neverendingbooks

In Belgium the hashtag-craze of the moment is #cestjoelle. Joelle Milquet is perceived to be the dark force behind everything, from the crisis in Greece, over DSK, to your mother-in-law coming over this weekend? #cestjoelle.

Sam Leith used the same meme in his book the coincidence engine.

A hurricane assembling a passenger jet out of old bean-cans? #cestGrothendieck

All shops in Alabama out of Chicken & Broccoli Rica-A-Roni? #cestGrothendieck

Frogs raining down on Atlanta? #cestGrothendieck

As this is a work of fiction, Alexandre Grothendieck‘s name is only mentioned in the ‘author’s note’:

“It is customary to announce on this page that all resemblances to characters living or dead are entirely coincidental. It seems only courteous to acknowledge, though, that in preparing the character of Nicolas Banacharski I was inspired by the true-life story of the eminent mathematician Alexandre Grothendieck.”

The name ‘Nicolas Banacharski’ is, of course, chosen on purpose (the old Bourbaki NB-joke even makes an appearance). The character ‘Isla Holderness’ is, of course, Leila Schneps, the ‘Banacharski ring’ is, of course, the Grothendieck circle. But, I’d love to know the name of the IRL-’Fred Nieman’, who’s described as ‘an operative for the military’.

Sam Leith surely knows all the Grothendieck-trivia which shouldn’t come as a surprise because he wrote in 2004 a piece for the Spectator on the ‘what is a metre?’ incident (see also this n-category cafe post).

The story of ‘the coincidence engine’ is that Grothendieck did a double (or was it triple) bluff when he dropped out of academia in protest of military money accepted by the IHES. He went into hiding only to work for a weapons company and to develop a ‘coincidence bomb’. As more and more unlikely events happen during a car-ride by a young Cambridge postdoc though the US (to propose to his American girlfriend), the true Grothendieck-aficianado (and there are still plenty of them in certain circles) will no doubt begin to believe that the old genius succeeded (once again) and that Ana’s (Grothendieck’s mother) $\infty$-ring is this devilish (pun intended) device…

However,

“There was no coincidence engine. Not in this world. It existed only in Banacharski’s imagination and in the imaginations he touched. But there was a world in which it worked, and this world was no further than a metre from our own. Its effect spilled across, like light through a lampshade.

And with that light there spilled, unappeased and peregrine, fragments of any number of versions of an old mathematician who had become his own ghost. Banacharski was neither quite alive nor quite dead, if you want the truth of it. He was a displaced person again, and nowhere was his home.”

Note to students following this year’s ‘seminar noncommutative geometry’ : the seminar starts friday september 30th at 13h in room G 0.16.

However, if you have another good reason to be in Ghent on thursday september 22nd, consider attending the inaugural lecture of Koen Thas at 17h in auditorium Emmy Noether, campus De Sterre, Krijgslaan 281, 9000 Gent.

Koen’s lecture has one of the longest titles i’ve seen : “De lange weg – een verhaal over wiskundige problemen die denkers al eeuwenlang teisteren, zonderlingen die in afgelegen berghutten de existentie van de duivel willen aantonen, en een mythisch object dat niet bestaat, maar waar we toch naar zoeken” (“The long road – a story on mathematical problems torturing scientists for centuries, weirdos trying to prove the existence of the Devil in desolated mountain-huts and the search for a mythical object that doesn’t exist”).

Knowing Koen a bit I’d say it will be on the Riemann hypothesis, Grothendieck’s theory of motives and the field with one element. A sneak preview of our upcoming seminar, quoi?

More information on the event and to register see here.

Last time we discovered that the mental picture to view prime numbers as knots in $S^3$ was first dreamed up by David Mumford. Today, we’ll focus on where and when this happened.

3. When did Mazur write his unpublished preprint?

According to his own website, Barry Mazur did write the paper Remarks on the Alexander polynomial in 1963 or 1964. A quick look at the references gives us a coarse lower- and upper-estimate.

Apart from a paper by Iwasawa and one by Milnor, all references predate 1962 giving us a lower-bound. More interesting is reference (14) to David Mumford’s Geometric Invariant Theory (GIT) which was first published in 1965 and is referred to as ‘in preparation’, so the paper was written no later than 1965. If we look a bit closer we see than some GIT-references are very precise

indicating that Mazur must have had the final version of GIT to consult, making it rather difficult to believe that the preprint was written late 1963 or early 1964.

Mazur’s dating of the preprint is probably based on this penciled note on the frontpage of the only surviving copy of the preprint

It reads : “Date from about 63/64, H.R. Morton”. Hugh Morton of Liverpool University confirms that it is indeed his writing on the preprint.

Further, he told me that early 64 Christopher Zeeman held a Topology Symposium in Cambridge UK, where Hugh was a graduate student at the time and, as far as he could recall, Mazur attended that conference and gave him the preprint on that occasion, whence the 63/64 dating. Hugh kindly offered to double-check this with Terry Wall who cannot remember Mazur attending that particular conference.

In fact, we will see that a more correct dating of the Mazur-preprint will be : late 1964 or early 1965.

4. The birthday : July 10th 1964

Clearly, Mumford’s insight predates the Mazur-preprint. In the first section, Mazur mentions ‘Grothendieck cohomology groups’ rather than ‘Etale cohomology groups’.

At the time, Artin’s seminar notes on Grothendieck topologies (spring 1962) were widely distributed, and Artin and Grothendieck were in the process of developing etale cohomology in their Paris 1963/64 seminar SGA 4, while Mumford was working on GIT in Harvard.

Mike Artin, David Mumford and Jean-Louis Verdier all attended the Woods Hole conference from july 6 till july 31 1964, famous for producing the Atiyah-Bott fixed point theorem (according to Fulton first proved by Verdier at the conference).

Etale cohomology was a hot topic at that conference. On july 10th there were three talks, Artin spoke on ‘Etale cohomology of schemes’, Verdier on ‘A duality theorem in the etale cohomology of schemes’ and John Tate on ‘Etale cohomology over number fields’.

After a first week of talks, more informal seminars were organized, including the Atiyah-Bott seminar leading to the ‘Woods hole duality theorem’ and one by Lubin-Tate and Serre on elliptic curves and formal groups. Two seminars adressed Etale Cohomology.

Artin and Verdier ran a seminar on the etale cohomology of number fields leading to their duality result, and, three young turks : Daniel Quillen, Steve Kleiman and Robin Hartshorne ran a Baby Seminar on Etale cohomology

Probably it is safe to say that the talks by Artin, Verdier and Tate on July 10th sparked the primes=knots idea, and if not then, a couple of days later.

5. The birthplace : the Whitney Estate

The ‘Woods Hole’ conference took place at the Whitney Estate and all the lectures took place in the rustic rooms of the main building and the participants (and their families) were housed in rented cottages in the neighborhood, for the duration of the summer.

The only picture i managed to find from the Whitney house comes from a rather surprising source : Gardeners and Caretakers ofWoods Hole. Anyway, here it is :

Probably, the knots=primes analogy was first dreamed up inside, or in the immediate neighborhood, on a walk to or from the cottages, overlooking the harbor.

Previously, we have recalled comparisons between approaches to define a geometry over the absolute point and art-historical movements, first those due to Yuri I. Manin, subsequently some extra ones due to Javier Lopez Pena and Oliver Lorscheid.

In these comparisons, the art trend appears to have been chosen more to illustrate a key feature of the approach or an appreciation of its importance, rather than giving a visual illustration of the varieties over $\mathbb{F}_1$ the approach proposes.

Some time ago, we’ve had a couple of posts trying to depict noncommutative varieties, first the illustrations used by Shahn Majid and Matilde Marcolli, and next my own mental picture of it.

In this post, we’ll try to do something similar for affine varieties over the absolute point. To simplify things drastically, I’ll divide the islands in the Lopez Pena-Lorscheid map of $\mathbb{F}_1$ land in two subsets : the former approaches (all but the $\Lambda$-schemes) and the current approach (the $\Lambda$-scheme approach due to James Borger).

The former approaches : Francis Bacon “The Pope” (1953)

The general consensus here was that in going from $\mathbb{Z}$ to $\mathbb{F}_1$ one looses the additive structure and retains only the multiplicative one. Hence, ‘commutative algebras’ over $\mathbb{F}_1$ are (commutative) monoids, and mimicking Grothendieck’s functor of points approach to algebraic geometry, a scheme over $\mathbb{F}_1$ would then correspond to a functor

$h_Z~:~\mathbf{monoids} \longrightarrow \mathbf{sets}$

Such functors are described largely by combinatorial data (see for example the recent blueprint-paper by Oliver Lorscheid), and, if the story would stop here, any Rothko painting could be used as illustration.

Most of the former approaches add something though (buzzwords include ‘Arakelov’, ‘completion at $\infty$’, ‘real place’ etc.) in order to connect the virtual geometric object over $\mathbb{F}_1$ with existing real, complex or integral schemes. For example, one can make the virtual object visible via an evaluation map $h_Z \rightarrow h_X$ which is a natural transformation, where $X$ is a complex variety with its usual functor of points $h_X$ and to connect both we associate to a monoid $M$ its complex monoid-algebra $\mathbb{C} M$. An integral scheme $Y$ can then be said to be ‘defined over $\mathbb{F}_1$’, if $h_Z$ becomes a subfunctor of its usual functor of points $h_Y$ (again, assigning to a monoid its integral monoid algebra $\mathbb{Z} M$) and $Y$ is the ‘best’ integral scheme approximation of the complex evaluation map.

To illustrate this, consider the painting Study after Velázquez’s Portrait of Pope Innocent X by Francis Bacon (right-hand painting above) which is a distorded version of the left-hand painting Portrait of Innocent X by Diego Velázquez.

Here, Velázquez’ painting plays the role of the complex variety which makes the combinatorial gadget $h_Z$ visible, and, Bacon’s painting depicts the integral scheme, build up from this combinatorial data, which approximates the evaluation map best.

All of the former approaches more or less give the same very small list of integral schemes defined over $\mathbb{F}_1$, none of them motivically interesting.

The current approach : Jackson Pollock “No. 8″ (1949)

An entirely different approach was proposed by James Borger in $\Lambda$-rings and the field with one element. He proposes another definition for commutative $\mathbb{F}_1$-algebras, namely $\lambda$-rings (in the sense of Grothendieck’s Riemann-Roch) and he argues that the $\lambda$-ring structure (which amounts in the sensible cases to a family of endomorphisms of the integral ring lifting the Frobenius morphisms) can be viewed as descent data from $\mathbb{Z}$ to $\mathbb{F}_1$.

The list of integral schemes of finite type with a $\lambda$-structure coincides roughly with the list of integral schemes defined over $\mathbb{F}_1$ in the other approaches, but Borger’s theory really shines in that it proposes long sought for mystery-objects such as $\mathbf{spec}(\mathbb{Z}) \times_{\mathbf{spec}(\mathbb{F}_1)} \mathbf{spec}(\mathbb{Z})$. If one accepts Borger’s premise, then this object should be the geometric object corresponding to the Witt-ring $W(\mathbb{Z})$. Recall that the role of Witt-rings in $\mathbb{F}_1$-geometry was anticipated by Manin in Cyclotomy and analytic geometry over $\mathbb{F}_1$.

But, Witt-rings and their associated Witt-spaces are huge objects, so one needs to extend arithmetic geometry drastically to include such ‘integral schemes of infinite type’. Borger has made a couple of steps in this direction in The basic geometry of Witt vectors, II: Spaces.

To depict these new infinite dimensional geometric objects I’ve chosen for Jackson Pollock‘s painting No. 8. It is no coincidence that Pollock-paintings also appeared in the depiction of noncommutative spaces. In fact, Matilde Marcolli has made the connection between $\lambda$-rings and noncommutative geometry in Cyclotomy and endomotives by showing that the Bost-Connes endomotives are universal for $\lambda$-rings.

One of the more surprising analogies around is that prime numbers can be viewed as knots in the 3-sphere $S^3$. The motivation behind it is that the (etale) fundamental group of $\pmb{spec}(\mathbb{Z}/(p))$ is equal to (the completion) of the fundamental group of a circle $S^1$ and that the embedding

$\pmb{spec}(\mathbb{Z}/(p)) \subset \pmb{spec}(\mathbb{Z})$

embeds this circle as a knot in a 3-dimensional simply connected manifold which, after Perelman, has to be $S^3$. For more see the what is the knot associated to a prime?-post.

In recent months new evidence has come to light allowing us to settle the genesis of this marvelous idea.

1. The former consensus

Until now, the generally accepted view (see for example the ‘Mazur-dictionary-post’ or Morishita’s expository paper) was that the analogy between knots and primes was first pointed out by Barry Mazur in the middle of the 1960′s when preparing for his lectures at the Summer Conference on Algebraic Geometry, at Bowdoin, in 1966. The lecture notes where later published in 1973 in the Annales of the ENS as ‘Notes on etale cohomology of number fields’.

For further use in this series of posts, please note the acknowledgement at the bottom of the first page, reproduced below : “It gives me pleasure to thank J.-P. Serre for his vigorous editing and his suggestions and corrections, which led to this revised version.”

Independently, Yuri I. Manin spotted the same analogy at around the same time. However, this point of view was quickly forgotten in favor of the more classical one of viewing number fields as analogous to algebraic function fields of one variable. Subsequently, in the mid 1990′s Mikhail Kapranov and Alexander Reznikov took up the analogy between number fields and 3-manifolds again, and called the resulting study arithmetic topology.

2. The new evidence

On december 13th 2010, David Feldman posted a MathOverflow-question Mazur’s unpublished manuscript on primes and knots?. He wrote : “The story of the analogy between knots and primes, which now has a literature, started with an unpublished note by Barry Mazur. I’m not absolutely sure this is the one I mean, but in his paper, Analogies between group actions on 3-manifolds and number fields, Adam Sikora cites B. Mazur, Remarks on the Alexander polynomial, unpublished notes.“

Two months later, on february 15th David Feldman suddenly found the missing preprint in his mail-box and made it available. The preprint is now also available from Barry Mazur’s website. Mazur adds the following comment :

“In 1963 or 1964 I wrote an article Remarks on the Alexander Polynomial [PDF] about the analogy between knots in the three-dimensional sphere and prime numbers (and, correspondingly, the relationship between the Alexander polynomial and Iwasawa Theory). I distributed some copies of my article but never published it, and I misplaced my own copy. In subsequent years I have had many requests for my article and would often try to search through my files to find it, but never did. A few weeks ago Minh-Tri Do asked me for my article, and when I said I had none, he very kindly went on the web and magically found a scanned copy of it. I’m extremely grateful to Minh-Tri Do for his efforts (and many thanks, too, to David Feldman who provided the lead).”

The opening paragraph of this unpublished preprint contains a major surprise!

Mazur points to David Mumford as the originator of the ‘primes-are-knots’ idea : “Mumford has suggested a most elegant model as a geometric interpretation of the above situation : $\pmb{spec}(\mathbb{Z}/p\mathbb{Z})$ is like a one-dimensional knot in $\pmb{spec}(\mathbb{Z})$ which is like a simply connected three-manifold.”

In a later post we will show that one can even pinpoint the time and place when and where this analogy was first dreamed-up to within a few days and a couple of miles.

For the impatient among you, have a sneak preview of the cradle of birth of the primes=knots idea…