Author 
Message 
Lubos Motl science forum beginner
Joined: 02 May 2005
Posts: 38

Posted: Fri Mar 11, 2005 11:19 pm Post subject:
Re: deconstruction



On Fri, 11 Mar 2005, Urs Schreiber wrote:
I suppose the question from Robert is
Quote:  So what is more in deconstruction that M(atrix)Tduality?

I may misunderstand what is Robert asking about because the answer is
obvious.
The models with Tduality he mentions always describe completely and
exactly continuous dimensions. There are really no "quivers" in these
models. The corresponding "quiver" is a smooth space. They're not
deconstruction.
The whole point of deconstruction is that the theory space (quiver or
moose) is *not* exactly continuous which makes a lot of difference. One
has two basic reasons why we consider discrete theory spaces:
1. First of all, there is a phenomenological reason: we may "deconstruct"
extra dimensions beyond the four dimensions we have, and obtain
effectively fourdimensional physics. This is done in the little Higgs
models and similar stuff where the quiver is pretty small (and as
different from a continuous extra dimensions as you can get) and these
fourdimensional theories with finite quivers are welldefined. If
the gauge groups are asymptotically free, the theories are defined
even nonperturbatively.
2. Even in the theoretical context, it's important to have a finite value
of N exactly because we don't know how to define a theory in which
something is explicitly infinite  but we know how to take the limit.
This allows you, for example, to write down an inprinciple definition
of the (2,0) SCFT and (1,1) little string theory in 6D in such a way
that the locality at least in 4 dimensions is manifest
http://arxiv.org/abs/hepth/0110146
The difference between this definition of these exotic 6D theories
and any other approach different from deconstruction is that
deconstruction can give the definition, but no other approach has
provided us with such a definition before. If Robert has another
definition of the (2,0) theory that allows calculations, at least in
principle, I am very interested in it.
Simply speaking, the "other side" that Robert says is equivalent to
deconstruction has nothing to do with deconstruction because all of its
dimensions are explicitly continuous. Deconstruction means that the "extra
dimensions" are really discrete.
All the best
Lubos
______________________________________________________________________________
Email: lumo@matfyz.cz fax: +1617/4960110 Web: http://lumo.matfyz.cz/
eFax: +1801/4541858 work: +1617/3849488 home: +1617/8684487 (call)
Webs: http://schwinger.harvard.edu/~motl/ http://motls.blogspot.com/
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 

Back to top 


Urs Schreiber science forum Guru Wannabe
Joined: 04 May 2005
Posts: 127

Posted: Sun Mar 13, 2005 2:59 pm Post subject:
Re: deconstruction



"Lubos Motl" <motl@feynman.harvard.edu> schrieb im Newsbeitrag
news:Pine.LNX.4.31.0503111908120.28181100000@feynman.harvard.edu...
Quote:  This allows you, for example, to write down an inprinciple definition
of the (2,0) SCFT and (1,1) little string theory in 6D in such a way
that the locality at least in 4 dimensions is manifest
http://arxiv.org/abs/hepth/0110146

By the way, can you identify n^3scaling in the deconstrcuted constructions
of these 6D theories? 

Back to top 


Lubos Motl science forum beginner
Joined: 02 May 2005
Posts: 38

Posted: Mon Mar 14, 2005 1:22 pm Post subject:
Re: deconstruction



Urs wrote:
Quote:  By the way, can you identify n^3scaling in the deconstrcuted constructions
of these 6D theories?

Great question. The simple answer is NO (so far). The longer answer may be
YES, but the argument does not depend on the details of the deconstructed
theory too much  it is based on the identification of two different
scales that differ by a factor of "n"  in the intermediate interval, the
entropy gets an extra kick by a factor of "n" by linear scaling. If
someone knows how to do it directly from the quiver theory, I would love
to hear it.
______________________________________________________________________________
Email: lumo@matfyz.cz fax: +1617/4960110 Web: http://lumo.matfyz.cz/
eFax: +1801/4541858 work: +1617/3849488 home: +1617/8684487 (call)
Webs: http://schwinger.harvard.edu/~motl/ http://motls.blogspot.com/
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 

Back to top 


Robert C. Helling science forum beginner
Joined: 30 Apr 2005
Posts: 22

Posted: Fri Mar 18, 2005 11:28 am Post subject:
Re: deconstruction



On Fri, 11 Mar 2005 19:19:53 0500, Lubos Motl <motl@feynman.harvard.edu> wrote:
Quote:  On Fri, 11 Mar 2005, Urs Schreiber wrote:
Over on his blog Robert Helling made an imteresting comment on
deconstruction.
http://atdotde.blogspot.com/2005/03/deconstructionhelpsought_10.html .
Does anyone have a good answer?
I suppose the question from Robert is
So what is more in deconstruction that M(atrix)Tduality?
I may misunderstand what is Robert asking about because the answer is
obvious.
The models with Tduality he mentions always describe completely and
exactly continuous dimensions. There are really no "quivers" in these
models. The corresponding "quiver" is a smooth space. They're not
deconstruction.
The whole point of deconstruction is that the theory space (quiver or
moose) is *not* exactly continuous which makes a lot of difference. One
has two basic reasons why we consider discrete theory spaces:

My question was about infinite N.
Quote:  2. Even in the theoretical context, it's important to have a finite value
of N exactly because we don't know how to define a theory in which
something is explicitly infinite  but we know how to take the limit.

Really? As I said, the result for the limit at least resembles the
theory for a finite number of D3's say (D0 in the original reference)
with a periodic transversal direction. Wati Tayler explained how to
treat this case by including the inifinte number of image branes. The
resulting theory is a 3+1 dimensional theory with infinte matrices
but after a Fourier transform can be brought into the language of a
4+1 dimensional theory with finite matrices. What makes you confident
that the above limit you mention exists _in the quantum theory_?
Quote:  The difference between this definition of these exotic 6D theories
and any other approach different from deconstruction is that
deconstruction can give the definition, but no other approach has
provided us with such a definition before. If Robert has another
definition of the (2,0) theory that allows calculations, at least in
principle, I am very interested in it.

No, of course I don't have an alternative description. I just mention
that this definition (assuming the limit exists) looks like another,
older version to write higher dimensional gauge theories in terms of
lower dimensional ones.
Quote:  Simply speaking, the "other side" that Robert says is equivalent to
deconstruction has nothing to do with deconstruction because all of its
dimensions are explicitly continuous. Deconstruction means that the "extra
dimensions" are really discrete.

Discrete yes, but if there is an infinite number of discrete points,
the Fourier transform has the potential to relate this to a
continious, periodic dimension.
Robert

..oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oOo.oO
Robert C. Helling School of Science and Engineering
International University Bremen
print "Just another Phone: +49 421200 3574
stupid .sig\n"; http://www.aeipotsdam.mpg.de/~helling 

Back to top 


Lubos Motl science forum beginner
Joined: 02 May 2005
Posts: 38

Posted: Fri Mar 18, 2005 12:07 pm Post subject:
Re: deconstruction



On Fri, 18 Mar 2005, Robert C. Helling wrote:
Quote:  My question was about infinite N.

That's OK, but in that case, your questions was *not* about deconstruction
because the point of deconstruction is to describe physics analogous to
continuous dimensions in terms of a theory with a *finite* N (number of
copies of the gauge group).
Quote:  2. Even in the theoretical context, it's important to have a finite value
of N exactly because we don't know how to define a theory in which
something is explicitly infinite  but we know how to take the limit.
Really? As I said, the result for the limit at least resembles the
theory for a finite number of D3's say (D0 in the original reference)
with a periodic transversal direction.

That's right. It's because it's the goal of deconstruction to resemble
physics in continuous dimensions  but resemble it using a controllable
theory where only a smaller number of dimensions are strictly continuous.
Quote:  Wati Tayler explained how to treat this case by including the inifinte
number of image branes.

This is a very naive viewpoint. Wati Taylor has shown how to make Fourier
transformation in the space of continuous indices. One can either view a
D(p+1)brane on a circle in terms of a new "continuous index" i.e. a new
dimension in its worldvolume theory, or as an array of Dpbranes in the
Tdual theory. Tduality is represented as a simple Fourier transform on
the space of gauge indices. This machinery says nothing about the
existence of the theories at quantum level.
Quote:  The resulting theory is a 3+1 dimensional theory with infinte matrices
but after a Fourier transform can be brought into the language of a 4+1
dimensional theory with finite matrices.

What you obtain with this procedure is the naive 4+1dimensional gauge
theory which is not renormalizable, and the whole quantum physics is
unknown. It's just a formal manipulation that happens to give the right
classical limit at low energies. But deconstruction gives you the quantum
consistent UV complete definition of the 4+1dimensional gauge theory,
namely the (2,0) theory on a circle. You can't get these things just by
the manipulations you describe.
You know, the procedure that you seem to like so much was initially
thought to be relevant for the definition of M(atrix) theory on
*arbitrary* tori. It was soon realized that it is only good up to the
threetorus where the matrix model is the 3+1dimensional maximally
supersymmetric gauge theory on a dual threetorus.
Beyond this point, one derives theories that can't be described in this
simple form. The "4+1dimensional YangMills" is a correct description of
the relevant matrix model for Mtheory on T^4 at low energies only. The
appropriate quantum completion is the sixdimensional (2,0) theory on a
circle and a fourtorus  more generally, on a fivetorus. (The same
theory one gets by deconstruction.) One may derive this from Tdualities 
Tdualizing D0branes four times gives D4branes, but if one is careful
about the string coupling, it diverges. Therefore one really works with
M5branes in Mtheory, and the relevant regime that enters the definition
of the matrix model is the SCFT which is the (2,0) theory.
Similarly, Mtheory on T^5 is not described just by "5+1dimensional
YangMills" that you obtain from your naive classical procedure. The right
UV completion is the (1,1) 6D little string theory on a fivetorus  that,
incidentally, inherits the Tduality group SO(5,5) from the full IIB
string theory, which is interpreted as the Uduality of Mtheory on T5.
The matrix description beyond T^5, namely for Mtheory on T^6 and higher,
does not exist because gravity does not decouple, and one expects that
there is no nongravitational UV completion of the gauge theory. The
toroidal compactification of BFSS is the main focus of Ashoke Sen's paper
http://arxiv.org/abs/hepth/9709220
who identified the correct theories before they were described and detail
and called "little string theory", for example.
Quote:  What makes you confident that the above limit you mention exists _in the
quantum theory_?

There are two main things that make me confident about that: the proof
that only relies on Tduality, the lowenergy description of D3branes on
orbifolds, and on the fact that the physics of cone with a small opening
angle approaches the physics of cylinder. Neither of these things makes
any unsatisfied assumption about the strength of the coupling, unlike your
classical procedure, and the reliability of the considerations is
strengthened by the large amount of preserved supersymmetry. The second
thing that makes me confident are consistency checks  the description has
the right limit "4+1dimensional YangMills" etc. by the usual
considerations, but it also has the right duality groups etc. so that it
is almost forced to be the advertised theory in all regimes  certainly
all asymptotic ones  at quantum level. It contains the right perturbative
excitations as well as monopoles, and I see no way how it could "fail" in
the middle, especially if there exists the proof.
Quote:  No, of course I don't have an alternative description. I just mention
that this definition (assuming the limit exists) looks like another,
older version to write higher dimensional gauge theories in terms of
lower dimensional ones.

There is no other way how to write the little string theory using its
lowdimensional counterparts, and therefore the words "another, older" in
your sentence are not correct. Be sure that if you try to make a generic
latticization of 4+1dimensional YangMills, you won't obtain the (2,0)
theory on a circle or any other consistent quantum theory unless you
finetune an infinite number of coupling constants. The special virtue of
the deconstruction is that it preserves some supersymmetries which
together with the discrete symmetries in the continuum limit guarantee
that the limiting theory has all the required supercharges, and therefore
it is unique given the degrees of freedom. Why deconstruction is clearly
superior for "latticizing" supersymmetry  see e.g. a talk
http://arxiv.org/abs/heplat/0208046
Quote:  Discrete yes, but if there is an infinite number of discrete points,
the Fourier transform has the potential to relate this to a
continious, periodic dimension.

That's right. This is a fact that deconstruction uses heavily.
______________________________________________________________________________
Email: lumo@matfyz.cz fax: +1617/4960110 Web: http://lumo.matfyz.cz/
eFax: +1801/4541858 work: +1617/3849488 home: +1617/8684487 (call)
Webs: http://schwinger.harvard.edu/~motl/ http://motls.blogspot.com/
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 

Back to top 


Urs Schreiber science forum Guru Wannabe
Joined: 04 May 2005
Posts: 127

Posted: Sun Mar 20, 2005 3:43 pm Post subject:
Re: deconstruction



Lubos Motl <motl@feynman.harvard.edu> wrote in message news:<Pine.LNX.4.62.0503140918450.2188@feynman.harvard.edu>...
Quote:  Urs wrote:
By the way, can you identify n^3scaling in the deconstrcuted constructions
of these 6D theories?
Great question. The simple answer is NO (so far). The longer answer may be
YES, but the argument does not depend on the details of the deconstructed
theory too much  it is based on the identification of two different
scales that differ by a factor of "n"  in the intermediate interval, the
entropy gets an extra kick by a factor of "n" by linear scaling. If
someone knows how to do it directly from the quiver theory, I would love
to hear it.

Here is another question:
Given a field theory on some spacetime M with fields taking values in
some space F we know that it is in general not correct to describe
these fields as Fvalued functions on M, but that they have to be
considered as sections of a (possibly nontrivial) bundle over M with
typical fiber F.
When deconstructing dimensions a quiver gets interpreted as (part of)
spacetime. There are fields living on the edges of the quiver which
are usually regarded as functions from these edges to some space of
homomorphisms.
Here is the question: Do we in the most general case have to allow
these functions to really be sections of a possibly nontrivial
"bundle" or something? 

Back to top 


Urs Schreiber science forum Guru Wannabe
Joined: 04 May 2005
Posts: 127

Posted: Tue Mar 22, 2005 7:55 am Post subject:
Re: deconstruction



Urs Schreiber <Urs.Schreiber@uniessen.de> wrote in message news:<206f2305.0503200214.15aa53ed100000@posting.google.com>...
Quote:  Given a field theory on some spacetime M with fields taking values in
some space F we know that it is in general not correct to describe
these fields as Fvalued functions on M, but that they have to be
considered as sections of a (possibly nontrivial) bundle over M with
typical fiber F.
When deconstructing dimensions a quiver gets interpreted as (part of)
spacetime. There are fields living on the edges of the quiver which
are usually regarded as functions from these edges to some space of
homomorphisms.
Here is the question: Do we in the most general case have to allow
these functions to really be sections of a possibly nontrivial
"bundle" or something?

Let me try to phrase the same question in a "dual" picture, where it
might be more transparent if it makes any sense:
So instead of thinking of the quiver as a deconstructed dimension, go
back to the Dbrane picture and think of every vertex of the quiver as
a fractional brane on some C^3/G orbifold CY and as the edges as
string states between these fractional branes.
Consider a quiver which contains at least one noncontractible loop.
Now I am going to ask if it makes sense to consider the case where the
nature of the fractional branes gradually changes as we go along this
loop such that by going once around we have effectively traced out a
noncontractible loop in moduli space.
I'll try to make that a little more precise as follows:
Consider a flow Phi through the moduli space of the CY. Given any
Dbrane E with respect to the CY at Phi(0) let Phi(t)(E) be the result
of transporting that Dbrane with the flow Phi a distance t through
moduli space.
Since I want all my fractional branes to live in the same CY invert
this picture:
The brane which, when transported a distance t _to_ the CY at Phi(0),
looks like E there is given by Phi(t)(E).
Consider for a moment a quiver with just one edge, i.e. a situation
where we have just two fractional branes E1 and E2 with strings
between them. Can we also have a physical situation with E1 and
Phi(t)(E2) instead of E1 and E2, for sufficiently small but
nonvanishing Phi(t)?
I have the vague idea that for small t this should not change anything
of the physics, but I don't really know. Does anyone?
If however this makes sense, then it should make sense to do the
following:
Given a quiver with a noncontractible loop, let the brane states
along that loop be labelled E0, E1, E2, ... En = E0.
Now, assuming the above is correct, it should make sense to gradually
go from
E0, E1 to Phi(0 epsilon)(E0) , Phi(1 epsilon)(E1)
and
E1, E2 to Phi(1 epsilon)(E1) , Phi(2 epsilon)(E2)
etc,
in a case where
Phi(n epsilon) = Phi(0) ,
i.e. where the flow Phi is a loop in moduli space.
So in this case the fractional brane Ei is of the same kind as E(i+1),
but En is equal to E0 only op to the effect of going around a closed
loop in moduli space. If that loop has nontrivial monodromy, as it may
happen, we would get the situation that I am intersted in.
The reason I am interested in this is because I believe there would be
an interesting solution to make this setup consistent, if indeed this
setup makes any sense. A classical case of a solution looking for a
problem. 

Back to top 


Aaron Bergman science forum addict
Joined: 24 Mar 2005
Posts: 94

Posted: Wed Mar 23, 2005 8:55 am Post subject:
Re: deconstruction



In article <206f2305.0503220030.2bd609a3@posting.google.com>,
Urs Schreiber <Urs.Schreiber@uniessen.de> wrote:
Quote:  Consider a flow Phi through the moduli space of the CY. Given any
Dbrane E with respect to the CY at Phi(0) let Phi(t)(E) be the result
of transporting that Dbrane with the flow Phi a distance t through
moduli space.
Since I want all my fractional branes to live in the same CY invert
this picture:
The brane which, when transported a distance t _to_ the CY at Phi(0),
looks like E there is given by Phi(t)(E).
Consider for a moment a quiver with just one edge, i.e. a situation
where we have just two fractional branes E1 and E2 with strings
between them. Can we also have a physical situation with E1 and
Phi(t)(E2) instead of E1 and E2, for sufficiently small but
nonvanishing Phi(t)?

I can't figure out what you're talking about here. But, as I said,
generally noncontractible loops in the moduli space give rise to
autoequivalences of the derived category when you look at the monodromy
around the loop.
Aaron 

Back to top 


Urs Schreiber science forum Guru Wannabe
Joined: 04 May 2005
Posts: 127

Posted: Wed Mar 23, 2005 9:08 am Post subject:
Re: deconstruction



On Wed, 23 Mar 2005, Aaron Bergman wrote:
Quote:  I can't figure out what you're talking about here.

I am trying to understand if the assignment of vector spaces and linear
maps to a quiver should just be a functor or rather really a 2section
of a possibly nontrivial 2bundle over the quiver.
The spacetime interpretation of the quiver in dimensional deconstruction
naively suggests the latter. I am trying to see if this would make sense
from the perspective of branes on orbifolds.
Quote:  But, as I said, generally
noncontractible loops in the moduli space give rise to autoequivalences of
the derived category when you look at the monodromy around the loop.

Yes. So the Picard group would naturally arise as the structure 2group of
such a 2bundle. Formally this idea looks quite suggestive. I am trying to
understand if it is physically meaningful. 

Back to top 


Aaron Bergman science forum addict
Joined: 24 Mar 2005
Posts: 94

Posted: Thu Mar 24, 2005 2:27 pm Post subject:
Re: deconstruction



In article <Pine.LNX.4.62.0503230458001.21151@feynman.harvard.edu>,
Urs Schreiber <Urs.Schreiber@uniessen.de> wrote:
[...]
Sorry, Urs.
I wish I could help some more, but you've completely lost me.
Aaron 

Back to top 


Urs Schreiber science forum Guru Wannabe
Joined: 04 May 2005
Posts: 127

Posted: Fri Mar 25, 2005 8:02 pm Post subject:
Re: deconstruction



Aaron Bergman wrote:
Quote:  [...]
Sorry, Urs.
I wish I could help some more, but you've completely lost me.

Yesterday I talked about this stuff with Jarah Evslin. He said what I
was telling him reminded him of the KlebanovStrassler cascade
mechanism that he had reformulated in terms of systems of 5branes and
3branes.
So I guess when we blow up the C^3/G singlularity his setup applies and
we then know that the gauge group living on the branes is defined only
up to a certain shift. U(m)xU(n) to U(mn)xU(m) or something.
But this gauge group is what is defined by the rep of the qiver. If
that's only well defined up to something we arrive essentially at my
idea that we should not think of the quiver rep as a function but as a
section. I think...
More later, I am still at Milano central station... 

Back to top 


Aaron Bergman science forum addict
Joined: 24 Mar 2005
Posts: 94

Posted: Sun Mar 27, 2005 5:05 pm Post subject:
Re: deconstruction



In article <1111780379.433364.80330@f14g2000cwb.googlegroups.com>,
Urs <Urs.Schreiber@uniessen.de> wrote:
Quote:  So I guess when we blow up the C^3/G singlularity his setup applies and
we then know that the gauge group living on the branes is defined only
up to a certain shift. U(m)xU(n) to U(mn)xU(m) or something.

That's not really what's going on. The idea in KS is that there's a sort
of 'cascade' of Seiberg dualities. It's not that the gauge group is only
defined up to a shift; it's that in different energy scale regimes, the
has different effective descriptions. Sort of. That's not quite right,
but KS is a little funky.
But, it's certainly true that, even for a superconformal theory, the
assignment of a quiver to a particular geometry is not unique. This is
because the actual SCFT is the IR limit of the quiver gauge theory. You
can have lots of different quivers that have the same IR limit. The
procedure of obtaining quivers from derived categories depends on the
choice of an exceptional collection which is far from unique. Herzog
proves in hepth/0405118 that many (all?) of these quivers are related
by Seiberg duality.
Aaron 

Back to top 


Urs Schreiber science forum Guru Wannabe
Joined: 04 May 2005
Posts: 127

Posted: Sun Mar 27, 2005 5:42 pm Post subject:
Re: deconstruction



On Sun, 27 Mar 2005, Aaron Bergman wrote:
Quote:  But, it's certainly true that, even for a superconformal theory, the
assignment of a quiver to a particular geometry is not unique. This is
because the actual SCFT is the IR limit of the quiver gauge theory. You can
have lots of different quivers that have the same IR limit. The procedure of
obtaining quivers from derived categories depends on the choice of an
exceptional collection which is far from unique. Herzog proves in
hepth/0405118 that many (all?) of these quivers are related by Seiberg
duality.

Thanks.
Familiy duties over easter holidays prevent me from doing
real work today, so once again just another followup question which in a
better world I would have time to look up in the literature:
When do we say that two quivers are equivalent? Or dual?
Given a representation R1 of a quiver Q1 which is equivalent or dual to
Q2, do we get an induced representation R2 of Q2 from R1? 

Back to top 


Aaron Bergman science forum addict
Joined: 24 Mar 2005
Posts: 94

Posted: Mon Mar 28, 2005 2:23 pm Post subject:
Re: deconstruction



In article <Pine.LNX.4.62.0503271437340.2929@feynman.harvard.edu>, Urs Schreiber <Urs.Schreiber@uniessen.de> wrote:
Quote:  On Sun, 27 Mar 2005, Aaron Bergman wrote:
But, it's certainly true that, even for a superconformal theory, the
assignment of a quiver to a particular geometry is not unique. This is
because the actual SCFT is the IR limit of the quiver gauge theory. You can
have lots of different quivers that have the same IR limit. The procedure
of
obtaining quivers from derived categories depends on the choice of an
exceptional collection which is far from unique. Herzog proves in
hepth/0405118 that many (all?) of these quivers are related by Seiberg
duality.
Thanks.
Familiy duties over easter holidays prevent me from doing
real work today, so once again just another followup question which in a
better world I would have time to look up in the literature:
When do we say that two quivers are equivalent? Or dual?
Given a representation R1 of a quiver Q1 which is equivalent or dual to
Q2, do we get an induced representation R2 of Q2 from R1?

The quivers all have the same nodes, I think, so the dimension vectors
are the same. Then, what I think everyone believes is that the quiver
variety, ie, the moduli space of the gauge theories, is the same.
To see the action of Seiberg duality, I'd recommend checking out Chris's
paper above, Aspinwall's paper 0405134 and references therein.
Aaron 

Back to top 


Urs Schreiber science forum Guru Wannabe
Joined: 04 May 2005
Posts: 127

Posted: Tue Mar 29, 2005 2:24 pm Post subject:
Re: deconstruction



"Aaron Bergman" <abergman@physics.utexas.edu> schrieb im Newsbeitrag news:abergman68A952.14112427032005@localhost...
Quote:  To see the action of Seiberg duality, I'd recommend checking out Chris's
paper above, Aspinwall's paper 0405134 and references therein.

Ok. Beginning to do so I have come across
Berenstein & Douglas
Seiberg Duality for Quiver Gauge Theories
hepth/0207027
They make explicit what I was going to ask as the next obvious question,
namely that these dualities are described by twosided tilting complexes.
That is, a duality between a gauge theory described by a a quiver with path
algebra A and one with path algebra B is given by an object
T in D( AmodB )
of the derived category of leftA/rightBbimodules which is weakly
invertible, i.e. such that there is a T' such that
T \otimes^L T' ~ A
and
T' \otimes^L T ~ B ,
where A and B here are the algebras themselves regarded as bimodules over
themselves and playing the role of the identity object.
The duality is then just given by \otimes^Lmultiplying a Dbrane state in
D(modA)
or
D(Bmod)
with T.
I note that a special case of D( AmodB ) this is the derived category of
Abimodules
D( AmodA ).
describing autoequivalences (and hence, as you emphasized, in particular
monodromies in moduli space ) which contains a weak 2group as a full
subcategory, namely that containing all the morphisms between weakly
invertible objects = tilting complexes.
I have asked Amnon Yekutieli if this 2group structure here is discussed in
the literature. He didn't know about it, so I guess it doesn't. Which is
strange, because the group you get by taking isomorphism classes of objects
of this 2group is the wellknown derived Picard group. From the
categorytheoretic viewpoint it seems strange to immediately pass to just
the isomorphism class of objects. There is much more information in the
2group than just that.
Now, from Berenstein and Douglas discussion I have not quite managed to
extract yet if that "more" of information is physically useful or just
redundant.
Anyway, there is a small comment on subquivers in Berenstein&Douglas which
makes me wonder (p. 43). Would you know of any more detailed discussion
which crucially uses embeddings of gaugetheory quivers into larger quivers?
Somehow?
Sorry, a rather vague question once again. 

Back to top 


Google


Back to top 



The time now is Sat Nov 17, 2018 5:37 pm  All times are GMT

