In the last lecture, we reduced the problem of cheating in (the k-times repeated m-cycle game) to finding a small set of edges
in
whose removal eliminates all topologically non-trivial cycles. Such a set
is called a spine. To get some intuition about how many edges such a spine should contain, let’s instead look at a continuous variant of the problem.
Spines, Foams, and Isoperimetry
Consider again the -dimensional torus
, which one can think of as
with opposite sides identified. Say that a nice set (e.g. a compact,
surface)
is a spine if it intersects every non-contractible loop in
. This is the continuous analog of a spine in
. We will try to find such a spine
with surface area, i.e.
, as small as possible.
Let’s consider some easy bounds. First, it is clear that the set
is a spine with . (A moment’s thought shows that this is “equivalent” to the provers playing independent games in each coordinate of
.)
To get a good lower bound, it helps to relate spines to foams which tile according to
, as follows. Take two potential spines.
To determine which curve is actually a spine, we can repeatedly tile them side-by-side.
The first tiling contains the blue bi-infinite curve, which obviously gives a non-trivial cycle in , while the second yields a tiling of
by bodies of volume 1. It is easy to deduce the following claim.
Claim: A surface is a spine if and only if it induces a tiling of the plane by bodies of volume 1 which is invariant under shifts by
.
By the isoperimetric inequality in , this immediately yields the bound
where is the unit
-dimensional sphere.
So the the surface area of an optimal spine lies somewhere between and
. On the one hand, cubes tile very nicely but have large surface area. On the other hand, we have sphere-like objects which have small surface area, but don’t seem (at least intuitively) to tile very well at all. As first evidence that this isn’t quite right, note that it is known how to cover
by disjoint bodies of volume at most 1 so that the surface area/volume ratio grows like
. See Lemma 3.16 in this paper, which is based on Chekuri, et. al. It’s just that these covers are not invariant under
shifts.
Before we reveal the answer, let’s see what consequences the corresponding discrete bounds would have for parallel repetition of the m-cycle game. If the “cube bound” were tight, we would have , which doesn’t rule out a strong parallel repetition theorem (
in the previous lecture). If the “sphere bound” were tight, we would have
, which shows that
. In the latter case, the approach to proving equivalence of the UGC and MAX-CUT conjectures doesn’t even get off the ground.
As the astute reader might have guessed, recently Ran Raz proved that for some constant
, showing that a strong parallel repetition theorem—even for unique games—is impossible. Subsequently, Kindler, O’Donnell, Rao, and Wigderson showed that there exists a spine
with
. While it is not difficult to show that the continuous result implies Raz’s discrete result, we will take a direct approach found recently by Alon and Klartag.
Spectral partitioning, Cheeger’s inequality, and Dirichlet boundary conditions
First, we will show that it suffices to find a subset so that the induced graph on
contains no non-trivial cycles, and
is small, where we use
to denote the set of edges from
to
in
. A variant of this reduction appears in KORW. For a subset
, let
.
Random Partitioning Lemma: For any subset which contains no non-trivial cycles, there exists a spine
with
.
Proof:
Let be i.i.d. uniformly random vectors, and put
, with addition done over
. Clearly (with probability 1), we have
for some finite time
. Let us define
, so that
is a partition into disjoint sets. Since each
is isomorphic to
, and
, no set
contains a non-trivial cycle. So if we define
, then we certainly get a spine.
To calculate , we will use a “charging” argument common to many random clustering analyses. If
, then we put
. Clearly we can write
, so it suffices to estimate
for a fixed
.
To this end, note that after conditioning on ,
is a uniformly random vertex of
, and thus
.
By linearity of expectation, it follows that .
So our task is reduced to finding a subset with
small and which contains no non-trivial cycles. A natural method for finding such a set
would be to prove a bound on the smallest non-zero eigenvalue of the Laplacian on
, and then use Cheeger’s inequality to conclude the existence of a good cut. But here we have to handle the special condition that
should contain no non-trivial cycles.
There is a very natural way to ensure this: If we impose that our eigenfunction vanishes on the boundary of (of course any other fundamental domain would do), then the standard “sweep” algorithm which chooses a level set of the eigenfunction will always find a set
that doesn’t wrap around (and therefore doesn’t contain any non-trivial cycles). The orange lines are examples of possible level sets.
Dirichlet boundary conditions for the Laplacian.
Fix a graph with
. We define its combinatorial Laplacian as the matrix
, where
is the diagonal degree matrix with
, and
is the adjacency matrix of
. Given a subset
, we can consider the first Dirichlet eigenvalue with boundary conditions on
:
.
By standard variational principles, there always exists a non-negative vector with
for
and
for
. If we take
, then this is the lowest-energy norm-1 function on
, subject to the boundary conditions.
Next, we need a version of Cheeger’s inequality for . As usual, we won’t actually need an eigenvector—any vector that has small Rayleigh quotient will do.
Discrete Cheeger inequality: If for all
, then
where is the maximum degree in
.
For the proof, I’ll refer to Theorem 4 in Alon-Klartag. Note that the proof for the Dirichlet version is even simpler than for the “standard” (Neumann) eigenvalues—it’s just a straightforward combination of Cauchy-Schwarz and the coarea formula.
Choosing a good eigenvector.
So to finish cheating at the odd-cycle game, we need only produce a low-energy vector on . We are aided greatly by the fact that this graph is a product of m-cycles, and thus we will only need to know about eigenvectors on the m-cycle
.
Let be the adjacency matrix of the m-cycle, and let
be the matrix obtained from it by replacing the last row and column by the zero vector (thus
is the adjacency matrix of the
-path, with an isolated vertex). Now, the adjacency matrix of
is simply
, where
denotes the k-fold tensor product of a matrix with itself. Note that if
satisfies
, then
(1)
It’s easy to see that the vector with
is an eigenvector of
with eigenvalue
and
. Therefore
is an eigenvector of
with eigenvalue
. Using (1), this implies that
Finally, using the fact that the Laplacian of is
, we see that
for .
Applying the Discrete Cheeger Inequality above, with implies that there exists a
, with
Since ,
contains no non-trivial cycles. Applying the Random Partitioning Lemma, and noting that the number of edges in
is
, we conclude that there exists a spine which cuts at most a
fraction of edges.
As discussed above, this finishes the proof that is the best possible exponent in the parallel repetition theorem (even for unique games), and thus this kind of gap amplification fails to mount an attack on the Unique Games Conjecture. In the next lecture, we’ll see a more benevolent use of eigenvectors.
Just a minor remark about the statement: “If the “sphere bound” were tight, […] the approach to proving equivalence of the UGC and MAX-CUT conjectures doesn’t even get off the ground.”
I guess, this is not quite accurate. The point is that the odd cycle is not a hard instance of MAX-CUT. So, Raz + KORW don’t rule out the approach you sketched in your first lecture. It could have still been the case that strong parallel repetition were true on a class of interesting (i.e. hard) instances of MAX-CUT. But you may have addressed this issue in your third lecture :-)
Hi Moritz,
The approach of Lecture 1 (that of Feige-Kindler-O’Donnell) was to prove a general parallel repetition theorem for unique games with exponent < 2, and use this to show that the MAX-CUT conjecture implies the UGC. Certainly this is ruled out by Raz’s example. As you say, it could still be the case that “hard instances” of MAX-CUT perform better under repetition. This will be ruled out in Lecture 3, via your SDP approach, but Lecture 3 is a bit delayed because I want to rework the presentation.
hello,
I know this is not the right place but I don’t know where to put my question
I heard a story about a student who slept in a lecture and the he woke up to find that the prof. wrote 2 math problems on the board and he thought that these two questions are homework so he tried to solve them and he solved only one, then he gave his answer to the prof. who said that this wasn’t a homework but it was a problem that scientest had NEVER proved it .. !!!
but the student solved it! because he tried his best (not expecting that this problem can’t be solved !)
is this a TRUE story ??
sorry for bothering you prof. .. but I don’t know where to ask as I mentioned above
Abdullah Abukar
Yes, your question has little to do with the post. But you are thinking of George Dantzig, the father of linear programming. See also a talk I gave once for some of his history.