## August 31, 2020

### Sphere Spectrum Analogue of PGL(2,Z)

#### Posted by John Baez Since I’ve been thinking about continued fractions I’ve been thinking about $PGL(2,\mathbb{Z})$, the group of transformations

$z \mapsto \frac{a z + b}{c z + d} , \qquad a,b,c,d \; \text{s.t.} \; a d - b c \ne 0$

mod its center. You can think of this as a group of transformations of the integral form of the projective line. When we see something like

$\frac{\sqrt{5} + 1}{2} = 1 + \frac{1}{1 + \frac{1}{1 + \frac{1}{1 + \ddots}}}$

or even

$\frac{\pi}{4} = \frac{1}{1 + \frac{1^2}{2 + \frac{3^2}{2 + \frac{5^2}{2 + \ddots}}}}$

But in modern mathematics the sphere spectrum is what Joyal called the “true integers”: it’s the initial ring spectrum just as $\mathbb{Z}$ is the initial ring. So there should be an enhanced version of $PGL(2,\mathbb{Z})$ with the sphere spectrum taking over the role of the integers, and a lot should be known about it.

What’s it called, and what do people know about it?

Posted at August 31, 2020 5:27 AM UTC

TrackBack URL for this Entry:   https://golem.ph.utexas.edu/cgi-bin/MT-3.0/dxy-tb.fcgi/3246

### Re: Sphere Spectrum Analogue of PGL(2,Z)

People also consider $P S L(2, \mathbb{Z})$ in the context of continued fractions, such as Todd Trimble here (though there with regard to the presentation $a_1 - 1/(a_2 - 1/(a_3 - 1/...$).

What does the difference between $P G L$ and $P S L$ amount to in this context?

Posted by: David Corfield on August 31, 2020 8:05 AM | Permalink | Reply to this

### Re: Sphere Spectrum Analogue of PGL(2,Z)

Perhaps I should have focused on $PSL(2,\mathbb{Z})$ because that’s the one number theorists always talk about. A sphere spectrum analogue of $PSL(2,\mathbb{Z})$ would make me perfectly happy.

James Dolan brainwashed me into thinking $PGL(2,-)$ is more fundamental because it’s the full symmetry group of the projective line, while $PSL(2,-)$ consists of orientation-preserving symmetries of the projective line.

I have to think about these things every time because the details depend on the commutative ring you put in the $-$, but I think $PSL(2,\mathbb{Z})$ is an index-two subgroup of $PGL(2,\mathbb{Z})$. The point is that the determinant of an invertible integer matrix must be either $1$ or $-1$, and the matrix is in $SL$ if the determinant $1$. For example

$\left(\begin{array}{cc} 1 & 0 \\ 0 & -1 \end{array} \right)$

is a guy in $GL(2,\mathbb{Z})$ that’s not in $SL(2,\mathbb{Z})$, which gives an element of $PGL(2,\mathbb{Z})$ that’s not in $PSL(2,\mathbb{Z})$ .

Posted by: John Baez on August 31, 2020 6:21 PM | Permalink | Reply to this

### Re: Sphere Spectrum Analogue of PGL(2,Z)

Hmm, can one define $\mathbb{P}^1$ of the sphere spectrum $\mathbb{S}$? This would probably be a spectral scheme. Then its automorphism group could be one form of $PGL(2,\mathbb{S})$

Posted by: David Roberts on August 31, 2020 11:43 AM | Permalink | Reply to this

### Re: Sphere Spectrum Analogue of PGL(2,Z)

With regard to the question in your first sentence, this is discussed for example in sections 5.4 and 19.2.6 of Lurie’s Spectral Algebraic Geometry.

Posted by: Richard Williamson on September 1, 2020 11:50 AM | Permalink | Reply to this

### Re: Sphere Spectrum Analogue of PGL(2,Z)

Here’s a thought. I claim that the functor $R \mapsto B\mathrm{PGL}(n,R)$, from commutative rings to connected pointed spaces, is naturally the restriction of a functor defined on $E_\infty$ ring spectra.

For a commutative ring $R$, the tensor product of $R$-modules induces a map of pointed spaces $B\mathrm{GL}(n,R) \times B\mathrm{GL}(m,R) \to B\mathrm{GL}(n m,R)$. This is part of a higher structure, making $B\mathrm{GL}(1,R)$ into an $E_\infty$-space whose underlying $E_1$-space acts on $B\mathrm{GL}(n,R)$. Hence we may form the homotopy orbit space $(B\mathrm{GL}(n,R))_{h B\mathrm{GL}(1,R)}$, which I think is naturally pointed homotopy equivalent to $B\mathrm{PGL}(n,R)$.

This should make sense for $E_\infty$ ring spectra as well, for example $B\mathrm{GL}(1,R)$ is an $E_\infty$-space, the basepoint component of what is nowadays called $\mathrm{Pic}(R)$. It acts on the space $B\mathrm{GL}(n,R)$ by $\otimes_R$, so we obtain a functor $R \mapsto (B\mathrm{GL}(n,R))_{h B\mathrm{GL}(1,R)}$ from $E_\infty$ ring spectra to connected pointed spaces.

It seems one could take $B\mathrm{PGL}(n,R) := (B\mathrm{GL}(n,R))_{h B\mathrm{GL}(1,R)}$ and $\mathrm{PGL}(n,R) = \Omega B\mathrm{PGL}(n,R)$. I don’t know how to justify whether those are good definitions though.

Posted by: anonymous on August 31, 2020 3:46 PM | Permalink | Reply to this

Post a New Comment