Abstract. Supercompact extender based forcings are used to construct models with cardinal structure different from those of . In particular, a model where all regular uncountable cardinals are measurable in is constructed.
Date: July 28, 2016.
2010 Mathematics Subject Classification. Primary 03E35, 03E55.
Key words and phrases. large cardinals, extender based forcing, HOD, Easton iteration.
The work of the first author was partially supported by ISF grant No.58/14.
In [3] the following result was proved:
Theorem. Suppose are cardinals such that , is inaccessible, and is a limit of -supercompact cardinals. Then there is a forcing poset that adds no bounded subsets of , and if is -generic then:
The supercompact extender based Prikry forcing, developed by the second author in [8], is applied to reduce largely the initial assumptions of this theorem and to give a simpler proof. Namely, we show the following:
Theorem 1. Suppose is a -supercompact cardinal1, and is an inaccessible cardinal above . Then there is a forcing poset that adds no bounded subsets of , and if is -generic then:
Actually, assuming the measurability (or supercompactness) of in , we obtain that is measurable (or supercompact) in .
In [2], a model with the property , for every infinite cardinal was constructed. We extend this result, using the supercompact extender based Magidor forcing of the second author [9], and show the following:
Theorem 22. Assume there is a Mitchell increasing sequence of extenders such that is measurable, and for each , , , and , where is the natural embedding. Then there is a model of ZFC where all regular uncountable cardinals are measurable in .
The work [1] obtained results similar to our last theorem using iteration of Radin forcing together with Cardinal collapsing.
This may be of some interest due to the following result of H. Woodin [10]:
Theorem (The HOD dichotomy theorem). Suppose is an extendible cardinal. Then exactly one of the following holds:
However, we do not have even inaccessibles in the model of theorem 2. It is possible to modify the construction in order to have measurable cardinals (and bit more) in the model. We do not know how to get supercompacts and it is very unlikely the method used will allow model with supercompacts.
The structure of this work is as follows. In section 2 we give definitions and claims about and homogeneous forcing notions which are well known. In section 3 we prove theorem 1. In section 4 we prove theorem 2.
We assume knowledge of large cardinals and forcing. In particular this work depends on the supercompact extender based Prikry-Magidor-Radin forcing.
Definition 2.1. Let be a class. The class contains the sets definable using ordinals and sets from , i.e., iff there is a formula , ordinals , and sets , such that .
The class contains the sets which are hereditarily in , i.e., iff .
We write and for and , respectively.
Note, if is a set of ordinals then .
We will work in of generic extensions, hence the relation between and , where is a generic extension, will be our main machinery.
Our main tool will be forcing notions which are homogeneous in some sense. A forcing notion is said to be cone homogeneous if for each pair of conditions there is a pair of conditions such that , , and .
A forcing notion is said to be weakly homogeneous if for each pair of conditions there is an automorphism so that and are compatible. It is evident a weakly homogeneous forcing notion is cone homogeneous.
An automorphism induces an automorphism on -terms by setting recursively .
Note ground model terms are fixed by automorphisms, i.e., , in particular for each ordinal , .
An essential fact about a cone homogeneous forcing notion is that for each formula , either or . If in addition the forcing is ordinal definable then we get , where is -generic.
In [4] it was shown that an arbitrary iteration of weakly (cone) homogeneous forcing notions is weakly (cone) homogeneous under the very mild assumption that the iterand is fixed by automorphisms. For the sake of completeness, we show here a special case of this theorem, which is enough for our purpose.
Theorem 2.2 (Special case of Dobrinen-Friedman [4]). Assume is a backward Easton iteration such that for each , and for each and automorphism , we have . Then is cone homogeneous.
Proof. Fix two conditions . We will construct two conditions and such that , by which we will be done. The construction is done by induction on as follows.
Assume , , , and were constructed. We know . Let be a function for which holds, whenever is generic and . If both and are the maximal element of then let and be the maximal element of and let be the trivial automorphism of . If either or is not the maximal element of then use the the cone homogeneity of to find -names , , and , such that , , and is an automorphism. Whatever way was constructed define the automorphism by letting , for each .
Assume is limit and for each we have , , and is an automorphism such that , whenever . For each let be the condition defined by setting for each , . □
The following claim is practically the successor case of the previous one. It is useful when we will have automorphism of forcing notions which are not necessarily cone homogeneous.
Claim 2.3. Assume and are forcing notions with being an isomorphism. Let be a -name of a cone homogeneous forcing notion such that , where .
Then for each pair and there are stronger conditions and such that .
Proof. Note there is a function taking -names to -names such that , where is generic and .
Set . By the cone homogeneity of in there are stronger conditions and , for which there is (a name of) an automorphism . Set . Since for generics as above we have we get is the required automorphism. □
While the forcing notions we will use are cone homogeneous we will deliberately break down some of their homogeneity. The relation between and will be as follows.
Claim 2.4. Assume is an ordinal definable cone homogeneous forcing notion. Let be a projection. Assume that for each condition , ordinals , and formula , if then . Then , where is the upward closure of .
Proof. Assume . Let be generic. Then in there are ordinals such that for each ,
Let be a maximal antichain such that for each ,
and for each ,
Let be a -name defined by setting for each .
Since is predense in we get , by which we are done. □
Let be the Cohen forcing for adding subsets to , i.e., . The following is well known.
Proof. Assume are conditions. Choose stronger conditions, and , such that . Define by setting for each . It is obvious is an automorphism. □
The following is immediate from the previous claim and theorem 2.2.
Let us switch to the cone-homogeneity of the Extender Based Prikry forcing. Extender based Prikry forcing was originally developed in [5]. We use a revision of the forcing where the extender can witness supercompactness. This was first developed in [8]. At the suggestion3 of the referee we add intuitive explanation of the extender based Prikry forcing. We will do so by introducing definitions going gradually from the Prikry forcing to the extender based Prikry forcing.
We begin with a definition of Prikry forcing which is more cumbersome than the standard definition. When generalizing to the extender base Prikry forcing this cumbersome definition becomes simpler than the standard definition of the extender based forcing.
Assume is an elementary embedding such that , , and is its sole generator. Define the measure by letting for each ,
Recall that a condition in Prikry forcing is of the form , where is a finite increasing sequence and . We can always assume that for each ordinal , . The condition is said to be a direct extension of the condition () in this forcing if . Let be an ordinal. Then . We say the condition is a -point extension of . By recursion define the -point extension of the condition to be . We say the condition is stronger than the condition if there are such that . This is clearly a valid definition of Prikry forcing. If is the generic object then letting we get that is an -sequence cofinal in .
Let us define again Prikry forcing, increasing the level of cumbersomeness. Starting from the same assumption as above proceed as follows. Define the measure by letting for each ,
Note a typical function is of the form . Define now a condition in Prikry forcing to be of the form , where is a function such that is a finite increasing sequence, and . Note we can assume for each , . The condition is said to be a direct extension of the condition () in this forcing is if . Assume is a condition and is a function. Define the function by letting . Define the set of functions in which are above as . A 1-point extension of is defined to be . By recursion define the -point extension of the condition to be . We say the condition is stronger than the condition if there is such that . This is clearly a valid, if somewhat bizarre, definition of Prikry forcing. If we let be the generic object and define the function by setting , then is an -sequence cofinal in .
Staying in the previous context, let us choose an ordinal . There is a function in the ground model such that . Then is a sequence Prikry generic for the measure generated by . We can, however, generalize somewhat the forcing so as to be able to read directly the generic sequence corresponding to from the generic object. For this define the measure by letting for each ,
Note a typical function is of the form such that . Define now a condition in Prikry forcing to be of the form , where is a function such that both and are finite increasing sequences, and . Note we can assume for each , and . The condition is a direct extension of the condition () in this forcing if . Assume is a condition and is a function. Define the function by setting
Set . Then define the condition to be . We say the condition is a -point extension of . By recursion define the -point extension of the condition to be . We say the condition is stronger than the condition () if there is such that . The above is clearly a valid, if strange, definition of Prikry forcing. Letting be the generic object we define the function by setting
Then both and are -sequences cofinal in .
Of course, seeing the above one can immediately generalize to any less than ordinals below , which still leaves us in the realm of Prikry forcing. Thus fix . It is technically useful to assume . Define the measure by letting for each ,
A typical function is of the form and for each such that we have . While it is not terribly important now, it should be observed that we also have . Define now a condition in Prikry forcing to be of the form , where is a function such that for each we have is increasing, and . Note if we can assume for that each we have . The condition is said to be a direct extension of the condition () in this forcing if . Assume is a condition and . The function is the function defined by setting for each ,
Set . Then define the condition to be . We say the condition is a -point extension of . By recursion define the -point extension of the condition to be . We say the condition is stronger than the condition () if there is such that . The above is still a valid definition of Prikry forcing. Letting be the generic object and defining the function by setting for each , , we get that for each , is an -sequence cofinal in . Note the function defined here is found in the generic extension of the standard Prikry forcing.
In fact, one can use in the previous definition also sets of size . The usefulness of the ultrafilter is more apparent in this case, when one views a typical function in a measure one set. Thus fix . It is technically useful to assume . Define the measure by letting for each ,
A typical function is of the form (note and not !), where , and for each such that we have . Moreover, . Define now a condition in Prikry forcing to be of the form , where is a function such that for each we have is a finite increasing sequence, and . We can always assume that for each we have for each The condition is said to be a direct extension of the condition () in this forcing if . Assume is a condition and is a function. Define the function by setting for each ,
Set . Then define the condition to be . We say the condition is a -point extension of . By recursion define the -point extension of the condition to be . We say the condition is stronger than the condition if there is such that . The above is still a valid definition of Prikry forcing. Letting be the generic object and defining the function by setting for each , , we get that for each , is an -sequence cofinal in . Defining the function as in the previous paragraph we get cofinal -sequences, but still all of them are generated by .
Since the ultrapower is closed only under -sequences, one cannot enlarge to be of size greater than while keeping the nice properties of Prikry type forcing notions. However, we can use conditions with different domains, thus adding sequence corresponding to each of the ordinals below . The domain change, however, causes the forcing to be non-isomorphic to Prikry forcing. Thus if then define the measure by letting for each ,
Define now a condition in the forcing to be of the form , where is a function such that for each , is a finite increasing sequence, , and . If and then set . The condition is said to be a direct extension of the condition () in this forcing if and . Note this definition of the direct order is a major change from all previous definitions. In fact the direct order is a Cohen forcing for adding subsets to . Assume is a condition and is a function. The function is defined by setting for each ,
Given a set set . Then define the condition to be . We say the condition is a -point extension of . By recursion define the -point extension of the condition to be . We say the condition is stronger than the condition () if there is such that . This forcing notion is no longer Prikry forcing. Using the same definition of as before we get cofinal -sequence into . However, while each of the sequences appears in a generic extension by Prikry forcing, the function itself does not belong to a Prikry generic extension.
The point of the previous forcing is that nothing restricts us from using it with elementary embeddings with many generators, thus we get the extender based Prikry forcing. Thus assume is an elementary embedding such that , , and is a cardinal in . Using the previous definition with conditions such that , we get that -many new -sequences cofinal in appear in the generic extension, thus . Working out the proof we get that no cardinals are collapsed, , and .
The final generalization achieved so far, along the lines above, is to begin with elementary embedding with even more closure properties, i.e., such that and , where and then work out the definition above to use of arbitrary size below . This yields a generic extension in which and blows up to whatever cardinal the model catches beginning with up to . However, the cardinals above and below are collapsed in this extension., which is to be expected when using a -supercompact cardinal.
Let be an extender as in [8]. Let be the extender based Prikry forcing derived from . We show is cone homogeneous.
Proof. Set . Set and . Choose a set so that both and are conditions. Define by setting for each , , where and . We claim is an isomorphism. Note the condition and are mostly identical. The condition can differ from when and differs from . Thus the one point which is not trivial is that is order preserving.
Assume . We show . We need to show there is such that . Choose such that . For the measure one sets we get at once
For the functions we have the following. If then
If then there is such that , thus , hence
□
For a generic filter define the function by setting .
Let us define the Easton products we are going to work with. Let be a set of ordinals. Let be the Easton product of the Cohen forcing notions yielding, in the generic extension, for each ,
When forcing with we will choose to be large enough so as not to interfere with our intended usage. Due to the (cone) homogeneity of , the sequences forced by are not in . We would like to break the homogeneity of so as to have the Prikry sequence enter . We will achieve this by coding the Prikry sequence into the power set function. We will want the Cohen forcing used to be stabilized by reasonable automorphisms of . Let be Prikry forcing using the measure . Define the function by setting , where . Note . Assume a pair of conditions are compatible in . I.e., there is a condition . Then , hence . Hence a maximal antichain in is also a maximal antichain in , thus the function is a projection. Thus if is generic then is -generic.
Until the end of the section set .
Claim 3.2. Assume are conditions such that and are compatible. Then there are extensions, and , such that .
Proof. Since and are compatible, we can choose conditions and such that . By claim 3.1 there are direct extensions and such that is an automorphism. Since , where is generic, we are done by claim 2.3. □
The following is immediate from the previous claim.
Proof. In order to show we will show a dense subset of conditions below forces . Let be an arbitrary condition. By claim 3.2 there is and such that . Thus . □
The previous corollary together with claim 2.4 yields the following.
We will get a special case of theorem 1 by invoking the last corollary in a model of the form .
Corollary 3.5. Assume , where is a set of ordinals, and is an extender witnessing is a -supercompact cardinal. There is a forcing notion preserving the extender such that in , where is -generic, , , and .
Proof. We will begin by defining the forcing notion so that for an -generic filter we will have .
Define by induction the forcing notions and sets , as follows. Set and . For each define as follows. In , where is generic over , let be the forcing notion for a large enough . Let be -generic over , i.e., is a code for . Set , where is an -name for . Let be the inverse limit of . Let be generic.
Invoking inside and calculating we get . For each , , thus . □
Hence we get:
Corollary 3.6. Assume is measurable and is -supercompact. Then there is a generic extension in which , and (of the generic extension) is measurable in .
In order to analyze , where , let us derive another line of corollaries stemming from claim 3.2. The problem we face when dealing with is an automorphism of might move , the name of . Thus we will need to fine tune the projection .
First we recall the notion of a good pair from [9]. We say the pair is a good pair if is a -internally approachable elementary substructure, , and there is a sequence such that witnesses the -internal approachability of , , is a -decreasing continuous sequence in , and for each , , , and .
Set . Define the function by setting for each , . Note . Fix two conditions . Assume they are compatible in , i.e., there is a condition such that . Thus there are and such that . Immediately we get . Hence a maximal antichain in is a maximal antichain in , hence the function is a projection.
Corollary 3.7. Assume is an elementary substructure such that is an -generic condition and is a good pair. Let be a -name such that . If , , and , then .
Proof. In order to show we will show a dense subset of conditions below forces .
Let be arbitrary condition. We can choose such that . By claim 3.1 there is and such that .
Recall that if , , and , then , where is such that . Thus for each , , and ,
Thus . Use claim 3.2 to find stronger conditions and such that is an automorphism. Since we get . We are done since . □
We will get theorem 1 by beginning with a model where . For this let us define the following coding. Let be an enumeration of all subsets of . Let be the Easton product of the Cohen forcing notions yielding, in the generic extension, for each and ,
Corollary 3.9. Let is an extender witnessing is a -supercompact cardinal. In , where is -generic, , and for each set , and is measurable in .
Proof. Let be a measure on . Then , hence , where is -generic.
Working in let be -generic. By corollary 3.8 there is such that , , and . The filter is -generic. Since we have , hence any measure (in ) over trivially lifts to a measure in over . In particular lifts to , which is definable by . Since we can define in , . Since we necessarily have . Thus is a measure on in . □
In this section we prove theorem 2. The extender based Radin forcing was originally developed in [7]. We use a generalization of the forcing where the extenders can witness supercompactness. This was developed in [9]. Since the introduction in the previous section was very detailed, we will give here only our version of Radin forcing.
Let us begin with with defining the Magidor forcing [6] using two measures. Assume are two normal measures over . For each let be the natural elementary embeddings. Define the measure by letting for each ,
Note a typical function is of the form . Define the measure by letting for each ,
A typical function is of the form , where and is a measure over . Define by recursion the conditions and ordering of the forcing notion as follows. A basic condition in Magidor forcing is of the form , where and is a function such that , where and for each , is a measure over . A sequence of the form is said to be -decreasing since we consider and . Assume is a condition and is a function. We define the functions and as follows. Assume . In this case we work essentialy as in the Prikry forcing case. We let . Define the function by letting . Note that since the sequence is -decreasing if is -decreasing. Assume is a function such that . In this case we define two functions and . If we would have let then we might have ended with a non -decreasing sequence. Thus we cut the possible problematic tail of as follows. Set . If the set over which the above is operating is empty then set . Then let . Whatever is the value of let . The tail removed from is ‘pushed down’ by letting be a function such that , where is if . Together with the ‘pushed down function’ we set the pushed down part of to be , where . Finally set and . Note is a condition in a Prikry forcing. A 1-point extension of is .
Assume is generic with , where . Letting and we get that is an sequence cofinal in . Moreover if then setting and . Then is an -sequence cofinal in .
Let us switch to the extender based Magidor forcing using two extenders. Assume are two extenders over . For each let be the natural elementary embeddings.
For each such that define the measure as follows. For each ,
A typical function is of the form where , and for each such that we have . Moreover, . For each such that define the ultrafilter as follows. For each ,
A typical function is of the form where , and for each such that we have . Moreover, . Define by recursion the conditions and order on the forcing notion as follows. A basic condition in the extender based Magidor forcing is of the form , where is a function such that for each we have is a finite -decreasing sequence, and . We can always assume that for each we have for each . Assume is a condition and is a function. We define the functions and , employing the same idea used on to each of the ’s, as follows. Assume . Let . Define the function by setting for each ,
Assume . Define the function by setting for each ,
where
Let be the function defined by setting for each , . Set . If then set . If then set , where defined by setting for each , . Then define and to be and , respectively. We say the condition is stronger than the condition (), if and .
Assume is generic with , where . Letting and we get that is an sequence cofinal in . Moreover if then setting and . Then is an -sequence cofinal in . Note there are new -sequences cofinal into . For each of the reflections down we get the reflected amount of -sequences. E.g., if then there are new -sequences cofinal in .
Letting be the generic object and defining the function by setting for each , , we get that for each , is an -sequence cofinal in . Defining the function as in the previous paragraph we get cofinal -sequences, but still all of them are generated by .
Since the ultrapower is closed only under -sequences, one cannot enlarge to be of size greater than while keeping the nice properties of Prikry type forcing notions. However, we can use conditions with different domains, thus adding sequence corresponding to each of the ordinals below . The domain change, however, causes the forcing to be non-isomorphic to Prikry forcing. Thus if then define the measure by letting for each ,
Define now a condition in the forcing to be of the form , where is a function such that for each , is a finite increasing sequence, , and . If and then set . The condition is said to be a direct extension of the condition () in this forcing if and . Note this definition of the direct order is a major change from all previous definitions. In fact the direct order is a Cohen forcing for adding subsets to . Assume is a condition and is a function. The function is defined by setting for each ,
Given a set set . Then define the condition to be . We say the condition is a -point extension of . By recursion define the -point extension of the condition to be . We say the condition is stronger than the condition () if there is such that . This forcing notion is no longer Prikry forcing. Using the same definition of as before we get cofinal -sequence into . However, while each of the sequences appears in a generic extension by Prikry forcing, the function itself does not below to a Prikry generic extension.
The point of the previous forcing is that nothing restricts us from using it with elementary embeddings with many generators, thus we get the extender based Prikry forcing. Thus assume is an elementary embedding such that , , and is a cardinal in . Using the previous definition with conditions such that , we get that new -sequences cofinal in appear in the generic extension, thus . Working out the proof we get that no cardinals are collapsed, , and .
The final generalization achieved so far is to begin with elementary embedding with even more closure properties, i.e., such that and , where and then work out the definition above to use of arbitrary size below . This yields a generic extension in which and blows up to whatever cardinal the model catches beginning with up to . However, the cardinals above and below are collapsed in this extension.
Thus throughout this section assume is a Mitchell increasing sequence of extenders such that is measurable, and for each , , , and , where is the natural embedding. (We demand since we want to be measurable in all ultrapowers, not only in ).
Let be the supercompact extender based Radin forcing using . (see [9]).
Let us recall the cardinal structure in . remains an inaccessible cardinal, hence is a model of ZFC. while remains a cardinal, the cardinals between and are collapsed. Both and are reflected down using the different extenders. Let be reflection of which a limit cardinal in . Let be the matching reflection of . Then is preserved while the -cardinals between and are collapsed.
Let us deal with the homogeneity of the Extender Based Radin forcing.
Proof. Set . Set and . Choose a set so that and are conditions, and . Define the isomorphism as follows. Thus assume . By the definition of the order there is such that . I.e., , where for each , and . Consider the condition . Note . Let , where . Finally set . Let us show the function is order preserving. Fix . We will show .
Since there is such that , where for each , and . Since there is such that , where for each , and . Thus and , where
Since there such that . Thus there is such that . I.e., for each , and . Noting that for each , and , we conclude that , hence .
Proceeding as above for each , (e.g., there is such that ) we get that . □
Recall that for a condition we have , where and . Thus the following is an immediate corollary of the above lemma by recursion.
For a condition define its projection to the normal measure by setting . Define by recursion the projection of arbitrary condition by setting . It is obvious is the Radin forcing using the measures . Moreover, if is -generic then is -generic.
Let be -generic. Work in . Let be the increasing enumeration of . Define the sequence by setting for each ,
Note: If is limit, then is measurable in since it is a reflection of being measurable in one of the -ultrapowers. On the other hand, if is successor then is measurable in since concentrates on measurables. Thus for each we can choose which is a measure in over . Define the backward Easton iteration by setting for each , . By theorem 2.2 the iteration is cone homogeneous. Let be generic.
Working in we want to pull into the of a generic extension the measures ’s. Define the backward Easton iteration by setting for each , , where, and . By theorem 2.2 is cone homogeneous.
One final definition is in order before the following claim. If then set . If then set by recursion . Note is the subset of decided by the condition .
Claim 4.3. Let . Assume are conditions such that and are compatible. Then there are stronger conditions, and , such that .
Proof. Since and are compatible there are stronger conditions and and Mitchell increasing sequences such that and . By the previous corollary there are direct extensions and such that . Most importantly we have is cone homogeneous. Thus by claim 2.3 we are done. □
Proof. We will prove a dense subset of conditions below force . Assume . Trivially and are compatible, hence by the previous corollary there are stronger conditions and such that . Necessarily . □
Letting be -generic over we get the following from the previous corollary together with claim 2.4.
Proof. Since the regulars in the range are , we will be done by showing for each the measure (in ) lifts to a measure in . In , is measurable. The set is the plain Radin forcing, hence any measure in over lifts trivially to a measure on in . In particular the measure in lifts to the measure in , which is definable by .
Since we get . Let . Then and . Necessarily is a measure on . □
We get theorem 2 by forcing in with .
[1] Omer Ben-Neria and Spencer Unger. Homogenerous changes in cofinalities with applications to HOD. Preprint.
[2] James Cummings, Sy David Friedman, and Mohammad Golshani. Collapsing the cardinals of hod. Journal of Mathematical Logic, 15(02):1550007, 2015.
[3] James Cummings, Sy-David Friedman, Menachem Magidor, Dima Sinapova, and Assaf Rinot. Ordinal definable subsets of singular cardinals. Preprint.
[4] Natash Dobrinen and Sy-David Friedman. Homogeneous iteration and measure one covering relative to HOD. Archive for Mathematical logic, 47(7–8):711–718, November 2008. doi:10.1007/s00153-008-0103-5.
[5] Moti Gitik and Menachem Magidor. The Singular Cardinal Hypothesis revisited. In Haim Judah, Winfried Just, and W. Hugh Woodin, editors, Set theory of the continuum, volume 26 of Mathematical Sciences Research Institute publications, pages 243–279. Springer, 1992.
[6] Menachem Magidor. Changing the cofinality of Cardinals. Fundamenta Mathematicae, 99(1):61–71, 1978.
[7] Carmi Merimovich. Extender-based Radin forcing. Transactions of the American Mathematical Society, 355:1729–1772, 2003. doi:10.1090/S0002-9947-03-03202-1.
[8] Carmi Merimovich. Supercompact Extender Based Prikry forcing. Archiv for Mathematical Logic, 50(5-6):592—601, June 2011. doi:10.1007/s00153-011-0234-y.
[9] Carmi Merimovich. Supercompact extender based Magidor-Radin forcing. Annals of Pure and Applied Logic, 168(8):1571–1587, 2017. doi:10.1016/j.apal.2017.02.006.
[10] W. Hugh Woodin. Suitable extender models I. Journal of Mathematical Logic, 10(01n02):101–339, 2010. doi:10.1142/S021906131000095X.
School of Mathematical Sciences, Raymond and Beverly Sackler Faculty of Exact Sciences, Tel Aviv University, Ramat Aviv 69978, Israel
Email address: [email protected]
Computer Science School, Tel Aviv Academic College, 2 Rabenum Yeroham St., Tel Aviv, Israel
Email address: [email protected]