$$\frac{1}{\mu(B)}\int_B \vert x \vert d\mu(x) = \frac{1}{p+1}$$

This formula holds for the unit ball in $\mathbb{Q_p}$. This formula also holds for $\mathbb{R}$ when $p=1$. Should one expect $$\mathrm{Frac}(W_{1^{\infty}} (\mathbb{F_1}))=\mathbb{R}?$$ What (mathematical) criteria do people use to rule-out field with one element phenomena? What makes point counting formulas better (or worse)?

  • 10
    $\begingroup$ What is $f$ in the formula that holds? $\endgroup$ – Mariano Suárez-Álvarez Jul 6 '11 at 0:58
  • 3
    $\begingroup$ It wouldn't hurt to explain your notation a bit more. Are you asking if the field of real numbers should be viewed as the ring of infinite Witt vectors on the field with one element? $\endgroup$ – S. Carnahan Jul 6 '11 at 2:05
  • $\begingroup$ Mariano: oops, $|x|$ is supposed to be the absolute value of a number. Carna: yes. $\endgroup$ – Taylor Dupuy Jul 6 '11 at 5:31
  • $\begingroup$ Googling for "Witt vectors + field with one element" turned up nothing promising, but did lead transitively to Manin's "Cyclotomy and analytic geometry over $\mathbb F_1$" (lanl.arxiv.org/abs/0809.1564). I am a bit mystified by a first reading, but the sentence "However, Witt's addition / multiplication becomes well defined only after extension to $\mathbb Z$" on p. 7 suggests that we don't yet know what, if anything, is the proper definition of the relevant ring of Witt vectors. (Also, one doesn't expect the ring of Witt vectors to be a field, I think ….) $\endgroup$ – LSpice Jul 6 '11 at 6:33
  • 3
    $\begingroup$ Connes has developed an analogue of the Witt functor in the setting of characteristic 1 that contains the field of real numbers in its image alainconnes.org/docs/henri65.pdf. Maybe this is related to your question... $\endgroup$ – user5831 Jul 6 '11 at 10:28

It is true that if $f(T)\in \mathbb Z[T]$ then,
$$ \frac{1}{\mu_p(B)}\int_{\mathbb Z_p} f(\vert x \vert_p) d\mu_p(x) \to \frac{1}{\vert B \vert}\int_{B}f(\vert x \vert)dx \mbox{ as } p\to 1.$$ It is not in general true that $$ \frac{1}{\mu_p(B)}\int_{\mathbb Z_p} \vert f( x )\vert_p d\mu_p(x) \to \frac{1}{\vert B \vert}\int_{B}\vert f( x )\vert dx \mbox{ as } p\to 1.$$

When $f(x) = x^2-1$ we have $$\int_{\mathbb Z_p} \vert x^2-1 \vert_p d\mu(x) = \frac{1+p(p-2)}{p}+ \frac{1}{(p+1)p} \to 1/2 \mbox{ as } p \to 1 $$ and $$\frac{1}{2}\int_{-1}^1 \vert x^2-1\vert dx = 2/3.$$ http://imgur.com/a/m6KYA

A couple remarks:

  1. You should have also written $\mathbb Q_1 = \mathbb{R}$ instead of that terrible notation $\mathrm{Frac}(W_{1^{\infty}}(\mathbb{F}_1))$.

  2. The question as posed is a little stupid since if someone had a procedure for "ruling our phenomena" not only would they would probably have category in mind, but they would be able to compute with it. I think the correct answer is $\mathbb F_1$ numerology is justified when it can be categorified''. This is kind of a weak version of the statement "a conjecture is true when you can prove it". I guess a vague question deserves a vague answer.

The first part of the question seems to be about Arakelov geometry and replacing the place at infinity with the place 1. It seems that the answer is no. Another bad question would be to ask for more examples where it does make sense.

The last part of the original question can be made more precise:

-Are there examples of formulas for $|X(\mathbb F_q)|$ such that $X$ is definable in either monoidal algebraic geometries or Borger's $\Lambda$-ring categorifications of schemes over $\mathbb F_1$ such that the point counting formula's in the category do not agree with $|X(\mathbb F_q)|$ as $q\to 1$? (if you know of another categorification that justifies $GL_n(\mathbb{F}_1)=S_n$ this question applies there too (Maybe Durov's Category?).)


Your Answer

By clicking "Post Your Answer", you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.