Small baseball triangles

This all started when CJ asked which three baseball stadiums formed the smallest triangle.  And we agreed it had to be the Brewers, the White Sox, and the Cubs, because Milwaukee and Chicago are really close together.

But it seems like cheating to use two teams in the same city.  The most elegant way to forbid that is to ask the question one league at a time.  Which three American League parks form the smallest triangle?  And what about the National League?

First of all, what does “smallest” mean?  There are lots of choices, but (perhaps inspired by the summer we played a lot of Ingress) we asked for the triangle with the smallest area.  Which means you don’t just want the parks to be close together, you want them to be almost collinear!

I asked on Twitter and got lots of proposed answers.  But it wasn’t obvious to me which, if any, were right, so I worked it out myself!  Seamheads has the longitude and latitude of every major league ballpark past and present in a nice .csv file.  How do you compute the area of a spherical triangle given longitudes and latitudes?  You probably already know that the area is given by the excess over pi of the sum of the angles.  But then you gotta look up a formula for the angles.  Or another way:  Distance on the sphere is standard, and then it turns out that there’s a spherical Heron formula for the area of a spherical triangle given its edgelengths!  I guess it’s clear there’s some formula like that, but it’s cool how Heron-like it looks.  Fifteen lines of Python and you’re ready to go!

So what are the answers?

We were right that Brewers-White Sox-Cubs form the smallest major league triangle.  And the smallest American League triangle is not so surprising:  Red Sox, Yankees, Orioles, forming a shortish line up the Eastern Seaboard.  But for the National League, the smallest triangle isn’t what you might expect!  A good guess, following what happened in the AL, is Mets-Phillies-Nationals.  And that’s actually the second-smallest.  But the smallest National League triangle is formed by the Phillies, the Nationals, and the Atlanta Braves!  Here’s a piece the geodesic path from SunTrust Park in Atlanta to Citizen’s Bank Park in Philly, courtesy of GPSVisualizer:

Not only does it go right through DC, it passes about a mile and a half from Nationals Park!

Another fun surprise is the second-smallest major league triangle:  you’d think it would be another triangle with two teams in the same city, but no!  It’s Baltimore-Cincinnati-St. Louis.  Here’s the geodesic path from Oriole Park at Camden Yards to Busch Stadium:

And here’s a closeup:

The geodesic path runs through the Ohio River, about 300m from the uppermost bleachers at Great American Ball Park.  Wow!

Now here’s a question:  should we find it surprising that the smallest triangles involve teams that are pretty far from each other?  If points are placed at random in a circle (which baseball teams are definitely not) do we expect the smallest-area triangles to have small diameter, or do we expect them to be long and skinny?  It’s the latter!  See this paper: “On Smallest Triangles,” by Grimmet and Janson.  Put down n points at random in the unit circle; the smallest-area triangle will typically have area on order 1/n^3, but will have diameter on order 1.  Should that have been what I expected?

PS:  the largest-area major league triangle is Boston-Miami-SF.  Until MLB expands to Mexico City, that is!

Dissecting squares into equal-area triangles: idle questions

Love this post from Matt Baker in which he explains the tropical / 2-adic proof (in fact the only proof!) that you can’t dissect a square into an odd number of triangles of equal area.  In fact, his argument proves more, I think — you can’t dissect a square into triangles whose areas are all rational numbers with odd denominator!

• The space of quadrilaterals in R^2, up to the action of affine linear transformations, is basically just R^2, right?  Because you can move three vertices to (0,0), (0,1), (1,0) and then you’re basically out of linear transformations.   And the property “can be decomposed into n triangles of equal area” is invariant under those transformations.  OK, so — for which choices of the “fourth vertex” do you get a quadrilateral that has a decomposition into an odd number of equal-area triangles? (I think once you’re not a parallelogram you lose the easy decomposition into 2 equal area triangles, so I suppose generically maybe there’s NO equal-area decomposition?)  When do you have a decomposition into triangles whose area has odd denominator?
• What if you replace the square with the torus R^2 / Z^2; for which n can you decompose the torus into equal-area triangles?  What about a Riemann surface with constant negative curvature?  (Now a “triangle” is understood to be a geodesic triangle.)  If I have this right, there are plenty of examples of such surfaces with equal-area triangulations — for instance, Voight gives lots of examples of Shimura curves corresponding to cocompact arithmetic subgroups which are finite index in triangle groups; I think that lets you decompose the Riemann surface into a union of fundamental domains each of which are geodesic triangles of the same area.

Abraham Wald and the volume of the 4-simplex

Abraham Wald is one of the figures who keeps popping up in HOW NOT TO BE WRONG.  I just learned yet another cool thing about him which doesn’t have a place in the book, so I’m putting it here.

You know Heron’s formula, which gives you the area of a triangle in terms of the lengths of its edges.  And there’s a generalization:  the Cayley-Menger determinant is a formula for the volume of an n-simplex in terms of its edge lengths.

What about other kinds of faces?  The volume of a tetrahedron is determined by the six edge lengths, but not by the areas of its four faces.  Imagine four identical isosceles right triangles formed into a flattened tetrahedron that’s really just one square laid on top of another; that tetrahedron has volume 0, but a regular tetrahedron with faces of the same area obviously has positive volume.

What about a 4-simplex?  The lengths of the 10 edges determine the volume.  So you might guess that the areas of the 10 2-faces would give you a way to compute the volume, too.  But nope!  Wald gave an example of two 4-simplices with the same face-areas but different volume.  I wonder what the space of 4-simplices with fixed face-areas looks like.

David Mumford says we should replace plane geometry with programming and I’m not sure he’s wrong

MAA Mathfest is in Madison this week — lots of interesting talks about mathematics, lots of interesting talks about mathematics education.  Yesterday morning, David Mumford gave a plenary lecture in which he decried the lack of engagement between pure and applied mathematics — lots of nodding heads — and then suggested that two-column proofs in plane geometry should be replaced with basic programming — lots of muttering and concerned looks.

But there’s something to what he’s saying.  The task of writing a good two-column proof has a lot in common with the task of writing a correct program.  Both ask you to construct a sequence that accomplishes a complicated goal from a relatively small set of simple elements.  Both have the useful feature that there’s not a unique “correct answer” — there are different proofs of the same proposition, and different programs delivering the same output.  Both quite difficult for novices and both are difficult to teach.  Both have the “weakest link” feature:  one wrong step means the whole thing is wrong.

Most importantly:  both provide the training in formalization of mental process that we mathematicians mostly consider a non-negotiable part of general education.

But teaching coding instead of two-column proofs has certain advantages.  I am not, in general, of the opinion that everything in school has to lead to employable skills.  But I can’t deny that “can’t write five lines of code” closes more doors to a kid than “can’t write or identify a correct proof.”  People say that really understanding what it means to prove a theorem helps you assess people’s deductive reasoning in domains outside mathematics, and I think that’s true; but really understanding what it means to write a five-line program helps you understand and construct deterministic processes in domains outside a terminal window, and that’s surely just as important!

Computer programs are easier to check, for the teacher and more importantly the student — you can tell whether the program is correct by running it, which means that the student can iterate the try-check-fail-try-again process many times without the need for intervention.

And then there’s this:  a computer program does something.  When you ask a kid to prove that a right triangle is similar to the triangle cut off by an altitude to the hypotenuse, she may well say “but that’s obvious, I can just see that it’s true.”  And she’s not exactly wrong!  “I know you know this, but you don’t really know this, despite the fact that it’s completely clear” is a hard sell, it devalues the geometric intuition we should be working to encourage.

 

 

 

 

 

 

In which I agree with Pushkin

“Imagination is as necessary in geometry as it is in poetry.”

Rush Hour, Jr.

OK, so a black toddler and a Chinese toddler stumble on an international drug-trafficking ring — no, actually, this is a game I just bought for CJ, a kid’s version of Nob Yoshigahara‘s classic game Rush Hour.  The object here is to get the small white truck to the edge of the board (the top edge, in the image here.)  The trucks in your way can’t move sideways or turn; they just go forward and back.

You play a captivating game like this and naturally you start abstracting out the underlying math problem.  Play Set enough and you can’t avoid thinking about affine caps.  Rush Hour has more to do with the geometry of configuration spaces; it reminds me of the “disk in a box” problems that people like Persi Diaconis and Matt Kahle work on.

So here’s a question — it doesn’t capture all the features of Rush Hour, but let’s start here.  Let X be the unit square, and let c be a parameter between 0 and 1, and let N be a large integer.  Let L be the set of line segments in X which are either horizontal of the form y = i/N or vertical of the form x = i/N.  A traffic jam is a choice of a length-c interval in each of the 2N +2 line segments in L, where we require that these intervals be pairwise disjoint.  The traffic jams naturally form a topological space, which we call T(N,c).  We say an interval (x,i/n),(x+c,i/n) in a traffic jam t is trapped if no traffic jam in the connected component of t contains the interval (0,i/n),(c,i/n).

Questions: For which values of (N,c) is T(N,c) connected?  In particular, is it connected almost always once it’s nonempty?  If not, when does T(N,c) have a “giant component”?  If there’s an interesting range of parameters where T(N,c) is not connected, what proportion of intervals do we expect to be trapped?

Square pegs, square pegs. Square, square pegs.

Lately I’ve been thinking again about the “square pegs” problem:  proving that any simple closed plane curve has an inscribed square.  (I’ve blogged about this before: here, here, here, here, here.)  This post is just to collect some recent links that are relevant to the problem, some of which contain new results.

Jason Cantarella has a page on the problem with lots of nice pictures of inscribed squares, like the one at the bottom of this post.

Igor Pak wrote a preprint giving two elegant proofs that every simple closed piecewise-linear curve in the plane has an inscribed square.  What’s more, Igor tells me about a nice generalized conjecture:  if Q is a quadrilateral with a circumscribed circle, then every smooth simple closed plane curve has an inscribed quadrilateral similar to Q.  Apparently this is not always true for piecewise-linear curves!

I had a nice generalization of this problem in mind, which has the advantage of being invariant under the whole group of affine-linear transformations and not just the affine-orthogonal ones:  show that every simple closed plane curve has an inscribed hexagon which is an affine-linear transform of a regular hexagon.  This is carried out for smooth curves in a November 2008 preprint of Vrecica and Zivaljevic.  What’s more, the conjecture apparently dates back to 1972 and is due to Branko Grunbaum.  I wonder whether Pak’s methods supply a nice proof in the piecewise linear case.

Disks in a box, update

In April, I blogged about the space of small disks in a box.  One question I mentioned there was the following:  if C(n,r) is the space of configurations of n non-overlapping disks of radius r in a box of sidelength 1, what kind of upper bounds on r assure that C(n,r) is connected?  A recent preprint of Matthew Kahle gives some insight into this question: he produces configurations of disks which are stable (each disk is hemmed in by its neighbors) with r on order of 1/n.  (In particular, the density of such configurations goes to 0 as n goes to infinity.)  Note that Kahle’s configurations are not obviously isolated points in C(n,r); it could be, and Kahle suggests it is likely to be, that his configurations can be deformed by moving several disks at once.

Also appearing in Kahle’s paper is the stable 5-disk configuration at left; this one is in fact an isolated point in C(5,r).

More Kahle: another recent paper, with Babson and Hoffman, features the theorem that a random 2-complex on n vertices, where the edges are all present and each 2-face appears with probability p, transitions from non-simply-connected to simply connected when p crosses n^{-1/2}.  This is in sharp contrast with the H_1 of the complex with Z/ ell Z coefficients, which disappears almost surely once p exceeds 2 log n / n, by a result of Meshulam and Wallach.  So in some huge range, the fundamental group is almost surely a big group with no nontrivial abelian quotient!  (I guess this doesn’t formally follow from Meshulam-Wallach unless you have some reasonable uniformity in ell…)

One naturally wonders:  Let pi_1(n,p) be the fundamental group of a random 2-complex on n vertices with facial probability p.   If G is a finite simple group, what is the expected number of surjections from pi_1(n,p) to G?  Does it sharply transition from nonzero to zero?  Is there a range of p in which pi_1(n,p) is almost certainly an infinite group with no finite quotients?

“Every curve is a Teichmuller curve,” or “Why SL_2(Z) has the congruence subgroup property.”

A Teichmüller curve in M_g, the moduli space of genus-g curves, is an algebraic curve V in M_g such that the inclusion V -> M_g induces an isometry between the constant-curvature metric on V and the restriction of the Teichmüller metric on M_g.

Alternatively:  the cotangent bundle of M_g, considered as a real manifold, admits a natural action of SL_2(R); the orbits are all copies of SL_2(R) / SO(2), or the upper half-plane.  Most of the time, when you project that hyperbolic plane H down to M_g, you get a dense orbit that wanders all over M_g.  But every once in a while, the fibers of the map H -> M_g are a lattice in H, and the image is actually an algebraic curve; that, again, is a Teichmüller curve.

Teichmüller curves are the subject of lots of recent research; for now, let me just say that they are interestingly canonical curves inside M_g.  Matt Bainbridge proved strong results about their intersection numbers in Hilbert modular surfaces.  McMullen classified Teichmuller curves in M_2, giving a very nice algebraic description of the 1-parameter families of genus-2 curves parametrized by Teichmüller curves.  (As far as I know, there’s no such description in higher genus.)  In a recent note, McMullen proved that they are all defined over number fields.

This leads one to ask:  which curves defined over algebraic number fields are Teichmüller curves?  This is the subject of a paper Ben McReynolds and I just posted to the arXiv, “Every curve is a Teichmüller curve.”  The title should be read birationally; what we prove is that every curve X over an algebraic number field is birational (over C) to a Teichmüller curve in some M_g.  (In the posted version, we prove the slightly weaker statement that X is birational to a Teichmüller curve in M_{g,n}), but we’ve since tweaked the argument to get the closed-surface version.)

So why does SL_2(Z) have the congruence subgroup property?  Especially given that it, y’know, doesn’t?

Here’s what I mean.  Let Gamma_{g,n} be the mapping class group of a genus-g surface with n punctures.  Then Gamma_{g,n} acts as a group of outer automorphisms of the fundamental group pi_{g,n} of the surface; and from this, you get an action of Gamma_{g,n} on the finite set

Hom(pi_{g,n},G)/~

where G is a finite group and ~ is conjugacy.

By a congruence subgroup of Gamma_{g,n} let’s mean a stabilizer in this action.  Why this definition?  Well, when g = 1, n = 0, and G = Z/NZ, the stabilizer is just the standard congruence subgroup Gamma_0(N).  And you can easily check that the class of congruence subgroups of Gamma_{1,0} is cofinal with the usual class of congruence subgroups in SL_2(Z).

Now Gamma_{1,1} is also isomorphic to SL_2(Z), but the notion of “congruence subgroup of SL_2(Z)” afforded by this isomorphism is much more general than the usual one.  So much so that one gets the following, which is really the main point of my paper with Ben:

Every finite-index subgroup of Gamma_{1,1} containing the center and contained in Gamma(2) is a congruence subgroup.

It turns out that the finite covers of the moduli space M_{1,1} corresponding to such finite-index subgroups are always Teichmüller curves; since, by Belyi’s theorem, every curve over a number field can be so expressed, we get the desired result.

The italicized assertion above can be thought of as a very strong kind of “congruence subgroup property.”  Of course, CSP usually refers to the property that every finite-index subgroup contains a principal congruence subgroup.  That finite-index subgroups Gamma_{1,1} (and even Gamma_{1,n}) always contain congruence subgroups as defined above is a theorem of Asada, and it’s conjectured to be true for all g,n.  But the statement that every finite-index subgroup of a mapping class group is a congruence subgroup on the nose is substantially stronger, and I imagine it’s true only for (1,1) and the closely related case (0,4), which was proved, in somewhat different language, in the paper “Every curve is a Hurwitz space,” by Diaz, Donagi, and Harbater.  Our argument is very much inspired by theirs — it was to emphasize this debt that we gave our paper more or less the same title.

Cited!

We’re in a new era of mathematical publishing indeed; a paper posted yesterday on the arXiv cites a post from this blog as a reference.

The paper, by Strashimir Popvassiliev, constructs for every positive integer n a simple closed plane curve with exactly n inscribed squares. (It’s an old conjecture of Toeplitz that every simple closed plane curve contains at least one inscribed square.) This seems to speak against philosophy, mentioned by Denne in her guest post here, that “the reason” every curve has at least one inscribed square is because every curve has an odd number of inscribed squares.

I’m not sure Popvassiliev’s example really contradicts this philosophy. Surely the squares should be counted with multiplicity, in the appropriate sense. With a more naive notion of “counting” you can’t expect parity conditions to hold. For instance, you certainly want to say that a straight line intersects a smooth closed curve an even number of times. Naively, you might complain that a tangent line to a circle intersects the circle only once! But of course, it really crosses the curve twice; it’s just that the two crossings are at the same point.