is this theory consistent?
in [Logic]
|
From: Rupert on 15 Dec 2009 00:33 On p. 218 of "Naturalism in Mathematics" Penelope Maddy writes "Consider the theory ZFC+V=L+there is a transitive model of 'ZFC+0# exists' " Is there an upper bound for the consistency strength of this? If there is a measurable cardinal then there must be a transitive model of ZFC +0# exists in L[U] where L[U] is the inner model with one measurable cardinal, but it doesn't seem obvious that there must be such a model in L.
From: David Libert on 15 Dec 2009 02:01 Rupert (rupertmccallum(a)yahoo.com) writes: > On p. 218 of "Naturalism in Mathematics" Penelope Maddy writes > > "Consider the theory > > ZFC+V=L+there is a transitive model of 'ZFC+0# exists' " > > Is there an upper bound for the consistency strength of this? If there > is a measurable cardinal then there must be a transitive model of ZFC > +0# exists in L[U] where L[U] is the inner model with one measurable > cardinal, but it doesn't seem obvious that there must be such a model > in L. This theory is equiconsistent with ZFC + there is a transitive model of ZFC + 0# exists. Ie, the V=L part does not affect consistency strength in this case. To see better what this consistency stength is, it is strictly above ZFC + Con(ZFC + 0# exists). In fact it is strictly above all reasonable iterations of Con^alpha(ZFC + 0#) exists for small concrete ordinal notations. But it is strictly below the consistency strength of ZFC + 0# exists and there is an inaccessible cardinal. So it is well below the consistency strength of ZFC + there is a measurable cardinal. To see the opening claimed equiconsitency of adding V=L, suppose we are in ZFC + there is a transitive model of 'ZFC + 0# exists' . So by downward Lowenhiem Skolem there is countrable elementary submodel of that transitive model. So this model is well-founded, being a submodel of a transitive model. We can now pull back the universe of this model isomorphically to a model with underlying universe the real omega. So this model's interpretation of epsilon is a subset of omega, ie a real. So the property we have our universe is there is a real interpretting epsilon over omega making a well-founded model of ZFC + 0# exists. To say the model is well-founded (we have AC) is equivalent to saying there is no omega sequence in our universe making an infinite descending sequence through the epsilon interpetation. So the existence of an epsilon interpeation on omega making a well founded model is equiocalent to exists a real (the epsilon interpretaion) such that for all reals (omega sequences) the omega sequence is not infinite descending through the epsilon interpretation. This amounts to (exists a real) (for all reals) a property expressible with wuantification over omega. So this is a Sigma^1_2 property in the universe. So by the Shoenfield Absoluteness Thoerem, since this holds in our possible large universe, it also holds in L of that universe. So in L, that statement true in L gives a real in L making a well founded model of the theory. Yes, I should hae said, we out into that sentence that the resulting model also models 'ZFC + 0# exists', but that is just number quantification to say that, so still stays in Sigma^1_2 , So the L inside our universe is a model of ZFC + V=L + there is a transitive model of ZFC + 0# exists. Some other interesting points about this. This got a countable well-founded model of ZFC + 0# exists into L. Suppose in our outer working universe we assume more: we assume there is an uncountable transtivie model of ZFC + 0# exists. We can't use the extra strength of this assumption to get a similar uncountable well-founded model of ZFC + 0# exists in L. Because if an uncountable well founded model ZFC believes a real is 0#, then it really is 0#. This from the theory of 0#. So in L, if inside there was an uncountable well-founded model inside of ZFC + 0# exists, then that real translating back to transitive would really be 0#, so we would have 0# in L. Which is impossible, by the theory of 0#. -- David Libert ah170(a)FreeNet.Carleton.CA
From: Rupert on 15 Dec 2009 02:20 On Dec 15, 6:01 pm, ah...(a)FreeNet.Carleton.CA (David Libert) wrote: > Rupert (rupertmccal...(a)yahoo.com) writes: > > On p. 218 of "Naturalism in Mathematics" Penelope Maddy writes > > > "Consider the theory > > > ZFC+V=L+there is a transitive model of 'ZFC+0# exists' " > > > Is there an upper bound for the consistency strength of this? If there > > is a measurable cardinal then there must be a transitive model of ZFC > > +0# exists in L[U] where L[U] is the inner model with one measurable > > cardinal, but it doesn't seem obvious that there must be such a model > > in L. > > This theory is equiconsistent with ZFC + there is a transitive model of > ZFC + 0# exists. > > Ie, the V=L part does not affect consistency strength in this case. > > To see better what this consistency stength is, it is strictly above > ZFC + Con(ZFC + 0# exists). > > In fact it is strictly above all reasonable iterations of > Con^alpha(ZFC + 0#) exists for small concrete ordinal notations. > > But it is strictly below the consistency strength of > ZFC + 0# exists and there is an inaccessible cardinal. > > So it is well below the consistency strength of ZFC + there is a measurable > cardinal. > > To see the opening claimed equiconsitency of adding V=L, suppose we are > in ZFC + there is a transitive model of 'ZFC + 0# exists' . > > So by downward Lowenhiem Skolem there is countrable elementary submodel > of that transitive model. So this model is well-founded, being a submodel > of a transitive model. > > We can now pull back the universe of this model isomorphically to a model > with underlying universe the real omega. So this model's interpretation of > epsilon is a subset of omega, ie a real. > > So the property we have our universe is there is a real interpretting > epsilon over omega making a well-founded model of > ZFC + 0# exists. > > To say the model is well-founded (we have AC) is equivalent to saying > there is no omega sequence in our universe making an infinite descending > sequence through the epsilon interpetation. > > So the existence of an epsilon interpeation on omega making a well > founded model is equiocalent to exists a real (the epsilon interpretaion) > such that for all reals (omega sequences) the omega sequence is not > infinite descending through the epsilon interpretation. > > This amounts to (exists a real) (for all reals) a property expressible > with wuantification over omega. > > So this is a Sigma^1_2 property in the universe. > > So by the Shoenfield Absoluteness Thoerem, since this holds in our > possible large universe, it also holds in L of that universe. > Yes, fantastic, Shoenfield Absoluteness Theorem. I think I must have forgotten the correct statement of that. > So in L, that statement true in L gives a real in L making > a well founded model of the theory. > > Yes, I should hae said, we out into that sentence that the resulting > model also models 'ZFC + 0# exists', but that is just number > quantification to say that, so still stays in Sigma^1_2 , > > So the L inside our universe is a model of > ZFC + V=L + there is a transitive model of ZFC + 0# exists. > > Some other interesting points about this. > > This got a countable well-founded model of ZFC + 0# exists into L. > > Suppose in our outer working universe we assume more: we assume there > is an uncountable transtivie model of ZFC + 0# exists. > > We can't use the extra strength of this assumption to get a similar > uncountable well-founded model of ZFC + 0# exists in L. > > Because if an uncountable well founded model ZFC believes a real is > 0#, then it really is 0#. This from the theory of 0#. > > So in L, if inside there was an uncountable well-founded model > inside of ZFC + 0# exists, then that real translating back to > transitive would really be 0#, so we would have 0# in L. > > Which is impossible, by the theory of 0#. > Yes, that's right, Maddy makes this observation. Thanks a lot.
|
Pages: 1 Prev: Does inductive reasoning lead to knowledge? Next: MADNESS IN EINSTEINIANA |