Abstract
In this paper, we consider fixed edgelength n-step random walks in . We give an explicit construction for the closest closed equilateral random walk to almost any open equilateral random walk based on the geometric median, providing a natural map from open polygons to closed polygons of the same edgelength. Using this, we first prove that a natural reconfiguration distance to closure converges in distribution to a Nakagami random variable as . We then strengthen this to an explicit probabilistic bound on the distance to closure for a random n-gon in any dimension with any collection of fixed edgelengths wi. Numerical evidence supports the conjecture that our closure map pushes forward the natural probability measure on open polygons to something very close to the natural probability measure on closed polygons; if this is so, we can draw some conclusions about the frequency of local knots in closed polygons of fixed edgelength.
Export citation and abstract BibTeX RIS
1. Introduction
Random walks in space with fixed edgelengths have been of interest to statistical physicists and chemists since Lord Rayleigh's day. These walks model polymers in solution (at least under θ-solvent conditions) [9, 14, 20] and are similarly interesting in computational geometry and mathematics as a space of 'linkages' [3, 16]. While 2- and 3-dimensional walks are the most relevant to this case, high-dimensional random walks often shed light on the lower dimensional situation [21].
In this paper, we will consider the relationship between open and closed random walks of fixed edgelengths (in polymeric language, the shapes of linear and ring polymers). We will provide an explicit algorithm for finding the nearest closed polygon with given edgelengths to almost any collection of edge directions, and use our construction to provide tail bounds on the fraction of polygon space within a fixed distance of the closed polygons in any dimension. Our results will be strongest for equilateral polygons, but provide explicit bounds for any collection of edgelengths.
To establish notation, we describe random walks in with (fixed) positive edgelengths wi by their edge clouds where is the direction of the ith edge. The space of polygonal arms is topologically equivalent to . If we let be the relative edgelengths, then we can define the submanifold of closed polygons .
Using Bernstein's inequality (e.g. [8]), there is an easy concentration inequality which suggests that the endpoints of random arms are close together. For equilateral polygons in , this takes the simple form
Theorem 1. If is chosen randomly in with edges ,
That is, the center of mass of a random edge cloud is very likely to be close to the origin. We can clearly close a random polygon in by subtracting the (small) from each edge. That closed polygon is clearly near the original arm, but it is no longer equilateral. This raises the question of whether we can generally close a polygon in (or ) while preserving edgelengths and changing the polygon only a small amount. This question is the focus of this paper.
Given and in , we view both as vectors in and measure the distance between them accordingly. We call this the chordal distance because it does not measure the arc on the spheres of radius wi for each pair of edges, but rather measures the straight line distance between edge vectors.
Our first main result is proposition 10, which shows that the chordal distance between a random and the nearest converges in distribution to a Nakagami- random variable with PDF proportional to as .
Our second main result is a general probabilistic bound on the chordal distance to closure for random polygons in any . For equilateral polygons in , our main theorem (corollary 19) takes the very simple form
for .
Here is a broad overview of our arguments. Given a polygon in , we will provide an explicit construction for a nearby closed polygon in , which we call the geometric median closure of (denoted ). It will be clear how to construct the geodesic in from to . For equilateral polygons, we show is the closest closed polygon to in chordal distance (theorem 8).
The distance between and depends on the norm of the geometric median (or Fermat–Weber point) of the edge cloud (proposition 12). For equilateral polygons, we will be able to leverage existing results of Niemiro [19] to find the asymptotic distribution of the geometric median of a random point cloud (proposition 9). Combining this with the matrix Chernoff inequalities proves our first main result (proposition 10).
The second main result follows from a concentration inequality for a random polygon in any , which bounds the probability of a large in terms of n, d, and w. This concentration result (theorem 18) follows from parallel uses of the scalar and matrix Bernstein inequalities to control the expected properties of a random edge cloud, together with the definition of the geometric median as the minimum of a convex function.
Last, we will observe that the pushforward measure on closed polygons obtained by closing random open polygons appears to converge rapidly to the uniform distribution on closed polygons (conjecture 22). Since these closures involve only very small motions of any part of the polygon, local features (such as small knots) should be preserved—it would follow (conjecture 23) that the rate of production of local knots in open and closed arcs should be almost the same.
2. Constructing a nearby closed polygon
As mentioned above, we view n-edge polygons (up to translation) in as collections of edge vectors4 . The vertices are obtained by summing the from an arbitrary basepoint. In this section of the paper, we will assume only that the lengths of the edges of the polygon are fixed to some arbitrary . We will think of these fixed edgelength polygons in two ways:
- as a weighted point cloud on the unit sphere where the points are denoted and the weights are the wi. We will call the edge cloud of the polygon.
- as a point (where Sd−1(r) is the sphere of radius r). We will call the vector of edges of the polygon.
The space of these polygons will be denoted . Within this space, there is a submanifold of closed polygons defined by the condition . (Equivalently, is closed if it lies in the codimension d subspace of normal to the , where is the standard basis in .) Both and are Riemannian manifolds with standard metrics, but it will be useful to use two additional metrics as well:
Definition 2. The chordal metric on is given by
The max-angular metric on is given by
We now make an important definition:
Definition 3. A geometric median (also known as a Fermat–Weber point) of an edge cloud is any which minimizes the weighted average distance function given by
where . To clarify notation, we will only use for points which are a geometric median of a weighted point cloud; the point cloud will be clear from the context.
This is a very old construction with a beautiful theory around it; see the nice review in [7]. We note that the geometric median differs from the center of mass (or geometric mean) of the points, which minimizes the weighted average of the squared distances between and the and that the geometric median is unique unless the points are all colinear and the geometric median is not one of the points.
This section is devoted to analyzing the following construction:
Definition 4. Suppose is a polygon and is a geometric median of its edge cloud which is not one of the . The geometric median closure of is the polygon whose edge cloud has the same weights and edge directions obtained by recentering the on and renormalizing: has edge cloud .
If every geometric median of is one of the , is not defined. If is defined, we say that is median-closeable.
Of course, we need to justify our choice of name by proving that is closed. The key observation is the following lemma, which follows by direct computation:
Lemma 5. The function is a convex function of . The gradient is given by
The Hessian of is given by
Proposition 6. If is median-closeable, is a unique closed polygon with edgelengths wi.
Proof. The proof follows from assembling several standard facts about the geometric median. These are in [18], but are easily checked by hand.
As it is a sum of convex functions, the average distance function is convex. Away from the , it is differentiable. If the points are not colinear, is strictly convex and is unique. If the points are colinear, either the geometric median is one of the or the set of geometric medians consists of the interval between two .
Any geometric median which is not one of the must be a critical point of the average distance function. For any such , using lemma 5,
This implies that is closed.
If is unique, then is obviously unique. If is not unique, the are colinear, and is on the line segment between two of the . In this case, it is not hard to see that (1) implies that the edges of are two antipodal groups of points on Sd−1, each containing n/2 points, regardless of where we take on the segment. □
Our next goal is to prove an optimality property for the geometric median closure. We will start by proving a more general fact about recentering and renormalizing:
Proposition 7. Let be a point cloud in , and take any not an . Given any set of weights wi we let denote the renormalized and recentered point cloud with weights wi and denote its weighted sum:
If and is the vector of edges corresponding to the edge cloud , then
that is, is the closest vector of edges to (in ) with edge weights wi and vector sum .
Proof. Suppose that is a point cloud with the same weights which also has and is the corresponding vector of edges in . Let . Since , we know .
Remembering that , we compute
Since is a positive scalar multiple of , this implies that , and so we have . Since , we see
or that . Using the facts and ,
so , as claimed. □
Note that the statement of proposition 7 is a little awkward for polygons which are not equilateral. We originally conjectured that was closest to , as the components of both are vectors in with the same lengths wi. But our proof shows that this conjecture was not true; instead, is closest to , whose components are all unit vectors.
This subtlety vanishes for equilateral polygons (where the wi = 1), which is the case of primary interest. In that case, combining propositions 6 and 7 with definition 4, we have
Theorem 8. If is a median-closeable equilateral polygon, is the closed equilateral polygon closest to in the chordal metric.
Remarks. This construction may seem unexpected, but it has deep roots. In [15], Kapovich and Millson provide an analogous closure construction which associates a unique closed equilateral polygon to any equilateral polygon where no more than half the edge vectors coincide by viewing the unit ball as the Poincaré ball model of hyperbolic space and (essentially) recentering and renormalizing in hyperbolic geometry around a point called the 'conformal median' (see [6]) which is in many ways parallel to the geometric median. This is an example of a 'Geometric Invariant Theory' (or GIT) quotient: see [13]. These ideas inspired our work above: we did not adopt them entirely only because working in hyperbolic geometry makes the whole endeavor seem much more abstract and because we have not managed to prove an optimality property for their construction analogous to theorem 8.
3. Asymptotics of the geometric median and the distance to closure
Now that we have established the connection between the geometric median and closure, we will establish some facts about the large-n behavior of the geometric median. Since the geometric median is a symmetric estimator of a large number of i.i.d. random variables, it seems natural to expect that the distribution of μ should converge to a multivariate normal, even though the classical central limit theorem does not apply. In fact, this is true:
Proposition 9. Let n points be sampled independently and uniformly on Sd−1, with geometric median . The random variable converges in distribution to as . This implies that converges in distribution to a Nakagami random variable.
Proof. We start by defining to be the expected distance from to the unit sphere; formally,
Observe that, by symmetry, the minimizer of is the origin. Now the geometric median of a finite collection of points uniformly sampled from the sphere is the minimizer of the average distance to the (definition 3). For a large number of points, we expect to be close to as a function, and hence that the minimizers of the functions should be nearby as well.
In fact, Niemiro studied exactly this situation, showing5 ([19, p 1517], see Haberman [11]) that
where is the covariance matrix of a random point on Sd−1 and is the Hessian of , evaluated at the origin.
The off-diagonal elements of are zero by symmetry. Using cylindrical coordinates on Sd−1 with axis , the ith diagonal entry in the covariance matrix is computed by the integral
We prove in the appendix (proposition A.1) that the expected distance function is given as a function of by
When d is odd, the standard Taylor series representation of the hypergeometric function truncates, and is a polynomial in r. For example, when d = 3 we have . In turn, a straightforward computation shows that the Hessian of evaluated at the origin is simply
where Id is the identity matrix. This completes the proof of the first statement. To get the second, we note that the norm of a Gaussian random variate is Nakagami-distributed. □
We now see that the geometric median is becoming asymptotically normal, and concentrating around the origin. In fact, numerical experiments show that the rate of convergence is rather fast (see figure 1). We can use this to prove an asymptotic result for the distance to closure for equilateral polygons.
Proposition 10. For a random equilateral n-gon with edges sampled independently and uniformly from Sd−1, the random variable converges in distribution to a Nakagami as .
Proof. We know from theorem 8 that is actually the chordal distance from to . To estimate this distance, we will make use of the recentering and renormalizing map from proposition 7.
When is small, we can estimate
where is the derivative of with respect to the vector in the direction of the unit vector (while leaving the variables constant).
Using the definition of , a direct computation reveals that
Since is a unit vector, the sum is the Rayleigh quotient for the matrix , and so obeys the estimates
Now and are also random variables depending on the , but we can use the matrix Chernoff inequalities [22, remark 5.3] to bound the probability that they are far from .
It's quite standard to prove that , so . The matrix Chernoff inequalities then reduce to
and
For any , the quantities raised to are <1, and so as the probability that the bounds in (2) and (3) both hold . In turn, this means that for any fixed ,
and so the random variable converges in probability to . By the continuous mapping theorem, this means that .
We can now rewrite the random variable as the product of , which by proposition 9 converges in distribution to a Nakagami random variable, and , which we have just proved converges in probability to the constant random variable .
Using Slutsky's theorem and a little algebra, this implies that the product converges in distribution to a Nakagami random variable, as claimed. □
We have now learned something interesting: the distribution of chordal distances to closure should be converging to a distribution which does not depend on the number of edges! This is surprising because the diameter of is clearly . This means that some arms might indeed be very far from closure—but they are very rare. We will look for this feature in the more specific probability inequalities to come.
We can also see how fast the tail of the distribution of dchordal can be expected to decay. The survival function of the Nakagami distribution is an incomplete Gamma function. Using [5, 8.10.1], we can show that there is a constant so that if x is Nakagami, then
Again, this is confirmed by numerical experiment for n = 1000, the empirical distribution of the distance to closure tracks the corresponding Nakagami distribution very well, see figure 2.
Download figure:
Standard image High-resolution image4. Concentration inequalities for and dchordal
We now know what to expect in the large-n limit, at least for equilateral polygons. For most applications in physics (4) is sufficient; even for small n, experimental evidence shows that the bound is quite close to the truth, see figure 1.
However, for mathematical applications, it's helpful to have a hard bound on the chordal distance to closure even if we sacrifice some accuracy. Learning from (4), we see that we should aim for a tail bound for dchordal which does not depend on n and is proportional to for some . We will get exactly such a bound in corollary 19 at the end of the section. Our bounds will apply for finite n, and also apply to the non-equilateral case, where it is not even clear what the large-n limit should mean.
4.1. A bound connecting and dchordal
To start with, we prove a hard bound on the relationship between the geometric median and our two measures of distance in polygon space. First, we note that our procedure of recentering and renormalizing changes each by a controlled amount.
Lemma 11. If and is any vector with , then and .
Proof. This is a calculus exercise; it is straightforward to establish the (sharp) bound
Further, it is easy to check that the right-hand side is a convex function of which is equal to 0 when , and when , so it is bounded above by the line . The angle bound is also straightforward. □
We now can give a bound on the distance between a given and in terms of the norm of the geometric median of the edge cloud .
Proposition 12. If the edge cloud has geometric median with ,
Proof. Since , our polygon is median-closeable and is a closed polygon with edge cloud . Lemma 11 immediately yields the bound on ; to get the chordal distance bound, we write
□
4.2. Strategy for the tail bound
To derive our explicit tail bound on the norm of the geometric median, our strategy is as follows. First, we will prove two probabilistic bounds: an upper bound on and a positive lower bound on . These will come from scalar and matrix versions of Bernstein's inequality.
If we restrict to a scalar function on a ray from the origin, these bounds yield an upper bound on and a lower bound on . We will get a uniform lower bound on for by showing that, on this interval. We prove this using the special structure of .
By Taylor's theorem, there is some z* in so that
This means that for , this directional derivative must be positive: in particular, since the geometric median is by definition a point where , can lie no farther than from the origin.
4.3. A probabilistic bound on
We want to bound the norm of the gradient , which we recall from lemma 5 is equal to . We start with a lemma which helps us understand the effect of variable weights .
Lemma 13. For any collection of n non-negative real numbers wi, if we define ,
where is the variance of . We have equality on the left precisely when all but one of the wi equal zero and equality on the right precisely when all the wi are equal.
Proof. Starting with the definition of variance, and remembering that ,
Solving for ,
which proves the central equality. Since with equality precisely when all the are equal, the inequality on the right follows easily.
To prove the inequality on the left, we invoke the Bhatia–Davis inequality [1], which says that since and the mean of the is , we have with equality precisely when one and the remainder are zero. □
Now we can give our first result:
Proposition 14. If we have n points sampled independently and uniformly from Sd−1, and n weights with and , then for any t > 0
If the are all equal (the polygon is equilateral), this simplifies to
Proof. We will use Bernstein's inequality [8, theorem 1.2]: suppose are independent random variables with for each i, the variance of each Xi is given by , and (with variance ). Then for any t > 0,
For any unit vector , we can set . These random variables clearly have expectation 0 and . Using cylindrical coordinates on Sd−1 with axis , the variance is computed by the integral
Using lemma 13, this implies
This proves that for any ,
Applying this inequality d times for , and using the union bound, we can bound the norm of :
But we know that for any we have , so
Replacing t by yields the statement of the proposition. □
The terms Ω and in the statement of proposition 14 at first seem mysterious. However, if you read them in light of lemma 13, they become clearer.
At one extreme, if one is close to 1 and the remaining are small, the sum regardless of n, and cannot concentrate on 0 as . To see this in the statement of the proposition, observe that in this case Ω and approach their maximum values of and , the n's in numerator and denominator cancel, and the exponent no longer depends on n at all.
At the other extreme, if the are all equal, Ω and are minimized: and . In this case, the denominator in the exponent does not depend on n and concentrates on 0 as fast as possible. We can compare this result to that of Khoi [17], who showed in a different sense that the equilateral polygons are the 'most flexible' of all the fixed edgelength polygons.
In the middle, if the are variable, but the number of comparably large increases, Ω and act to slow the rate of concentration, but they do not stop it: still concentrates on 0 as .
4.4. A probabilistic bound on
We now want to bound the lowest eigenvalue of the Hessian of at the origin. Again using lemma 5, we see that , where the quantities being summed are outer products of the vectors . That is, they are the symmetric, positive semidefinite projection matrices which project to the lines spanned by the . We now show
Proposition 15. If we have n points sampled independently and uniformly from Sd−1, and n weights with and , then for any t > 0
If the are all equal (the polygon is equilateral), this simplifies to
Proof. The statement is similar to the statement of proposition 14, so it should not be surprising that this also follows from a Bernstein inequality, this time for matrices [22, theorem 1.4]: suppose are independent random symmetric matrices, , , the 'matrix variance' of each Xi is given by , and (with 'scalar variance' ). Then for any t > 0,
We will set . These are clearly symmetric matrices and it is a standard computation to show that .
We now prove that . For any matrix A, the eigenvalues of A + kId are simply k added to the eigenvalues of A (see [12, theorem 2.4.8.1]). So
since the largest eigenvalue of a projection matrix like is 1.
Next, we want to show that . A direct computation reveals
and the result follows from our previous computation that . Summing the and taking the operator norm, we get
Plugging b and σ into (5) yields a bound on the probability that or, since , that . This completes the proof. □
We note that this concentration inequality is better than proposition 14: there is an extra factor of d in the numerator which means that the concentration gets faster as d increases. The effect of variable edgelengths is to slow (or stop) the concentration, just as in proposition 14; the same comments on the role of Ω and apply here.
4.5. A bound on the change in the radial second derivative
For any point , the second derivative of along the ray through is given by evaluating the Hessian as a quadratic form on the vector itself. Our last proposition gave us a lower bound on the result at the origin; we now show that this can not change too fast as we move away from the origin.
Since the fraction at right is increasing in , we can easily simplify the statement given a better upper bound on . In particular, for , the right-hand side .
Proof. Using lemma 5, we see that
Using the estimates and recalling that , we can underestimate the right hand side by
where the second part follows from finding a common denominator, expanding, and cancelling, using Cauchy–Schwartz carefully to underestimate the inner product terms as needed. Observing that allows us to underestimate , completing the proof. □
4.6. Bounding the norm of the geometric median
We are now in a position to bound the norm of the geometric median! This will proceed in two stages: first, we will use the Poincaré–Hopf index theorem to show that under certain hypotheses. Then we can immediately bootstrap to get a sharper bound.
Proposition 17. If , , and , then .
Proof. Given our hypothesis on of the Hessian, we know that the are not all colinear. This means that is the unique point inside Sd−1 where the vector field vanishes. We will now show that has a zero inside the sphere of radius ; by uniqueness, this point must be the geometric median.
Along any ray from the origin, we may restrict to a scalar function . Using proposition 16, on the interval our hypotheses imply
By Taylor's theorem, there is some so that
This means that the directional derivative of in the outward direction is positive on the boundary of the sphere of radius , or that points outward on this sphere. In particular, this implies that the vector field has index 1 on the sphere, and so by the Poincaré–Hopf index theorem must vanish at some point inside the sphere. □
We can now prove our main theorem.
Theorem 18. If we have n points sampled uniformly on Sd−1 (), n weights so that , and , then for any we have
If all the are equal (the polygon is equilateral)
For d = 3, we have the further simplification
Proof. We first define two random events: (event A) and (event B), which will happen for some choices of . Suppose both events occur.
As in proposition 17, we restrict to a scalar function on a ray; this time, the ray is assumed to pass through . By Taylor's theorem, if we evaluate at , there is some so that
Since we are assuming , the hypotheses of proposition 17 are satisfied and . In turn, this means that proposition 16 holds at z*, and
Since A, we have . As before, since is a directional derivative, it satisfies . We can plug both estimates into (7) and solve for , obtaining
If we call this event C, we have shown that , and hence that . This means that
Now was bounded above in proposition 15, while was bounded above in proposition 14. We now compare these upper bounds, noting that we have chosen in the statement of proposition 15 while the t in proposition 14 is smaller—less than . The bounds are
Of course, it suffices to compare the absolute values of the fractions inside the exponential functions (since both are negative). We can simplify the comparison by rewriting these as
It is now evident that if we compare the right fraction with the second fraction on the left, the numerator on the right is smaller and each term in the denominator is larger (recall t < t*). Multiplying by makes the left hand side even larger. Restoring the minus sign reverses this conclusion, and we see that our bound on is larger than our bound on , as claimed. Returning this conclusion to (8), we see , which is the first statement of the theorem.
The simplification when all the is an immediate consequence. To simplify to d = 3, we observe that ; substituting on the right hand side yields an expression in the form , where is a rational function bounded below by for . □
We now make a few remarks. First, if you carefully examine proposition 16, the lower bound on improves as . One can wring some extra information out of this, but the improvement in the final bound is minimal. Similarly, it is clear that one could set in our bound on without losing the conclusion, as long as t* < t. Again, this does not significantly improve things.
5. Distances and angles
We now want to restate our main theorem 18 in terms of the chordal and max-angular distance from a random arm to the nearest closed polygon using proposition 12.
Corollary 19. If we have n points sampled uniformly on Sd−1 (), n weights so that , and , then for any we have
If all the are equal (the polygon is equilateral), for we have
In dimension 3, this simplifies (again, for ), as
The problem with corollary 19 is that the hypotheses (on t) are disappointingly restrictive: for , we need to extend the domain of t to the point where the right-hand side becomes positive! On the other hand, numerical experiments (figure 3) comparing our bounds to experimental data and to the large-n Nakagami distribution proved in proposition 10 show that the conclusions of corollary 19 cannot be made much stronger. Further, these experiments suggest that, at least in the equilateral case, one should be able to entirely remove the upper bound on t—we leave this as
Download figure:
Standard image High-resolution imageConjecture 20. The conclusions of corollary 19 hold for any t > 0.
We now proceed to prove corollary 19.
Proof of corollary 19. Proposition 12 tells us , so to get a bound on the probability that we need to make the substitution on the right hand side of (6). Recalling that lemma 13 shows and carefully simplifying yields the first result.
For the second result, it follows immediately from the assumption that that the first result simplifies to
Using our upper bound on t, we see that the right hand side obeys
which immediately implies the second result.
For the third result, we simplify the fraction on the left hand side and substitute as we did above in the simplification of theorem 18; the complicated constant that results as the coefficient of t2 in the exponent is slightly less than . □
The statements for the maximum angular change in edge direction are similar, but somewhat easier to prove because the relationship between and the max-angular distance is simpler.
Corollary 21. If we have n points sampled uniformly on Sd−1 (), n weights so that , and , then for any we have
If all the are equal (the polygon is equilateral), for we have
In dimension 3, this simplifies (again, for ), as
Proof. We know from proposition 12 that . Since we're only going to apply this bound when (since and ), we can safely make the overestimate .
Substituting in (6) leads us to replace 3 by in the coefficient of nt2 in the numerator and 2 by in the coefficient of t in the denominator. Simplifying gives us the first statement.
To reach the second statement, we first observe that means and . Substituting these into the first statement (and overestimating the t in the denominator by ) yields the result.
Finally, the third statement (as in the proof of corollary 19) requires us to substitute to simplify the left-hand side. The resulting complicated coefficient of nt2 on the right-hand side is about . □
6. Discussion
From corollary 21 we see that closing a random arm is unlikely to change any edge very much. In particular, we should expect local features to be preserved by closure, as in the case of the local trefoil knot shown in figure 4. This suggests that closing up an arm is unlikely to destroy any local knots: in other words, the probability of local knotting in the standard measure on should be essentially the same as the probability of local knotting in the pushforward measure on via the map .
Download figure:
Standard image High-resolution imageOf course, this map is not defined on all of , but we know from theorem 18 that it is defined on all but an exponentially small fraction of ; pushing forward the restriction of the product measure to the domain of produces what we will call the pushforward measure on . On the other hand, the standard probability measure on is simply the volume measure induced by the Riemannian metric it inherits from . Since we have seen in corollaries 19 and 21 that almost all of is within a fixed distance of , it is reasonable to expect that this pushforward measure is close to the standard measure.
Indeed, this seems to be true. Rayleigh [20] showed that the distribution of end-to-end distances in a random element of is
We note that a closed form for is classical (see [14, 2.181]). Since a random closed polygon is formed from two random arms, conditioned on the hypothesis that their end-to-end distances are the same, the pdf of the length of the chord connecting vertices 0 and k in an polygon of n edges turns out to be given by
where the factor of comes from the fact that vertex k lies on a sphere of radius and is the volume of polygon space (which is known; see [2] for an identification between polygon space and a certain polytope which yields an explicit, though complicated, formula for ).
Therefore, the extent to which the distributions of the chordlengths match gives a sense of how close a given distribution on is to the standard one. For n = 4 and 5, we can see in figure 5 that the pushforward measure from is not particularly close to the standard measure. However, as n increases these statistics cannot distinguish between the pushforward measure and the standard measure; see figure 6.
Download figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution imageConjecture 22. As , the pushforward measure from to converges to the standard measure.
Assuming the truth of this conjecture implies that, at least for large n, random elements of look essentially like geometric median closures of random elements of . Since corollary 21 implies that individual edges are practically unchanged by closure, this would mean that all local phenomena happen at essentially the same rate in and .
When d = 3, a particularly important local phenomenon is that of local knotting. Say that a subsegment of is an r-local knot if it only intersects the boundary of a ball B of radius r at its endpoints and forms a knotted ball-arc pair. Let be the probability that a length-k arc of a random element of is an r-local knot, and similarly for .
Acknowledgments
This paper is a contribution to the Festschrift for Stu Whittington, a giant in the area of random polymers and random knots. We are indebted to Stu for years of insightful talks, perceptive questions, and remarkable mathematical results. His interest, enthusiasm, and explanation of the importance of these questions to the polymer science community have shaped our mathematical trajectory more than we can say.
We are also grateful for the continued support of the Simons Foundation (#524120 to Cantarella, #354225 to Shonkwiler), the German Research Foundation (DFG-Grant RE 3930/1–1, to Reiter), and the organizers of the 'Workshop on Topological Knots and Polymers' at Ochanomizu University, where key steps in the present work came together. In particular, we are indebted to Cristian Micheletti, Tetsuo Deguchi, Rob Kusner (for reducing everything to conformal geometry yet again!), Eric Rawdon, and Erik Schreyer for many helpful conversations. As always, we look to Yuanan Diao for inspiration—one of the motivations for this paper was the desire to find an alternate proof of [4].
Appendix. Proof of the hypergeometric formula for the expected distance to the sphere
Recall that is the expected distance from the point to the unit sphere. Since it is spherically symmetric, depends only on .
Proposition A.1. is given as a function of by
We can compute by evaluating at :
where the integrand only depends on the last coordinate xd of the point on Sd−1. Then the formula for will follow from a more general formula for functions on the sphere which only depend on a single coordinate:
Lemma A.2. Suppose depends only on xi; i.e. . Then
Proof. Let be projection to the ith coordinate. Then the projection of to the tangent space of Sd−1 has norm , and hence the smooth coarea formula implies that
Since is a -dimensional sphere of radius , which has area form , and since ϕ is constant on each level set, the above reduces to
as desired. □
Combining this result with (A.1) shows that
which is the starting point of our derivation of the hypergeometric formula.
Proof of proposition A.1. Using and the gamma function duplication formula , we can write the ratio of sphere volumes as
Substituting this into (A.2) and completing the square inside the square root yields
Making the substitution produces
which is the standard integral representation of . In turn, applying Kummer's quadratic transformation [10, 9.134.3] yields the desired formula
□
Footnotes
- 4
Throughout this paper, we use boldface to indicate elements of , which we usually think of as vectors of edge vectors. We use a superscript arrow—as in —to denote an arbitrary element of , though any such vector which is definitionally a unit vector we mark with a hat rather than an arrow.
- 5
Under some technical hypotheses which are obviously satisfied in our case.