reading of this question was different from Andreas', because Norman asked for order preserving maps of the indiscernibles to extend to embeddings $j:L[U]_\theta \rightarrow$ $L[U] _\theta$

i.e. the indiscernibles should be below $\theta$ as the ordinal height of the structures mentioned is $\theta$?

In that case the measurable or Ramsey above $\theta$ only guarantees indiscernibles above $\theta$ and so "missing the target"?

In any case there are generalisations of "double mice" that you surmise that provide a positive answer. Let $M$ be the "least" in a certain canonical well-ordering of all such iterable structure that have a measurable cardinal $\kappa$ which is, in $M$, the limit of measurables cardinals below $\kappa$. (Such structures are called "mice".) By Loewenheim Skolem we may assume that $M$ is countable. `Iterability' here means we may form all iterated ultrapowers of the first structure $M=M_0$ using this top measure repeatedly; call the ultrapower structures $M_\tau$ for all ordinal $\tau$; then all the $M_\tau$ will be wellfounded. In the $\tau$'th model let $\kappa_\tau$ be the top-most measurable cardinal. Then $\tau ,\mu \rightarrow \kappa_\tau<\kappa_\mu$ and the class of all such $\kappa_\tau$ forms a closed unbounded class of indiscernibles for the model $W$ "left behind" by these iterates.

[Fornally $W = \bigcup _ \tau H(\kappa_\tau)^{M_\tau}$.]

Then $W$ is the `least' inner model with a proper class of measurables cardinals, (because as $\tau$ increases the order type of the measurables in the models $M_\tau$ increases; but these measures are left behind in the lower $H(\kappa_\tau)^{M_\tau}$ part of the model that goes into $W$). The $\kappa_\tau$ are indiscernibles for it, just as the Silver indiscernibles are for $L$.

The large cardinal assumption is then that "This iterable $M$ exists" just as a parallel to "$0$-sharp exists" is to $L$.

  • $\begingroup$ You're right; I misread the question. $\endgroup$ Commented Feb 17, 2012 at 0:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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