3
$\begingroup$

Is it true that any finite dimensional division algebra over a pseudo-algebraically closed field is trivial? We know that this is true for algebraically closed field.

$\endgroup$
4
  • 3
    $\begingroup$ Presumably the answer below has to do with the following issue. Some mathematicians use "division algebra over $k$" to mean a finite dimensional $k$-algebra that is a division algebra and whose center equals $k$. That is presumably also what you intend. However, other mathematicians make no such hypothesis about the center of the division algebra. In that case, also field extensions of $k$ are "division algebras over $k$". The term "central simple algebras" clarifies this, because "central" is part of the name. $\endgroup$ Aug 24 '17 at 13:03
  • $\begingroup$ @JasonStarr Yeah, if I remember correctly, EGA or SGA makes this distinction between R-rings, R-algebras, and R-extensions, the first being noncentral ring maps, the second being central ring maps, and the third being central maps into commutative rings in the section about something involving derived functors, possibly the degree 1 Quillen-Andre ones. $\endgroup$ Aug 24 '17 at 16:39
  • $\begingroup$ Oops ran out of time to edit it, but yeah it's in the section on ExAlComm $\endgroup$ Aug 24 '17 at 16:45
  • $\begingroup$ $\operatorname{EGA}_0 \S 18$ got it for ya $\endgroup$ Aug 24 '17 at 17:11
6
$\begingroup$

The fact that every finite dimensional division algebra is trivial is not only true over an algebraically closed field but it is in fact equivalent to the field being algebraically closed. Remember that (extension-)fields are just a special case of division algebras. Thus, if it would be true over pseudo-algebraically closed fields, we would have that a field is algebraically closed if and only if it is pseudo-algebraically closed.
However, this is not the case, as for example finite fields are pseudo-algebraically closed.

$\endgroup$
3
  • $\begingroup$ It seems from the proof of Proposition $6.2.3$, page $171$ of "Central simple algebras and Galois cohomology" by Gille and Szamuely, that this is true for $C_1$-fields. But I am not an expert. $\endgroup$
    – user43198
    Aug 24 '17 at 12:53
  • 3
    $\begingroup$ @user43198: yes, if you consider division algebras whose center is the given field, then these things are classified by the Brauer group, and the Brauer group is trivial for PAC fields. $\endgroup$ Aug 24 '17 at 12:55
  • $\begingroup$ No. As mentioned in the answer, finite fields are $C_1$, and every extension field of them is such an algebra with itself (so not the base field) as center. $\endgroup$
    – Dirk
    Aug 24 '17 at 13:10
4
$\begingroup$

The fact that the Brauer group of a pseudo-algebraically closed field is trivial is noted in: J.Ax, The elementary theory of finite fields, Annals Math. 88 (1968), p. 269. A more detailed proof is given in: M. Fried and M. Jarden, Field Arithmetic (3rd edition), Springer, p. 209.

$\endgroup$

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.