Wild division algebras over Laurent series fields , , November 27, 2006

In this paper we study some class of division algebras over a Laurent series field with arbitrary residue field. Namely, we study division algebras which satisfy the following condition: there exists a section

\bar{D} ↪ D

of the residue homomorphism

D \to \bar{D}

, where

D

is a central division algebra over a complete discrete valued field

F = k ((t))

. We say that these division algebras are splittable. If

c h a r k = 0

, all such division algebras are tame and therefore belong to the group of tame division algebras, which was carefully studied in the papers [6] and [10] even in a much more general situation of a henselian field

F

of arbitrary characteristic. So, we consider mostly wild division algebras.

An extensive analysis of the wild division algebras of degree

p

over a field

F

with complete discrete rank 1 valuation with

c h a r (\bar{F}) = p

was given by Saltman in [11] (Tignol in [13] analyzed more general case of the defectless division algebras of degree

p

over a fild

F

with Henselian valuation). Here we study splittable division algebras of arbitrary index. This class (which is not a subgroup in

B r (F)

) contains a class of good splittable division algebras (see the definition in section 2), which posess several beautiful properties.

In particular, we prove a decomposition theorem for such algebras. This theorem is a generalization of the decomposition theorems for tame division algebras given by Jacob and Wadsworth in [6] .

For arbitrary splittable division algebras we give only several assorted results, and the study of this class is far from to be complete. Nevertheless, we investigate here technical tools, which are important for the study of such algebras, and prove a relation between the level and a higher order level for some splittable division algebras (see section 6).

We hope this technique will be applied to the study of the cyclisity question for certain division algebras od degree

p^{k}

As an application we get several results, which are partly well known (see proposition 6 ) and party not. In particular, we get the positive answer on the following conjecture:

the exponent of

A

is equal to its index for any division algebra

A

over a

C_{2}

-field

F = F_{1} ((t_{2}))

, where

F_{1}

is a

C_{1}

-field.

Here is a brief overview of this paper.

In section 2 we give a definition of splittable and good splittable division algebras and prove that all tame division algebras over

F = k ((t))

are good splittable.

Section 3 contains the most important technical tools for the study of splittable division algebras. We define a notion of

δ

-maps and investigate a theory of

δ

-maps for such algebras. In this section we define also the notion of a local height, which is a possible generalization of Saltman's level.

In section 4 we prove the period-index conjecture metioned above. This section contains also a small history of the question known to the author. We note that the proof does not use all the results from section 3.

In section 5 we study good splittable division algebras and prove the decomposition theorem.

In section 6 we reprove some results of Saltman about semiramified division algebras of index

p

over

F

using the technique from section 3. Then we define a notion of a higher order level and prove several general properties of splittable division algebras satisfying the following condition:

Z (\bar{D}) / \bar{F}

is a simple extension. At the end of section we putseveral open questions.

We use the notation of [6] . We always denote by

D

a division algebra finite dimensional over its center

F = k ((t)) = Z (D)

. Recall that any Henselian valuation on

F

has a unique extension to a valuation on

D

. We denote the valuation on

F

v

and its unique extension on

D

w

Given a valuation

w

D

, we denote by

Γ_{D}

its value group, by

V_{D}

its valuation ring, by

M_{D}

its maximal ideal and by

\bar{D} = V_{D} / M_{D}

its residue division ring.

By [12] , p.21 one has the fundamental inequality

[D : F] \geq | Γ_{D} : Γ_{E} | \cdot [\bar{D} : \bar{F}] .

D

is called defectless over

F

if equality holds and defective otherwise. It is known that

D

is defectless if it has a discrete valuation of rank 1.

Jacob and Wadsworth in [6] introduced the basic homomorphism

θ_{D} : Γ_{D} / Γ_{F} \to G a l (Z (\bar{D}) / \bar{F})

induced by conjugation by elements of

D

. They showed that

θ_{D}

is surjective and

Z (\bar{D})

is the compositum of an abelian Galois and a purely inseparable extension of

\bar{F}

We say

D

is tame division algebra if

c h a r (\bar{F}) = 0

c h a r (\bar{F}) = q \neq 0

D

is defectless over

F

Z (\bar{D})

is separable over

\bar{F}

, and

q |̸ | k e r (θ_{D}) |

. We say

D

is wild division algebra if it is non tame.

We call a division algebra

D

inertially split if

Z (\bar{D})

is separable over

\bar{F}

, the map

θ_{D}

is an isomorphism, and

D

is defectless over

F

Acknowledgements I am grateful to Professor A. N. Parshin, Professor E.-W. Zink, and M. Grabitz for useful discussions and attention to my work. I am very grateful to Professor A.Wadsworth for carefully reading my paper and for showing me a mistake in the very first version of this paper and to Professor V.I.Yanchevskii for valuable discussions during his visit in Berlin. Finally, I thank my wife Olga for her support and encouragement.

2 Cohen's theorem

Recall one definition from [14] .

Definition 1 A division algebra

D

is said to be splittable if there is a homomorphism

\bar{D} ↪ O_{D} \subset D

that is a section of the map

O_{D} \to \bar{D}

There is a natural question if there exists a generalization of Cohen's theorem, i.e. is any central division algebra splittable or not. It is not true if a division algebra is not finite dimensional over its centre, as Dubrovin's example in [14] shows. It is not true also for some finite dimensional division algebras, as the example to theorem 2.7. in [11] shows.

But it is true for tame division algebras over complete discrete valued fields. This easily follows from results of Jacob and Wadsworth [6] (compare with [14] , Th.1).

Theorem 1 Let

(F, v)

be a valued field which is complete with respect to a discrete rank 1 valuation

v

. Suppose

c h a r F = c h a r \bar{F}

. Let

D

be a tame division algebra with

Z (D) = F

and

[D : F] < \infty

Then there exists a section

\bar{D} ↪ D

of the residue homomorphism

D \to \bar{D}

Proof. Since

F

is a complete field,

F

is a Henselian field and

v

extends uniquely to a valuation

w

D

. Since

D

is tame,

Z (\bar{D}) / \bar{Z (D)}

is a cyclic Galois extension. There exists an inertial lift

Z

Z (\bar{D})

over

F

Z

is Galois over

F

, and by classical Cohen's theorem there exists a section

\tilde{Z} (\bar{D}) ↪ Z

Consider the centraliser

C = C_{D} (Z)

Z

D

. Then we have

\bar{C} = \bar{D}

Indeed, by Double Centraliser Theorem we have

[D : F] = [C : F] [Z : F]

and

[Z : F] = | G a l (Z (\bar{D}) / \bar{F}) |

. By [6] , prop.1.7 a homomorphism

θ_{D} : Γ_{D} / Γ_{F} \to G a l (Z (\bar{D}) / \bar{F})

is surjective, so for any parameter

z

we have

θ_{D} (w (z)) = σ

, where

< σ > = G a l (Z (\bar{D}) / \bar{F})

It is clear that

z / \in C

. Now let

u_{1}, \dots, u_{[C : F]}

be a

F

-basis of

C

. It is easy to see that the elements

u_{j}, z u_{j}, \dots, z^{n - 1} u_{j}

j = 1, \dots, [C : F]

, where

n = o r d (σ)

, the order of

σ

, are linearly independent, so form a basis for

D

over

F

. Since

w (F 〈 z u_{j}, \dots, z^{n - 1} u_{j}, j = 1, \dots, [C : F] 〉) \cap Γ_{C} = 0,

where

F 〈 z u_{j}, \dots, z^{n - 1} u_{j}, j = 1, \dots, [C : F] 〉

denote a vector space in

D

over

F

generated by elements

u_{j} z^{i}

, this implies that for any element

x \in D

with

w (x) = 0

we can find elements

r_{1}, \dots r_{[C : F]} \in F

such that

x = r_{1} u_{1} + \dots + r_{[C : F]} u_{[C : F]} mod M_{D}

Hence

\bar{C} = \bar{D}

Note that

C

is an unramified division algebra. Indeed, by [6] , th.2.8, th.2.9

C

contains a copy of the inertial lift of a maximal separable subfield in

\bar{C}

, say

\tilde{C}

. Then the centralizer

C_{C} (\tilde{C})

must be a totally ramified division algebra, i.e. it is trivial and

\tilde{C}

is a maximal subfield. So,

C

must be unramified.

Fix an embedding

i : \bar{F} ↪ F

. It can be extended to the embedding

i^{'} : \bar{Z} ↪ Z

i^{'} |_{\bar{F}} = i

by Hensel lemma. Now consider the algebra

A = \bar{C} \otimes_{\bar{Z}} Z (C)

. It is easy to see that

A

is an unramified division algebra with

\bar{A} = \bar{C} = \bar{D}

. Therefore by [3] , Th.31,

A \sim = C

; so there exists a section

\bar{D} ↪ C

The theorem is proved.

□

Later we will see that much more can be said about good splittable algebras:

Definition 2 A division algebra

D

is called good splittable if there exists a section

s : \bar{D} ↪ D

compatible with an embedding

i : \bar{Z (D)} ↪ Z (D)

, i.e.

s (\bar{Z (D)}) = i (\bar{Z (D)}) \subset Z (D)

It's easy to see that all tame division algebras are good splittable, because by Hensel lemma any embedding

\bar{Z (D)} ↪ Z (D)

can be uniquely extended to any separable extension of

Z (D)

It is interesting to know what kind of splittable division algebras are good splittable.

By theorem 3.9. in [11] even a splittable division algebra

D

of degree

p = c h a r D

is not a good splittable algebra if the level of

D

(the notion of level we will recall in section 3, see remark to lemma 7 ) is divisible by

p

. Nevertheless, it is an open question whether it is true or not, for example, for division algebras with

\bar{D} = Z (\bar{D})

such that

\bar{D} / \bar{F}

is a simple extension and the local height (see the definition in the same remark) is not divisible by

p

. We will discuss this question in section 6.

3 Delta-maps of splittable algebras

In this section we develop some ideas from [14] , where some properties of

δ

-maps for special kind of local skew fields were studied. Technical properties of

δ

-maps play the main role in all our results. Here we will give a list of these properties.

Let

D

be a finite dimensional division algebra over a complete valued field

F = k ((t))

Let

w

be a unique extension of the valuation

v

D

. We will denote by

z

any parameter of

D

, i.e. any element with

〈 w (z) 〉 = Γ_{D}

. Consider the ring

Z 〈 α, δ 〉

of noncommutative polinomials in two variables. Define the map

σ : Z 〈 α, σ 〉 \to Z 〈 α, δ, δ_{i}; i \geq 1 〉,

σ (α^{a_{1}} δ^{b_{1}} \dots α^{a_{n}} δ^{b_{n}}) = α^{a_{1}} δ_{b_{1}} \dots δ_{b_{n - 1}} α^{a_{n} - 1} δ^{b_{n}},

where

a_{1}, b_{n} \geq 0

a_{i}, b_{j} \geq 1

i > 1

j < n

for every word in

Z 〈 α, δ 〉

Let

S_{i}^{k} \in Z 〈 α, δ 〉

i \geq k

i \geq 1

be polynomials given by the following formula:

S_{i}^{k} = \sum_{τ \in S_{i} / G} τ ({\underset{︸}{α \dots α}}_{i - k} {\underset{︸}{δ \dots δ}}_{k}),

where

S_{i}

is a permutation group and

G

is an isotropy subgroup.

Lemma 1 ([14] , lemma 2) The polynomials

S_{i}^{k}

satisfy the following property:

S_{i}^{i} = δ^{i}, S_{i}^{0} = α^{i}, S_{i + 1}^{k + 1} = α S_{i}^{k + 1} + δ S_{i}^{k}

For any splittable division algebra can be defined a notion of

δ

-maps:

Proposition 1 ([14] , prop. 1,2) Let

D

be a splittable division algebra. Fix some parameter

z

and some embedding

u : \bar{D} ↪ D

. Then

D

is isomorphic to a division algebra

\bar{D} ((z))

, which is defined to be the vector space of series with multiplication defined by the formula

z a z^{- 1} = α (a) + δ_{1} (a) z + δ_{2} (a) z^{2} + \dots, a \in \bar{D},

where

α : \bar{D} \to \bar{D}

is an automorphism and

δ_{i} : \bar{D} \to \bar{D}

are linear maps such that the map

δ_{i}

satisfy the identity

δ_{i} (a b) = \sum_{k = 0}^{i} σ (δ^{i - k} α) (a) σ (S_{i}^{k} α) (b), a, b \in \bar{D}

Remark Note that the values

σ (S_{i}^{k} α)

and

σ (δ^{i - k} α)

belong to the subring

Z 〈 α, δ_{i}, i \geq 1 〉

, so the formula is well defined.

Note that

δ

-maps depend on the choice of a parameter and an embedding. The automorphism

α

, as it easy to see, depend only on the choice of a parameter. In the proposition we identify

\bar{D}

with

u (\bar{D})

Corollary 1 ([14] , corol. 1) Suppose

α = I d

. Then

δ_{i} (a b) = δ_{i} (a) b + \sum_{k = 1}^{i} δ_{i - k} (a) \sum_{(j_{1}, \dots, j_{l})} C_{i - k + 1}^{l} δ_{j_{1}} \dots δ_{j_{l}} (b),

where

δ_{0} = α

and the second sum is taken over all the vectors

(j_{1}, \dots, j_{l})

such that

0 < l \leq m i n {i - k + 1, k}

j_{m} \geq 1

\sum j_{m} = k

Further we will need even more general definition.

Definition 3 In the situation of proposition 1 let us define maps

_{m}^{(z, u)} δ_{i} : \bar{D} \to \bar{D}

m \in Z

i \in N

as follows.

z^{m} a z^{- m} = u (^{(z)} α^{m} (\bar{a})) + u (_{m}^{(z, u)} δ_{1} (\bar{a})) z + u (_{m}^{(z, u)} δ_{2} (\bar{a})) z^{2} + \dots, a \in u (\bar{D}) .

m = 0

, put

_{m}^{(z, u)} δ_{i} = 0

Note that

^{(z)} α |_{Z (\bar{D})}

does not depend on the choice of

z

Note that if

^{(z)} α = i d

, then

_{m}^{(z, u)} δ_{i} = 0

for

m = p^{k}

, where

k

is sufficiently large,

k

depends on

i

. Moreover,

_{m}^{(z, u)} δ_{i} =_{m + p^{k}}^{(z, u)} δ_{i}

for

k

sufficiently large. We will use also the following notation:

_{m}^{(z, u)} {\tilde{δ}}_{i} =_{- m}^{(z, u)} δ_{i},_{1}^{(z, u)} δ_{i} =^{(z, u)} δ_{i}

Sometimes, we will write

_{m} δ_{i}

instead of

_{m}^{(z, u)} δ_{i}

and

_{m}^{(z, u)} δ_{i} (a)

instead of

u (_{m}^{(z, u)} δ_{i} (\bar{a}))

whenever the context is clear.

Immediately from the definition follows

Lemma 2 In the situation of definition 3 we have (i) for

| m | > 1

_{m}^{(z, u)} δ_{i} (a) =^{(z)} α^{s i g n (m)} (_{s i g n (m) (| m | - 1)}^{(z, u)} δ_{i} (a)) +_{s i g n (m)}^{(z, u)} δ_{i} (^{(z)} α^{s i g n (m) (| m | - 1)} (a)) +

{\sum_{j = 1}^{i - 1}}_{s i g n (m)}^{(z, u)} δ_{j} (_{s i g n (m) (| m | - 1)}^{(z, u)} δ_{i - j} (a)),

where

s i g n (m) = m / | m |

a \in \bar{D}

; (ii) for any

m \neq 0

^{(z)} α^{- m} (_{m}^{(z, u)} δ_{i}) +_{- m}^{(z, u)} δ_{i} (^{(z)} α^{m}) + {\sum_{j = 1}^{i - 1}}_{- m}^{(z, u)} δ_{j} (_{m}^{(z, u)} δ_{i - j}) = 0

Proposition 2 For fixed

z, u

from proposition 1 we have (i) The maps

_{m}^{(z, u)} δ_{i}

satisfy the following identities:

_{m} δ_{i} (a b) =_{m} δ_{i} (a) α^{i + m} (b) + α^{m} (a)_{m} δ_{i} (b) + {\sum_{k = 1}^{i - 1}}_{m} δ_{i - k} (a)_{i - k + m} δ_{k} (b)

(ii) Suppose

α = i d

. Then the maps

_{m}^{(z, u)} δ_{i}

satisfy the following identities:

_{m} δ_{i} (a b) =_{m} δ_{i} (a) b + a_{m} δ_{i} (b) + {\sum_{k = 1}^{i - 1}}_{m} δ_{i - k} (a) \sum_{(j_{1}, \dots, j_{l})} C_{i - k + m}^{l} δ_{j_{1}} \dots δ_{j_{l}} (b)

where the second sum is taken over all the vectors

(j_{1}, \dots, j_{l})

such that

0 < l \leq m i n {i - k + m, k}

j_{m} \geq 1

\sum j_{m} = k

;

C_{j}^{k} = 0

j = 0

, and

C_{j}^{k} = C_{j + p^{q}}^{k}

for

q > > 0

j \leq 0

Proof. For any

a, b \in \bar{D}

we have

α^{m} (a b) z^{m} +_{m} δ_{1} (a b) z^{m + 1} +_{m} δ_{2} (a b) z^{m + 2} + \dots = z^{m} (a b) =

\begin{matrix} (α^{m} (a) z^{m} +_{m} δ_{1} (a) z^{m + 1} +_{m} δ_{2} (a) z^{m + 2} + \dots) b \end{matrix}

(1)

If we represent the right-hand side of ( 1 ) as a series with coeffitients shifted to the left and then compare the corresponding coeffitients on the left-hand side and right-hand side, we get some formulas for

_{m} δ_{i} (a b)

. We have to prove that these formulas are the same as in our proposition.

Let

z^{i + m - k} b = α^{i + m - k} (b) z^{i + m - k} + \dots + x_{k}^{'} z^{i + m} + \dots

and

(α^{m} (a) z^{m} +_{m} δ_{1} (a) z^{m + 1} +_{m} δ_{2} (a) z^{m + 2} + \dots) b = α^{m} (a b) z^{m} + y_{m + 1} z^{m + 1} + y_{m + 2} z^{m + 2} + \dots

Then we have

y_{i + m} = α^{m} (a) x_{i}^{'} + {\sum_{k = 0}^{i - 1}}_{m} δ_{i - k} (a) x_{k}^{'}

In the proof of [14] , prop.2 we have shown that

z^{i + 1 - k} b = α^{i + 1 - k} (b) z^{i + 1 - k} + \dots + σ (S_{i}^{k} α) (b) z^{i + 1} + \dots

Hence

x_{k}^{'} = σ (S_{i + m - 1}^{k} α) (b)

for

k < i

. It is easy to see that

x_{i}^{'} =_{m} δ_{i} (b)

x_{0}^{'} = α^{i + m} (b)

and

σ (S_{i + m - 1}^{k} α) =_{i + m - k} δ_{k}

, which proves (i).

For

α = i d

, by corollary 1 ,

σ (S_{i + m - 1}^{k} α) (b) = \sum_{(j_{1}, \dots, j_{l})} C_{i - k + m}^{l} δ_{j_{1}} \dots δ_{j_{l}} (b),

where

l, j_{1}, \dots, j_{l}

were defined in our proposition. This proves (ii).

The proposition is proved.

□

Lemma 3 ([14] , lemma 3 ) In the situation of proposition 1 suppose

_{i}^{(z, u)} δ_{j}

is the first map such that

_{i}^{(z, u)} δ_{j} (a) \neq 0

for given

a \in \bar{D}

i \in Z \ {0}

, i.e.

_{i}^{(z, u)} δ_{1} (a) = \dots =_{i}^{(z, u)} δ_{j - 1} (a) = 0

_{i}^{(z, u)} δ_{j} (a) \neq 0

(so we have a map

i \mapsto j (i)

). Then (i) for

z^{'} = z + u (b) z^{q + 1}

b \in \bar{D}

we have

^{(z^{'})} α^{i} (a) =^{(z)} α^{i} (a)

_{i}^{(z^{'}, u)} δ_{k} (a) =_{i}^{(z, u)} δ_{k} (a)

for

k < q

and

_{i}^{(z^{'}, u)} δ_{q} (a) =_{i}^{(z, u)} δ_{q} (a) + {b^{'}}^{(z)} α^{q + i} (a) -^{(z)} α^{i} (a) b^{'},

where

b^{'} = {\sum_{k = 0}^{i - 1}}^{(z)} α^{k} (b)

(ii) Suppose

^{(z)} α^{n} |_{Z (\bar{D})} = i d

n \geq 1

a \in Z (\bar{D})

and

_{1}^{(z, u)} δ_{1} (^{(z)} α^{k} (a)) = \dots =_{1}^{(z, u)} δ_{j - 1} (^{(z)} α^{k} (a)) = 0

for any

k

Then for

z^{'} = z + u (b) z^{q + 1}

b \in \bar{D}

we have

^{(z^{'})} α^{i} (a) =^{(z)} α^{i} (a)

_{i}^{(z^{'}, u)} δ_{k} (a) =_{i}^{(z, u)} δ_{k} (a)

for

k < q + j

and

_{i}^{(z^{'}, u)} δ_{q + j} (a) =_{i}^{(z, u)} δ_{q + j} (a) + {b^{'}}^{(z)} α^{q} (_{i}^{(z, u)} δ_{j} (a)) -_{i}^{(z, u)} δ_{j} (a)^{(z)} α^{j} (b^{'}) +

b^{'} {\sum_{k = 1}^{q}}^{(z)} α^{q - k} (^{(z, u)} δ_{j} (^{(z)} α^{k + i - 1} (a))) -_{i}^{(z, u)} δ_{j} (a) {\sum_{k = 0}^{j - 1}}^{(z)} α^{k} (b),

where

b^{'} = {\sum_{k = 0}^{i - 1}}^{(z)} α^{k} (b)

, if

n | q

^{(z)} α (a) = a

In particular, if

^{(z)} α = i d

and

(i, p) = 1

, then

_{i}^{(z^{'}, u)} δ_{q + j} (a) =_{i}^{(z, u)} δ_{q + j} (a) + (q - j)_{i}^{(z, u)} δ_{j} (a) b

(iii) for

z^{'} = u (b) z

b \in Z (\bar{D})

b \neq 0

we have

^{(z^{'})} α (a) =^{(z)} α (a)

^{(z^{'}, u)} δ_{k} (a) =^{(z, u)} δ_{k} (a)

for

k < j

and

^{(z^{'}, u)} δ_{j} (a) =^{(z, u)} δ_{j} (a)^{(z)} α (b^{- 1}) \dots^{(z)} α^{j} (b^{- 1})

i = 1

Proof. (i) We have

{z^{'}}^{i} a {z^{'}}^{- i} = (1 + b^{'} z^{q} + \dots) z^{i} a z^{- i} (1 + b^{'} z^{q} + \dots)^{- 1} = (z^{i} a z^{- i} + b^{'} z^{q} z^{i} a z^{- i} + \dots) (1 - b^{'} z^{q} + \dots) =

(z^{i} a z^{- i} - z^{i} a z^{- i} b^{'} z^{q} + \dots + b^{'} z^{q} z^{i} a z^{- i} - \dots) =

(z^{i} a z^{- i} - [^{(z)} α^{i} (a) +_{i}^{(z, u)} δ_{j} (a) z^{j} + \dots] b^{'} z^{q} + b^{'} z^{q} [^{(z)} α^{i} (a) +_{i}^{(z, u)} δ_{j} (a) z^{j} + \dots] + \dots) =

(z^{i} a z^{- i} - [^{(z)} α^{i} (a) b^{'} +_{i}^{(z, u)} δ_{j} (a)^{(z)} α^{j} (b^{'}) z^{j} + \dots] z^{q} + {b^{'}}^{(z)} α^{q + i} (a) z^{q} + \dots) =

(z^{i} a z^{- i} + (-^{(z)} α^{i} (a) b^{'} + {b^{'}}^{(z)} α^{q + i} (a)) z^{q} + \dots) =^{(z)} α^{i} (a) + \dots +_{i}^{(z, u)} δ_{q - 1} (a) {z^{'}}^{q - 1} +

(_{i}^{(z, u)} δ_{q} (a) + {b^{'}}^{(z)} α^{q + i} (a) -^{(z)} α^{i} (a) b^{'}) {z^{'}}^{q} + \dots

(ii) Put

c = {z^{'}}^{i} z^{- i} - 1 - b^{'} z^{q + i}

. So,

w (c) > q + i

. Note that

c^{(z)} α^{k} (a) =^{(z)} α^{k} (a) c

, since

n | q

^{(z)} α (a) = a

and

a \in Z (\bar{D})

. We have

{z^{'}}^{i} a {z^{'}}^{- i} = (1 + b^{'} z^{q} + c) z^{i} a z^{- i} (1 + b^{'} z^{q} + c)^{- 1} = (z^{i} a z^{- i} + b^{'} z^{q} z^{i} a z^{- i} + c z^{i} a z^{- i}) (1 + b^{'} z^{q} + c)^{- 1} =

(^{(z)} α^{i} (a) +_{i}^{(z, u)} δ_{j} (a) z^{j} + \dots +_{i}^{(z, u)} δ_{q + j} (a) z^{q + j} + \dots + b^{'} z^{q} (^{(z)} α^{i} (a) +_{i}^{(z, u)} δ_{j} (a) z^{j} + \dots)) (1 + b^{'} z^{q} + c)^{- 1} =

(^{(z)} α^{i} (a) + {b^{'}}^{(z)} α^{q + i} (a) z^{q} +^{(z)} α^{i} (a) c +_{i}^{(z, u)} δ_{j} (a) z^{j} + \dots +_{i}^{(z, u)} δ_{q + j} (a) z^{q + j} + \dots +

b^{'} \sum_{k = 1}^{q} (^{(z)} α^{q - k} (^{(z, u)} δ_{j} (^{(z)} α^{k + i - 1} (a)))) z^{q + j} + b^{'} (^{(z)} α^{q} (_{i}^{(z, u)} δ_{j} (a))) z^{q + j} + \dots) (1 + b^{'} z^{q} + c)^{- 1} =

^{(z)} α^{i} (a) + [_{i}^{(z, u)} δ_{j} (a) z^{j} + \dots +_{i}^{(z, u)} δ_{q + j} (a) z^{q + j} + \dots + b^{'} \sum_{k = 1}^{q} (^{(z)} α^{q - k} (^{(z, u)} δ_{j} (^{(z)} α^{k + i - 1} (a)))) z^{q + j} +

b^{'} (^{(z)} α^{q} (_{i}^{(z, u)} δ_{j} (a))) z^{q + j} + \dots)] (1 - b^{'} z^{q} - c + \dots) =

^{(z)} α^{i} (a) +_{i}^{(z, u)} δ_{j} (a) z^{j} + \dots +_{i}^{(z, u)} δ_{q + j} (a) z^{q + j} + \dots + b^{'} \sum_{k = 1}^{q} (^{(z)} α^{q - k} (^{(z, u)} δ_{j} (^{(z)} α^{k + i - 1} (a)))) z^{q + j} +

b^{'} (^{(z)} α^{q} (_{i}^{(z, u)} δ_{j} (a)) z^{q + j} + \dots -_{i}^{(z, u)} δ_{j} (a)^{(z)} α^{j} (b^{'}) z^{q + j} + \dots =

^{(z)} α^{i} (a) + \dots +_{i}^{(z, u)} δ_{q + j - 1} (a) {z^{'}}^{q + j - 1} + (_{i}^{(z, u)} δ_{q + j} (a) + {b^{'}}^{(z)} α^{q} (_{i}^{(z, u)} δ_{j} (a)) -_{i}^{(z, u)} δ_{j} (a)^{(z)} α^{j} (b^{'})

+ b^{'} {\sum_{k = 1}^{q}}^{(z)} α^{q - k} (^{(z, u)} δ_{j} (^{(z)} α^{k + i - 1} (a))) -_{i}^{(z, u)} δ_{j} (a) {\sum_{k = 0}^{j - 1}}^{(z)} α^{k} (b)) {z^{'}}^{q + j} + \dots,

since

{z^{'}}^{j} = z^{j} + {\sum_{k = 0}^{j - 1}}^{(z)} α^{k} (b) z^{q + j} + \dots

(iii) We have

z^{'} a {z^{'}}^{- 1} = b z a z^{- 1} b^{- 1} =^{(z)} α (a) + b^{(z, u)} δ_{j} (a)^{(z)} α^{j} (b^{- 1}) z^{j} + \dots =

^{(z)} α (a) +^{(z, u)} δ_{j} (a)^{(z)} α (b^{- 1}) \dots^{(z)} α^{j} (b^{- 1}) {z^{'}}^{j} + \dots,

since

^{(z^{'})} α |_{Z (\bar{D})} =^{(z)} α |_{Z (\bar{D})}

□

Corollary 2 In the situation of lemma 3 we have

j = w (x u (a) x^{- 1} - u (a)),

where

x \in D

is any element with

w (x) = i

, if

a \in Z (\bar{D})

α (a) = a

and

(i, p) = 1

, where

p = c h a r D

i = 1

, we will denote

j

j (u, a)

or by

i (u, a)

Proof. Since for some parameter

z

we have

x = b (1 + x_{1} z + \dots) z^{i}

, where

b, x_{k} \in u (\bar{D})

, the proof is easily follows from the proof of (ii) in lemma 3 .

□

In the sequel we will need the following definition.

Definition 4 Let

(α, β)

be endomorphisms of a division algebra

D

. A map

δ :

D \to D^{'}

, where

D \subset D^{'}

are algebras, is called a

(α, β)

-derivation if it is linear and satisfy the following identity

δ (a b) = δ (a) α (b) + β (a) δ (b)

where

a, b \in D

We will say that

(α, 1)

-derivation is an

α

-derivation.

Lemma 4 (cf. [14] , lemma 4) Let

δ

be an

(α, β)

-derivation of an arbitrary division algebra

D

such that

α, β

preserve

Z (D)

and

α |_{Z (D)} \neq β |_{Z (D)}

Then

δ

is an inner derivation, i.e. there exists

d \in D

such that

δ (a) = d α (a) - β (a) d

for all

a \in D

Proof. Put

d = δ (a) (a^{α} - a^{β})^{- 1}

, where

a \in Z (D)

is any element such that

α (a) \neq β (a)

. Put

δ_{i n} (x) = d α (x) - β (x) d

. We claim that

δ = δ_{i n}

. Indeed, consider the map

\bar{δ} = δ - δ_{i n}

. It is an

(α, β)

-derivation. Take arbitrary

b \in D

. Then

\bar{δ} (a b) = \bar{δ} (b a)

. But we have

\bar{δ} (a b) = \bar{δ} (a) α (b) + β (a) \bar{δ} (b) = β (a) \bar{δ} (b),

and

\bar{δ} (b a) = \bar{δ} (b) α (a) + β (b) \bar{δ} (a) = α (a) \bar{δ} (b)

Therefore,

\bar{δ} (b) = 0

for any

b

□

Proposition 3 (cf. [14] , lemma 10) Let

D

be a splittable division algebra. Let

n = G a l (Z (\bar{D}) / \bar{Z (D)})

. There exists a parameter

z^{'}

such that

_{m}^{(z^{'}, u)} δ_{j} = 0

n |̸ j

Proof. Since for

n = 1

there is nothing to prove, we will assume that

n > 1

. Let

z

be some fixed parameter. By [6] , prop. 1.7

^{(z)} α |_{Z (\bar{D})}

has order

n

By proposition 2 ,

^{(z, u)} δ_{1}

is a

(^{(z)} α^{2},^{(z)} α)

-derivation. Since

n > 1

^{(z)} α^{2} |_{Z (\bar{D})} \neq^{(z)} α |_{Z (\bar{D})}

. Therefore, by lemma 4 ,

^{(z, u)} δ_{1}

is an inner derivation and

^{(z, u)} δ_{1} (a) = d^{(z)} α^{2} (a) -^{(z)} α (a) d

a \in \bar{D}

. Put

z_{1} = z - u (d) z^{2}

. By lemma 3 , (i) we have for any

a \in \bar{D}

^{(z_{1}, u)} δ_{1} (a) = 0

and

^{(z)} α (a) =^{(z_{1})} α (a)

. So,

^{(z_{1}, u)} δ_{1} = 0

and

^{(z)} α =^{(z_{1})} α

By proposition 2 ,

^{(z_{1}, u)} δ_{2}

is a

(^{(z_{1})} α^{3},^{(z_{1})} α)

-derivation. If

n \neq 2

then it is inner and we can apply lemma 3 . By induction we get that there exists a parameter

z_{n - 1}

such that

^{(z_{n - 1}, u)} δ_{j} = 0

for

j < n

and

^{(z)} α =^{(z_{n - 1})} α

. It is easy to see that then

_{m}^{(z_{n - 1}, u)} δ_{j} = 0

for

j < n

and all

m \in Z

. Note that

^{(z_{n - 1}, u)} δ_{n}

is a

(^{(z_{n - 1})} α^{n + 1},^{(z_{n - 1})} α) = (^{(z_{n - 1})} α,^{(z_{n - 1})} α)

-derivation, i.e.

^{(z_{n - 1}, u)} {δ_{n}}^{(z_{n - 1})} α^{- 1}

is a derivation.

Note that

^{(z_{n - 1}, u)} δ_{n + 1}

is a

(^{(z_{n - 1})} α^{2},^{(z_{n - 1})} α)

-derivation. This follows by proposition 2 , since

_{m}^{(z_{n - 1}, u)} δ_{j} = 0

for

j < n

and all

m \in Z

. So, by lemma 4 ,

^{(z_{n - 1}, u)} δ_{n + 1}

is an inner derivation. Using lemma 3 , (i) with

z_{n + 1} = z_{n - 1} + b z_{n - 1}^{n + 2}

for an appropriate

b

, we have

^{(z_{n + 1}, u)} δ_{j} = 0

for

j < n + 2

n |̸ j

and

^{(z)} α =^{(z_{n + 1})} α

. Moreover,

_{m}^{(z_{n + 1}, u)} δ_{j} = 0

for

j < n + 2

n |̸ j

and all

m \in Z

. This easily follows from lemma 2 .

By induction we can assume that there exists a parameter

z_{k}

such that

_{m}^{(z_{k}, u)} δ_{j} = 0

for

j < k + 1

n |̸ j

and all

m \in Z

, and

^{(z)} α =^{(z_{k})} α

So, by proposition 2 , if

n |̸ k + 1

, then

^{(z_{k}, u)} δ_{k + 1}

is an inner

(^{(z_{k})} α^{k + 2},^{(z_{k})} α)

-derivation.

And if

n | k + 1

, we can apply the same arguments and conclude that

^{(z_{k}, u)} δ_{k + 2}

is a

(^{(z_{k})} α^{k + 2},^{(z_{k})} α)

-derivation. Therefore, by lemma 3 there exists a parameter

z_{k + 1} = z_{k} + b z_{k}^{k + 2}

(

z_{k} + b z_{k}^{k + 3}

n | k + 1

) such that

_{m}^{(z_{k + 1}, u)} δ_{j} = 0

for

j < k + 2

n |̸ j

and all

m \in Z

, and

^{(z)} α =^{(z_{k + 1})} α

(or

_{m}^{(z_{k + 1}, u)} δ_{j} = 0

for

j < k + 3

n |̸ j

and all

m \in Z

, and

^{(z)} α =^{(z_{k + 1})} α

n | k + 1

Since

z_{l + 1} = (1 + b_{l} z_{l}^{k_{l}}) z_{l}

for every

l

, the sequence

{z_{l}}_{l = 1}^{\infty}

converges in

D

, which completes the proof of the proposition.

□

Lemma 5 Let

D

be a splittable division algebra as in proposition 1 , of characteristic

p > 0

. Let

t \in Z (\bar{D})

be an element such that

α (t) = t

Let

j = i (u, t)

be the minimal positive integer such that

^{(z, u)} δ_{j} |_{F_{p} (t)} \neq 0

(see corollary 2 ), and we assume

j < \infty

. Then the maps

_{n}^{(z, u)} δ_{m}

k j \leq m < (k + 1) j

k \in {1, \dots, p - 1}

satisfy the following properties:

i) there exist elements

c_{n, m, k} \in \bar{D}

such that

_{n}^{(z, u)} δ_{m} |_{F_{p} (t)} = c_{n, m, 1} δ + \dots + c_{n, m, k} δ^{k},

where

δ : F_{p} (t) \to F_{p} (t)

is a derivation such that

δ (t) = 1

, and

c_{n, k j, k} = (k! {)^{- 1}}_{n}^{(z, u)} δ_{j} (t)_{n + j}^{(z, u)} δ_{j} (t) \dots_{n + (k - 1) j}^{(z, u)} δ_{j} (t) .

ii) Let

ζ = o r d (^{(z)} α |_{Z (\bar{D})})

. Then

ζ | j

and

c_{n, k j, k} \neq 0

(n, j) = 1

and

^{(z)} α (^{(z, u)} δ_{j} (t)) \neq^{(z, u)} δ_{j} (t)

;

c_{n, k j, k} \neq 0

^{(z)} α (^{(z, u)} δ_{j} (t)) =^{(z, u)} δ_{j} (t)

and

n, (n + j), \dots, (n + (k - 1) j) \neq 0 mod p

^{(z)} α = i d

, then

c_{n, k j, k} \neq 0

iff

n, (n + j), \dots, (n + (k - 1) j) \neq 0 mod p

Proof. i) The proof is by induction on

k

. Let

a, b \in F_{p} (t)

. For

k = 1

, by proposition 2 , (ii) we have

_{n} δ_{m} (a b) =_{n} δ_{m} (a) b + a_{n} δ_{m} (b)

because all the maps

δ_{q}

q < j

are equal to zero on

F_{p} (t)

. Hence,

_{n} δ_{m}

is a derivation on

F_{p} (t)

_{n} δ_{m} |_{F_{p} (t)} = c_{n, m, 1} δ

and

c_{n, j, 1} =_{n} δ_{j} (t)

For arbitrary

k

, by proposition 2 , (i) and by the induction hypothesis we have

_{n} δ_{m} (t^{q}) = q_{n} δ_{m} (t) t^{q - 1} +_{n} δ_{j} (t) (\sum_{l = 0}^{q - 2} (c_{n + j, m - j, 1} δ + \dots + c_{n + j, m - j, k - 1} δ^{k - 1}) (t^{q - 1 - l}) t^{l}) +

\begin{matrix} \dots +_{n} δ_{m - j} (t) (\sum_{l = 0}^{q - 2} (c_{m - j + n, m - s, 1} δ) (t^{q - 1 - l}) t^{l}) . \end{matrix}

(2)

Therefore,

_{n} δ_{m} (t^{p}) = 0

, because

k \leq p - 1

and

\sum_{l = 0}^{p - 2} δ^{i} (t^{p - 1 - l}) t^{l} = 0

for

i \leq p - 2

Hence,

_{n} δ_{m} |_{F_{p} (t)} = c_{n, m, 1} δ + \dots + c_{n, m, p - 1} δ^{p - 1}

and we only have to show that

c_{n, m, q} = 0

for

q > k

Using ( 2 ) we can calculate

c_{n, m, j}

. We have

c_{n, m, 1} =_{n} δ_{m} (t);

c_{n, m, 2} = \frac{1}{2!} (_{n} δ_{m} (t^{2}) - 2 c_{n, m, 1} t) = \frac{1}{2} (_{n} δ_{j} (t) (c_{n + j, m - j, 1} δ (t)) + \dots +_{n} δ_{s} (t) (c_{s + n, m - s, 1} δ (t)))

\dots

c_{n, m, q} = \frac{1}{q!} (_{n} δ_{j} (t) (\sum_{l = 0}^{q - 2} c_{n + j, m - j, q - 1} δ^{q - 1} (t^{q - 1 - l}) t^{l}) + \dots

+_{n} δ_{m - (q - 1) j} (t) (\sum_{l = 0}^{q - 2} c_{m + n - (q - 1) j, (q - 1) j, q - 1} δ^{q - 1} (t^{q - 1 - l}) t^{l}))

\begin{matrix} = \frac{1}{q} (_{n} δ_{j} (t) c_{n + j, m - j, q - 1} + \dots +_{n} δ_{m - (q - 1) j} (t) c_{m + n - (q - 1) j, (q - 1) j, q - 1}) \end{matrix}

(3)

Hence,

c_{n, m, k + 1} = \dots = c_{n, m, p - 1} = 0

and

c_{n, k j, k} = q_{n}^{- 1} δ_{j} (t) c_{n + j, k j - j, k - 1} = (k! {)^{- 1}}_{n}^{(z, u)} δ_{j} (t)_{n + j}^{(z, u)} δ_{j} (t) \dots_{n + (k - 1) j}^{(z, u)} δ_{j} (t) .

ii) Let us prove first that

ζ

divide

i

. For, if

i

is not divisible by

ζ

, we have, by proposition 2 ,

^{(z, u)} δ_{j} (t x) =^{(z, u)} δ_{j} (t)^{(z)} α^{j + 1} (x) +^{(z)} α (t)^{(z, u)} δ_{j} (x) =^{(z, u)} δ_{j} (x t) =

^{(z, u)} δ_{j} (x)^{(z)} α^{j + 1} (t) +^{(z)} α (x)^{(z, u)} δ_{j} (t),

where

x \in Z (\bar{D})

α (x) \neq x

. But then

^{(z)} α^{j + 1} (x) =^{(z)} α (x)

, a contradiction.

^{(z)} α = i d

, the same arguments show that

^{(z, u)} δ_{j} (t) \in Z (\bar{D})

x \in \bar{D}

is an arbitrary element, this formulae shows

^{(z)} α^{j}

is an inner automorphism

a d (^{(z, u)} δ_{j} (t)^{- 1})

. Therefore,

^{(z)} α^{j} (^{(z, u)} δ_{j} (t)) =^{(z, u)} δ_{j} (t)

Assume

^{(z)} α (^{(z, u)} δ_{j} (t)) \neq^{(z, u)} δ_{j} (t)

. It's clear then that

_{n + q j}^{(z, u)} δ_{j} (t) = {\sum_{l = 0}^{n + q j - 1}}^{(z)} α^{l} (^{(z, u)} δ_{j} (t)) \neq 0

(n, j) = 1

. So,

c_{n, k j, k} \neq 0

by (i) in this case.

^{(z)} α (^{(z, u)} δ_{j} (t)) =^{(z, u)} δ_{j} (t)

, then

_{n + q j}^{(z, u)} δ_{j} (t) = (n + q j)^{(z, u)} δ_{j} (t) \neq 0

iff

p

does not divide

(n + q j)

. So, by (i)

c_{n, k j, k} \neq 0

in this case iff

n, (n + j), \dots, (n + (k - 1) j) \neq 0 mod p

The lemma is proved.

□

Lemma 6 Let

D

be a splittable division algebra as in lemma 5 . Let

s \in Z (\bar{D})

be an element such that

α (s) = s

. Let

i = i (u, s)

be the minimal positive integer such that

^{(z, u)} δ_{i} (s) \neq 0

(see corollary 2 ).

p | i

, then for any positive integral

k

there exists a map

^{(z, u)} δ_{j (k)}

such that

^{(z, u)} δ_{j (k)} (s^{p^{k}}) \neq 0

Proof. We claim that

^{(z, u)} δ_{p^{q} i}

is the first map such that

^{(z, u)} δ_{p^{q} i} |_{F_{p} (s^{p^{q}})} \neq 0

. The proof is by induction on

q

. For

q = 0

, there is nothing to prove. For arbitrary

q

, put

t = s^{p^{q - 1}}

By proposition 2 we have

δ_{p^{q} i} (t^{p}) = δ_{p^{q - 1} i} (t) {\sum_{r = 0}^{p - 2}}_{1 + p^{q - 1} i} δ_{p^{q - 1} i (p - 1)} (t^{p - 1 - r}) t^{r} + \sum_{l = p^{q - 1} i + 1}^{p^{q} i - 1} δ_{l} (t) {\sum_{r = 0}^{p - 2}}_{1 + l} δ_{p^{q} i - l} (t^{p - 1 - r}) t^{r}

By induction and lemma 5 ,

_{1 + l} δ_{p^{q} i - l} |_{F_{p} (t)} = c_{1 + l, p^{q} i - l, 1} δ + \dots + c_{1 + l, p^{q} i - 1, p - 2} δ^{p - 2}

for

l > p^{q - 1} i

. Therefore,

{\sum_{r = 0}^{p - 2}}_{1 + l} δ_{p^{q} i - l} (t^{p - 1 - r}) t^{r} = 0

. By lemma 5 , (ii),

_{1 + p^{q - 1} i} δ_{p^{q - 1} i (p - 1)} |_{F_{p} (t)} = c_{1 + p^{q - 1} i, p^{q - 1} i (p - 1), 1} δ + \dots + c_{1 + p^{q - 1} i, p^{q - 1} i (p - 1), p - 1} δ^{p - 1}

with

c_{1 + p^{q - 1} i, p^{q - 1} i (p - 1), p - 1} \neq 0

. Hence,

δ_{p^{q} i} (t^{p}) = - c_{1 + p^{q - 1} i, p^{q - 1} i (p - 1), p - 1} δ_{p^{q - 1} i} (t) \neq 0

The same arguments show that

^{(z, u)} δ_{j} (t^{p}) = 0

for

j < p^{q} i

. So,

^{(z, u)} δ_{p^{q} i}

is the first non-zero map on

F_{p} (s^{p^{q}})

□

Lemma 7 Let

D

be a splittable division algebra. Let

z

be a fixed parameter and

^{(z)} α = i d

, let

u

be some fixed embedding

u : \bar{D} ↪ D

Let

^{(z, u)} δ_{i}

i \in N \cup \infty

be the first non-zero map on

\bar{D}

. Assume

(i, p) = 1

, where

p = c h a r D

. Let

^{(z, u)} δ_{j}

j > i

j \in N \cup \infty

be the first map such that

^{(z, u)} δ_{j} \neq 0

j

is not divisible by

i

and

^{(z, u)} δ_{j} \neq {c_{j / i}}^{(z, u)} δ_{i}^{j / i}

for some

c_{j / i} \in \bar{D}

otherwise. Then a) for

k < p = c h a r D

(arbitrary

k

c h a r D = 0

) we have

^{(z, u)} δ_{k i} = {c_{k}}^{(z, u)} δ_{i}^{k}

, where

\begin{matrix} c_{k} = \frac{(i + 1) \dots (i (k - 1) + 1)}{k!}, \end{matrix}

(4)

k i < j

b) if condition ( 4 ) is satisfied for any

k

with

k i < j

, then

_{- i}^{(z, u)} δ_{q} = 0

for

i < q < j

and

_{- i}^{(z, u)} δ_{j}

is a derivation.

Remark. We will call the number

i (u, z) = {min}_{a \in \bar{D}} {w (z u (a) z^{- 1} - u (a))}

defined in this lemma a local height. The number

i = i (z, u)

in lemma coinside with the level of

D

defined in [11] if

D

has index

p = c h a r D

and

D

is splittable. As it follows from lemmas 3 , 10 (see below),

i (z, u)

does not depend on

z, u

in this case. Corollary 2 completes then the proof that it coinside with the level defined by Saltman in the case

D

is splittable. This number will play an important role in this work. It was one of the important parameters in [14] . Recall the definition of level:

h (D) = min {w (a b - b a) - w (a) - w (b)}

Proof. If we compare coefficients in formulae for

δ_{k i} (a b)

from proposition 2 with coefficients in formulae for

δ_{i}^{k} (a b)

multiplied by

c_{k}

, we must have

c_{k} k = ((k - 1) i + 1) c_{k - 1},

where from follows a).

¿From the other hand side, if

_{- i} δ_{q}

q > i

is the first nonzero map after

_{- i} δ_{i}

, it must be a derivation by proposition 2 , (i). Note that in characterictic zero case this can happens only if

q \geq j

, because a map

c δ_{i}^{k}

can not be a derivation if

k > 1

, which proves b) in this case.

Since the maps

δ_{q}

are uniquely defined, by lemma 2 , by the maps

{\tilde{δ}}_{l}

l \leq q

, and the maps

{\tilde{δ}}_{q}

are uniquely defined by the maps

_{- i} δ_{l}

l \leq q

, and

_{- i} δ_{q}

are linear combinations of

δ_{l}

l \leq q

with integer coefficients, we see that b) holds in arbitrary characteristic.

□

Remark. So we see that the maps

_{i} δ_{q}

in this lemma satisfy the same identities as

δ_{q / i}

. This can be thought of as a possible reduction from level

i

to level

1

Definition 5 Let

D

be a splittable division algebra. Let

u

be some fixed embedding

u : \bar{D} ↪ D

. Let

s \in Z (\bar{D})

be an element such that

α (s) = s

. Let

i = i (u, s)

be the minimal positive integer such that

^{(z, u)} δ_{i} (s) \neq 0

( corollary 2 shows that

i

does not depend on

z

Assume

(i, p) = 1

, where

p = c h a r D

. Define

d (u, s) = {max}_{z} {w (z^{- i} u (s) z^{i} - u (s) - u (_{- i}^{(z, u)} δ_{i} (s)) z^{i})} \in N \cup \infty,

As we can see from lemma 7 b),

d (u, s)

can be interpreted under some conditions as the number

j

there. So, this definition was motivated by this lemma.

Lemma 8 In the definition above for

p = c h a r D > 0

and

^{(z)} α |_{Z (\bar{D})} = i d

we have i)

d (u, s) = 2 i mod p

d (u, s) < \infty

; ii) If

_{- i}^{(z, u)} δ_{i} (s) \neq 0

, the map

_{- i}^{(z, u)} δ_{d (u, s) + (p - 1) i}

is the first map such that

_{- i}^{(z, u)} δ_{d (u, s) + (p - 1) i} (s^{p}) \neq 0

for any parameter

z

. In particular, if

d (u, s) = \infty

[u (s^{p}), z^{i}] = 0

Proof. (ii) Let

_{- i}^{(z, u)} δ_{κ}

be the first map such that

_{- i}^{(z, u)} δ_{κ} (s^{p}) \neq 0

. By corollary 2

κ

does not depend on

z

. By the same reason,

_{- i}^{(z, u)} δ_{i}

is the first map such that

_{- i}^{(z, u)} δ_{i} (s) \neq 0

for any

z

Put

w : = d (u, s) + (p - 1) i

and fix

u, z

. By proposition 2 we have

_{- i} δ_{w} (s^{p}) =_{- i} δ_{d (u, s)} (s) {\sum_{q = 0}^{p - 2}}_{d (u, s) - i} δ_{(p - 1) i} (s^{p - 1 - q}) s^{q} +

{\sum_{k = d (u, s) + 1}^{w - 1}}_{- i} δ_{k} (s) {\sum_{q = 0}^{p - 2}}_{k - i} δ_{w - k} (s^{p - 1 - q}) s^{q}

By lemma 5 ,

_{k - i} δ_{w - k} |_{F_{p} (s)} = c_{k - i, w - k, 1} δ + \dots + c_{k - i, w - k, p - 2} δ^{p - 2}

for

w - k < (p - 1) i

and

_{d (u, s) - i} δ_{(p - 1) i} |_{F_{p} (s)} = c_{d (u, s) - i, (p - 1) i, 1} δ + \dots + c_{d (u, s) - i, (p - 1) i, p - 1} δ^{p - 1}

with

c_{d (u, s) - i, (p - 1) i, p - 1} \neq 0

d (u, s) - i = i mod p

. Indeed, as we have shown in the proof of lemma 5 , (ii), the order

n

of the automorphism

^{(z)} α

^{(z, u)} δ_{i} (s)

must divide

i

, so

(n, p) = 1

. Now we have two possibilities:

n |̸ d (u, s)

and

n | d (u, s)

In the first case we can repeat the arguments to the first assertion in lemma 5 , (ii) to show that

c_{d (u, s) - i, (p - 1) i, p - 1} \neq 0

. In the second case we have

_{d (u, s) - i + q i} δ_{i} (s) = (d (u, s) - i + q i) / i_{i} δ_{i} (s) \neq 0

d (u, s) - i + q i

is not divided by

p

. So, by lemma 5 , (i)

c_{d (u, s) - i, (p - 1) i, p - 1} \neq 0

iff

d (u, s) - i = i mod p

in this case.

Hence,

_{- i} δ_{w} (s^{p}) = -_{- i} δ_{d (u, s)} (s) c_{d (u, s) - i, (p - 1) i, p - 1} \neq 0

d (u, s) - i = i mod p

This also shows that

_{- i} δ_{w}

is the first map such that

_{- i} δ_{w} |_{F_{p} (s^{p})} \neq 0

d (u, s) - i = i mod p

i) By Skolem-Noether theorem there exists a parameter

z^{'}

D

such that

^{(z^{'})} α = i d

Put

d^{'} (u, z^{'}, s) = w ({z^{'}}^{- j} u (s) {z^{'}}^{j} - u (s) - u (_{- i}^{(z^{'}, u)} δ_{i} (s)) {z^{'}}^{i}) .

Since

^{(z^{'})} α = i d

, the map

^{(z^{'}, u)} δ_{i}

is the first map such that

_{- i}^{(z^{'}, u)} δ_{i} (s) \neq 0

. If

d^{'} (u, z^{'}, s) \neq 2 i mod p

, we can find a parameter

z^{''}

such that

d^{'} (u, z^{''}, s) > d^{'} (u, z^{'}, s)

using lemma 3 , (ii). Continuing this procedure, we find a parameter

z

such that

d^{'} (u, z, s) = 2 i mod p

d^{'} (u, z, s) = \infty

Using arguments from ii) we get that the map

_{- i}^{(z, u)} δ_{d^{'} (u, z, s) + (p - 1) i}

is the first map such that

_{- i}^{(z, u)} δ_{d^{'} (u, z, s) + (p - 1) i} (s^{p}) \neq 0

for the parameter

z

. As it was noted in the beginning of the proof, the number

κ = d^{'} (u, z, s) + (p - 1) i

does not depend on the parameter.

Since

d^{'} (u, z, s) \leq d (u, s)

, we get

d^{'} (u, z, s) = d (u, s)

. For, otherwise we can repeat the arguments from (ii) and conclude that

_{- i}^{(z, u)} δ_{d (u, s) + (p - 1) i} (s^{p}) = 0

, a contradiction. The lemma is proved.

□

It would be interesting to know more about a behaviour of

_{m}^{(z, u)} δ_{j}

with respect to the embedding

u

. We will give an answer in one special case, namely, when

\bar{D} = Z (\bar{D})

and

Z (\bar{D}) / \bar{Z (D)}

is a simple extension.

Lemma 9 Let

D

be a division algebra such that

c h a r D = p > 0

\bar{D} = Z (\bar{D})

Z (\bar{D})

is not perfect and

Z (\bar{D}) / \bar{Z (D)}

is a simple extension (so,

D

is splittable). Let

\bar{u}

be a primitive element of the extension

Z (\bar{D}) / \bar{Z (D)}

such that

\bar{u} / \in (Z (\bar{D}))^{p}

and let

u

be any lift of

\bar{u}

D

Then there exists an embedding

u : \bar{D} ↪ D

such that

u (\bar{u}) = u

and any map

_{m}^{(z, u)} δ_{j}

is uniqely defined by the values

_{m}^{(z, u)} δ_{j} (u^{q})

or, equivalently, by the values

_{l}^{(z, u)} δ_{k} (u)

k \leq j

In particular, if

_{m}^{(z, u)} δ_{k} (u) = 0

for

k \leq j

, then

_{m}^{(z, u)} δ_{j} = 0

Proof. Consider a field

Z (D) (u)

. It is a complete discrete valued field as a finite extension of

Z (D)

. By classical Cohen theorem, there exists an embedding

\bar{Z (D) (u)} = \bar{D} ↪ Z (D) (u) \subset D

. By [4] , lemmas 11,12 the embedding is completely defined by a

p

-basis

Γ

of the field

\bar{Z (D) (u)}

. Namely, for any lift

G

of a given

p

-basis

Γ

there exists an embedding

s

such that

G \subset s (\bar{Z (D) (u)})

Let's show that there exists a

p

-basis

Γ

of the field

\bar{D}

such that

\bar{u} \in Γ

and

Γ ∋ γ \in \bar{Z (D)}

γ \neq \bar{u}

Consider a set of all non-void sets

Γ^{'}

of elements

γ_{τ} \in \bar{D}

satisfying the following property:

\bar{u} \in Γ^{'}

Γ^{'} ∋ γ \in \bar{Z (D)}

γ \neq \bar{u}

and

[{\bar{D}}^{p} (γ_{1}, \dots, γ_{r}) : {\bar{D}}^{p}] = p^{r}

for any

r

distinct elements of

Γ^{'}

This set is not void, since it contains the set

Γ^{'} = {\bar{u}}

. By Zorn's lemma, there exists a maximal set

Γ

satisfying A). Then

\bar{D} = {\bar{D}}^{p} (Γ)

. Indeed, since

{\bar{Z (D)}}^{p} (\bar{u}) \subset {\bar{D}}^{p} (Γ)

, it suffice to show that any element from

\bar{Z (D)}

lies in

{\bar{D}}^{p} (Γ)

. Suppose

a \in \bar{Z (D)}

a / \in {\bar{D}}^{p} (Γ)

. Then the set

Γ^{'} = {a \cup Γ}

satisfy A), a contradiction with maximality of

Γ

Now, we can take a lift of

Γ

in the following way. We take

u

as a lift of

\bar{u}

, and we take lifts of all other elements in

Z (D)

. This lift defines an embedding

u : \bar{D} ↪ D

Let us show that any map

_{m}^{(z, u)} δ_{j}

(for some fixed

z

) is uniqely defined by the values

_{l}^{(z, u)} δ_{k} (u)

k \leq j

. We have

u (\bar{D}) = u (\bar{Z (D)}) (u)

and any element

a \in u (\bar{D})

can be represented as a polynomial in finite number of elements from

Γ

with coefficients from

u (\bar{D})^{p^{k}}

for any

k > 0

Note that for any

j

there exists

k > 0

such that for any

b \in {\bar{Z (D)}}^{p^{k}}

_{l}^{(z, u)} δ_{q} (b) = 0

for all

q \leq j

and all

l

. Indeed, assume

_{1}^{(z, u)} δ_{q} (b) \neq 0

for some

q \leq j

b \in {\bar{Z (D)}}^{p^{k}}

and

_{l}^{(z, u)} δ_{s} (c) = 0

for all

l

, all

c \in {\bar{Z (D)}}^{p^{k}}

and all

s < q

. Then, since

^{(z)} α |_{\bar{Z (D)}} = i d

and by proposition 2 ,

_{l}^{(z, u)} δ_{s} (b^{p}) = 0

for all

b \in {\bar{Z (D)}}^{p^{k}}

, all

l

and all

s \leq q

Now, since

u (\bar{D})^{p^{k}} = u (\bar{Z (D)})^{p^{k}} (u^{p^{k}})

, any element

a \in u (\bar{D})

can be represented as a polynomial in finite number of elements from

Γ

with coefficients from

u (\bar{Z (D)})^{p^{k}}

. Since all elements except

u

Γ

belong to the center

Z (D)

, the value of

_{m}^{(z, u)} δ_{j} (a)

is uniqely determined by the values

_{m}^{(z, u)} δ_{j} (u^{l})

that are uniqely defined, by proposition 2 , by the values

_{l}^{(z, u)} δ_{k} (u)

k \leq j

□

Remark In the case

Z (\bar{D})

perfect field there is only one embedding

u

, which is compatible with the embedding

\bar{Z (D)} ↪ Z (D)

. So, the assertion of lemma is easy in this case.

Lemma 10 (cf. [14] , lemma 8) In the situation of lemma 9 suppose

_{m}^{(z, u)} δ_{1} = \dots =_{m}^{(z, u)} δ_{j - 1} = 0

_{m}^{(z, u)} δ_{j} \neq 0

. Let

n

be the order of

^{(z)} α

. Then (i) for

u^{'} = u + b z^{q}

b \in u (\bar{D})

n | q

we have

_{m}^{(z, u^{'})} δ_{l} =_{m}^{(z, u)} δ_{l}

l < q

and

_{m}^{(z, u^{'})} δ_{q} (\bar{u}) =_{m}^{(z, u)} δ_{q} (\bar{u}) +^{(z)} α^{m} (\bar{b}) - \frac{\partial}{\partial \bar{u}} (^{(z)} α^{m} (\bar{u})) \bar{b},

where the derivative is taken in the field

\bar{D} = {\bar{D}}^{p} (Γ)

(ii) Suppose

^{(z)} α = i d

. Then for

u^{'} = u + b z^{q}

b \in u (\bar{D})

we have

_{m}^{(z, u^{'})} δ_{l} =_{m}^{(z, u)} δ_{l}

l < q + j

and

_{m}^{(z, u^{'})} δ_{q + j} (\bar{u}) =_{m}^{(z, u)} δ_{q + j} (\bar{u}) +_{m}^{(z, u)} δ_{j} (\bar{b}) - \frac{\partial}{\partial \bar{u}} (_{m}^{(z, u)} δ_{j} (\bar{u})) \bar{b},

where the derivative is taken in the field

\bar{D} = {\bar{D}}^{p} (Γ)

(iii) Suppose

^{(z)} α = i d

. Let

{\bar{u}}^{'} \in \bar{D}

be any primitive element of the extension

\bar{D} / \bar{Z (D)}

satisfying the conditions of lemma 9 , and let

u^{'} \in D

be any lift of

{\bar{u}}^{'}

. Then we have

_{m}^{(z, u^{'})} δ_{l} =_{m}^{(z, u)} δ_{l}

l < j

and

_{m}^{(z, u^{'})} δ_{j} ({\bar{u}}^{'}) =_{m}^{(z, u)} δ_{j} (\bar{u}) \frac{\partial}{\partial \bar{u}} ({\bar{u}}^{'}),

where the derivative is taken in the field

\bar{D} = {\bar{D}}^{p} (Γ)

Proof. First of all, let's note that there exists

k \in N

such that for any

a \in {\bar{Z (D)}}^{p^{k}}

holds

u (a) - u^{'} (a) = 0 mod M_{D}^{q + 1}

, where

u^{'}

is any another embedding,

q \in N

is any given number.

Indeed, assume for any

c \in {\bar{Z (D)}}^{p^{s}}

holds

u (c) - u^{'} (c) = 0 mod M_{D}^{l}

, i.e.

u (c) = u^{'} (c) + c_{l} z^{l} + \dots

, where

c_{l} \in u^{'} (\bar{D})

. Then

u (c^{p}) = (u (c))^{p} = (u^{'} (c))^{p} + p u^{'} (c)^{p - 1} c_{l} z^{l} + \dots

, so

u (c^{p}) - u^{'} (c^{p}) = 0 mod M_{D}^{l + 1}

. ¿From this immediately follows that

u (a) - u^{'} (a) = 0 mod M_{D}^{q}

for any

a \in \bar{D}

u^{'}

is defined by the element

u^{'} = u + b z^{q}

, because

u (\bar{u}) - u^{'} (\bar{u}) = b z^{q}

. Moreover, if we represent

a

as some polynomial

P (γ_{1}, \dots, γ_{r}, \bar{u})

with coefficients from

{\bar{Z (D)}}^{p^{k}}

, then it is clear that

[u (a) - u^{'} (a)] z^{- q} = - \frac{\partial}{\partial \bar{u}} (P (γ_{1}, \dots, γ_{r}, \bar{u})) \bar{b} mod M_{D}

n | q

, since

u (γ_{l}) = u^{'} (γ_{l})

for any

l

and

z^{q} u z^{- q} = u mod M_{D}

. It is also clear that the derivative can be taken even in the field

{\bar{D}}^{p} (Γ)

. So, we have (i)

z^{m} u^{'} z^{- m} = z^{m} (u + b z^{q}) z^{- m} = u (^{(z)} α^{m} (\bar{u})) + u (_{m}^{(z, u)} δ_{j} (\bar{u})) z^{j} + \dots + (u (^{(z)} α^{m} (\bar{b}))

+ u (_{m}^{(z, u)} δ_{j} (\bar{b})) z^{j} + \dots) z^{q} = u (^{(z)} α^{m} (\bar{u})) + \dots + (u (^{(z, u)} δ_{q} (\bar{u})) + u (^{(z)} α^{m} (\bar{b}))) z^{q} + \dots =

u^{'} (^{(z)} α^{m} (\bar{u})) + \dots + (u^{'} (_{m}^{(z, u)} δ_{q} (\bar{u})) + u^{'} (^{(z)} α^{m} (\bar{b})) - u^{'} ({\frac{\partial}{\partial \bar{u}}}^{(z)} α^{m} (\bar{u}) \bar{b})) z^{q} + \dots,

(ii) We have

z^{m} u^{'} z^{- m} = z^{m} (u + b z^{q}) z^{- m} = u (\bar{u}) + u (_{m}^{(z, u)} δ_{j} (\bar{u})) z^{j} + \dots + (u (\bar{b}) + u (_{m}^{(z, u)} δ_{j} (\bar{b})) z^{j} + \dots) z^{q} =

u (\bar{u}) + u (_{m}^{(z, u)} δ_{j} (\bar{u})) z^{j} + \dots + (u (_{m}^{(z, u)} δ_{q} (\bar{u})) + u (\bar{b})) z^{q} + u (_{m}^{(z, u)} δ_{q + 1} (\bar{u})) z^{q + 1} + \dots

+ u (_{m}^{(z, u)} δ_{q + j - 1} (\bar{u})) z^{q + j - 1} + (u (_{m}^{(z, u)} δ_{q + j} (\bar{u})) + u (_{m}^{(z, u)} δ_{j} (\bar{b}))) z^{q + j} + \dots =

u^{'} (\bar{u}) + u^{'} (_{m}^{(z, u)} δ_{j} (\bar{u})) z^{j} + \dots + u^{'} (_{m}^{(z, u)} δ_{q + j - 1} (\bar{u})) z^{q + j - 1} + (u^{'} (_{m}^{(z, u)} δ_{q + j} (\bar{u})) + u^{'} (_{m}^{(z, u)} δ_{j} (\bar{b})) -

u^{'} (\frac{\partial}{\partial \bar{u}} (_{m}^{(z, u)} δ_{j} (\bar{u})) \bar{b})) z^{q + j} + \dots

(iii) Assume

u^{'} = u ({\bar{u}}^{'}) + a_{1} z + \dots

, where

a_{i} \in u (\bar{D})

. Since, by proposition 2 , the map

_{m}^{(z, u)} δ_{j}

is a derivation, we have

z^{m} u^{'} z^{- m} = [u ({\bar{u}}^{'}) + u (_{m}^{(z, u)} δ_{j} ({\bar{u}}^{'})) z^{j} + \dots] + [a_{1} + u (_{m}^{(z, u)} δ_{j} (a_{1}) z^{j} + \dots] z + \dots =

u^{'} + u (_{m}^{(z, u)} δ_{j} ({\bar{u}}^{'})) z^{j} + \dots = u^{'} + u (_{m}^{(z, u)} δ_{j} (\bar{u}) \frac{\partial}{\partial \bar{u}} ({\bar{u}}^{'})) z^{j} + \dots = u^{'} + u^{'} (_{m}^{(z, u)} δ_{j} (\bar{u}) \frac{\partial}{\partial \bar{u}} ({\bar{u}}^{'})) z^{j} + \dots

□

4 The period-index problem

In this section we will prove the following theorem.

Theorem 2 The following conjecture: the exponent of

A

is equal to its index for any division algebra

A

over a

C_{2}

-field

F

has the positive answer for

F = F_{1} ((t))

, where

F_{1}

is a

C_{1}

-field.

Recall that a field

F

is called a

C_{i}

-field if any homogeneous form

f (x_{1}, \dots, x_{n})

of degree

d

n > d^{i}

variables with coefficients in

F

has a non-trivial zero. Some basic properties of

C_{i}

-fields see, for example, in [10] .

This conjecture was proposed by M. Artin and was solved for some another examples of the field

F

by many authors. As it is known for me, the positive answer for all division algebras of index

i n d A = 2^{a} 3^{b}

was given in [10] , for division algebras over the field

F = k ((X)) ((Y))

, where

k

is a perfect field of characteristic

p \neq 0

such that

{dim}_{F_{p}} k / ℘ (k) = 1

, was given by Tignol in the Appendix in [2] (we include this case though

F

may not be a

C_{2}

-field), for division algebras of index prime to the characterictic of

F

, where

F

is a function field of a surface, was given in [7] . I propose, the positive answer was also known for division algebras over

F = F_{1} ((t))

of characteristic 0. We will give the prove of the theorem above in any characteristic.

Proof. 1) Recall that any extension of a

C_{1}

-field is simple. Indeed, suppose

E = \bar{F} (u_{1}, \dots, u_{r})

. Consider the field

K = \bar{F} (u_{1}^{p}, \dots, u_{r}^{p})

. By Tsen's theorem,

K

and

E

are

C_{1}

-fields. So, the form

x_{1}^{p} + x_{2}^{p} u_{1} + \dots + x_{p}^{p} u_{1}^{p - 1} + x_{p + 1}^{p} u_{2}

has a non-trivial zero in

E

. But

x_{i}^{p} \in K

and elements

1, u_{1}, \dots, u_{1}^{p - 1}, u_{2}

are linearly independent over

K

, a contradiction.

2) Assume the theorem is known in the prime exponent case. We deduce the theorem by ascending induction on

e = e x p A

. If

e

is not a prime number, then write

e = l m

By assumption

A^{\otimes m}

can be split by a field extension

F \subset F^{'}

of degree

l

. This implies that

A_{F^{'}}

has exponent dividing

m

. Note that

F^{'}

is also a Laurent series field. By the induction hypothesis applied to the pair

(F^{'}, A_{F^{'}})

, there exists a field extension

F^{'} \subset L

of degree dividing

m

splitting

A_{F^{'}}

. Therefore

A

is split by the extension

F \subset L

of degree dividing

l m

and we conclude the theorem.

3) So, let

e x p A = l

be a prime number. By the basic properties of the exponent and the index (see, e.g. [10] ) we have then

i n d A = l^{k}

for some natural

k

Suppose

(l, p = c h a r F) = 1

It is known that the conjecture is true for all division algebras of index

i n d A = 2^{a} 3^{b}

, so we can assume

l \neq 2, 3

. We can assume

F

contains the group

μ_{l}

l

-roots of unity, because

[F (μ_{l}) : F] < l

and we can reduce the problem to the algebra

A \otimes_{F} F (μ_{l})

. Then by the Merkuriev-Suslin theorem

A

is similar to the tensor product of symbol-algebras of index

l

To conclude the statement of the corollary it is sufficient to prove that every two symbol algebras

A_{1}, A_{2}

contain

F

-isomorphic maximal subfields.

Since every division algebra over a

C_{1}

-field is trivial and every field extension is simple, every symbol-algebra of index

l

over

F

is splittable. Since

(l, p) = 1

, it is good splittable and its residue field is a cyclic Galois extension of

\bar{F}

. So, if

z_{i}

is a parameter from proposition 3 for algebra

A_{i}

, then

z_{i}

acts on

{\bar{A}}_{i}

as a Galois automorphism and

z_{i}^{l} \in F

. We have

v (z_{i}^{l}) = 1

(

v

is the valuation on

F

). Let us show that

A_{1}

contains a

l

-root of any element

u

F

with

v (u) \neq 0

. So,

A_{1}

will contain a subfield isomorphic to

F (z_{2})

. Since for any element

1 + b

v (b) > 0

there exists a

l

-root

(1 + b)^{1 / l} \in F

, it is sufficient to prove that

A_{1}

contains any

l

-root of elements

c t

c \in u (\bar{F})

, where

u

is some fixed embedding

u : {\bar{A}}_{1} ↪ A_{1}

Assume

z_{1}^{l} = c_{1} t

c_{1} \in u (\bar{F})

. Note that for any element

b \in u ({\bar{A}}_{1})

we have

(b z_{1})^{l} = u (N_{{\bar{A}}_{1} / \bar{F}} (b)) z_{1}^{l}

. But the norm map

N_{{\bar{A}}_{1} / \bar{F}}

is surjective, since

\bar{F}

is a

C_{1}

-field (see, e.g.

[10] , 3.4.2), so there exists

b

such that

(b z)^{l} = c t

4) Suppose now

e x p A = p

. Then

i n d A = p^{k}

By Albert's theorem (in [1] ) there exists a field

F^{'} = F (u_{1}^{1 / p}, \dots, u_{k}^{1 / p})

which splits

A

Using the same arguments as in 1) one can show that every such a field has maximum two generators, say

F^{'} = F (u_{1}^{1 / p}, u_{2}^{1 / p})

. Therefore,

i n d A \leq p^{2}

. If

i n d A = p

, there is nothing to prove, so we assume

i n d A = p^{2}

and

F^{'}

is a maximal subfield in

A

5) Suppose

F_{1}

is a perfect field.

By Albert's theorem,

A \sim = A_{1} \otimes_{F} A_{2}

, where

A_{1}, A_{2}

are cyclic algebras of degree

p

A_{1} = (L_{1} / F, σ_{1}, u_{1})

A_{2} = (L_{2} / F, σ_{2}, u_{2})

. Since

F_{1}

is perfect,

{\bar{A}}_{1} / \bar{F}

{\bar{A}}_{2} / \bar{F}

are Galois extensions. So,

A_{1}, A_{2}

are good splittable. Let us show that

A_{1}, A_{2}

have common splitting field of degree

p

over

F

. This leads to a contradiction.

By proposition 3 there exist parameters

z_{1} \in A_{1}

z_{2} \in A_{2}

such that they act on

{\bar{A}}_{1}

{\bar{A}}_{2}

as Galois automorphisms. Note that then

z_{1}^{p}, z_{2}^{p} \in F

. Let us show that

F (z_{1})

splits

A_{2}

Consider the centralizer

D = C_{A} (F (z_{1}))

. Consider the element

t_{1} = z_{2} z_{1}^{- 1}

. We have

t_{1}^{p} \in F

w (t_{1}) = 0

, where

w

denote the unique extension of the valuation

v

F

. Since

\bar{D} / \bar{Z (D)}

is a Galois extension, there exists an element

b_{1} \in F

such that

w (t_{1} - b_{1}) > 0

Since

(t_{1} - b_{1})^{p} \in F

, there exists natural

k_{1}

such that

w ((t_{1} - b_{1}) z_{1}^{- k_{1}}) = 0

. Denote

t_{2} = (t_{1} - b_{1}) z_{1}^{- k_{1}}

. We have again

t_{2}^{p} \in F

. Repeating this arguments and using the completeness of

D \subset A

we get

z_{2} = t_{1} z_{1} = (t_{2} z_{1}^{k_{1}} + b_{1}) z_{1} = \dots = b_{1} z_{1} + b_{2} z_{1}^{k_{1} + 1} + \dots

, so,

z_{2} \in F (z_{1}) = Z (D)

6) Suppose

F_{1}

is not perfect.

Since

F^{'}

is generated by two elements over

F

, it contains all

p

-roots of

F

. Then, every two elements

u, z \in F

such that

z^{1 / p} / \in F (u^{1 / p})

, where

z^{1 / p}, u^{1 / p} \in F^{'}

, also generate

F^{'}

over

F

. This follows from the same arguments as in 1), 4).

Now take

u \in F_{1} \ F_{1}^{p}

z = u + t

. It's clear that

p

-roots of these elements generate

F^{'}

over

F

. Moreover, the fields

F (u^{1 / p}), F (z^{1 / p})

are ”unramified” over

F

, i.e.

[\bar{F (u^{1 / p})} : \bar{F}] = p = [F (u^{1 / p}) : F]

[\bar{F (z^{1 / p})} : \bar{F}] = p

. Denote

u_{1} = u^{1 / p}

u_{2} = z^{1 / p}

F^{'}

. Then by Albert's theorem,

A \sim = A_{1} \otimes_{F} A_{2}

, where

A_{1}, A_{2}

are cyclic algebras of degree

p

A_{1} = (L_{1} / F, σ_{1}, u)

A_{2} = (L_{2} / F, σ_{2}, z)

Concider the centralizer

D = C_{A} (F (u_{1}))

. Suppose

\bar{D} / \bar{Z (D)}

is a separable extension.

Then there exist a lift

u : \bar{D} ↪ D

of arbitrary embedding

u^{'} : \bar{F (u_{1})} ↪ F (u_{1})

. Consider the embedding

u^{'} = u_{1}

defined in lemma 9 . Since

F (u_{1}) / F

is a purely inseparable extension,

u^{'}

is a good embedding, so

u

is a good embedding of

\bar{D} = \bar{A}

D \subset A

. So, we get

A

is a good splittable algebra, and

u (\bar{A})

contain a purely inseparable over

F

element. But this is a contradiction with lemma 6 . So,

\bar{A} / \bar{F}

can not contain a separable subextension, because in this case

\bar{D} / \bar{Z (D)}

must be a separable extension. Now we can use, for shorteness, lemmas A.4., A.6. of Tignol in Appendix to the paper [2] . These lemmas show that a tensor product

A_{1} \otimes A_{2}

of any two symbols

A_{1}, A_{2}

is similar either to a single symbol in

B r (F)

(in which case we are done) or to a product of two symbols of level zero. Recall that, by Saltman's results in [11] , every division algebra of level zero is tame, which means in our case that the residue division algebra is a separable extension over

\bar{F}

. A notion of level was already discussed above in remark to lemma 7 .

So, assume

A \sim D_{1} \otimes D_{2}

, where

D_{1}, D_{2}

are tame division algebras of degree

p

over

F

. We can assume

A

and

D_{1} \otimes D_{2}

are division algebras, so

A \sim = D_{1} \otimes D_{2}

. Since

D_{1}, D_{2}

are tame, we conclude

\bar{A}

must contain a separable element, a contradiction.

The theorem is proved.

□

5 Good splittable algebras

In this section we prove a decomposition theorem for good splittable division algebras.

This theorem shows how the studying of good splittable division algebras can be reduced to the studying of division algebras with simple described structure. So, good splittable algebras are the most easy and good algebras to study.

Lemma 11 Let

D

be a good splittable division algebra,

F = Z (D)

, and let

Z (\bar{D}) = \bar{F} (s)

be a purely inseparable over

\bar{F}

field of degree

p = c h a r D > 0

. Let

u : \bar{D} ↪ D

be a good embedding.

Then there exists a parameter

z

such that

_{- i}^{(z, u)} δ_{j} = 0

for

j > i

, where

i = i (z, u)

is a local height, and

u (^{(z, u)} δ_{i} (s)) = x

, where

x \in Z (D)

. Moreover,

(i, p) = 1

Proof. Since

Z (\bar{D}) / \bar{F}

is a purely inseparable extension,

^{(z)} α |_{Z (\bar{D})} = i d

for any parameter

z

. By Skolem-Noether theorem there exists a parameter

z

D

such that

^{(z)} α = i d

. Suppose

^{(z, u)} δ_{i} (s) = 0

, where

i = i (z, u)

. Then

^{(z, u)} δ_{i} |_{Z (\bar{D})} = 0

, since

u

is a good embedding and

Z (\bar{D}) / \bar{F}

is a simple extension. So,

^{(z, u)} δ_{i}

is an inner derivation by Scolem-Noether theorem, and by lemma 3 , (i) there exists a parameter

z^{'}

such that

^{(z^{'}, u)} δ_{i} = 0

^{(z^{'})} α = i d

So, we can assume

^{(z, u)} δ_{i} (s) \neq 0

for some parameter

z

. Since

s^{p} \in Z (D)

, by lemma 6 we have

(i, p) = 1

. Since

^{(z, u)} δ_{i}

is a derivation,

^{(z, u)} δ_{i} (s) \in Z (\bar{D})

(see the arguments in lemma 5 , (ii)). Since

(i, p) = 1

, there exists

k

such that

p | (1 - k i)

. So, by lemma 3 , (iii), for the parameter

z^{'} = (^{(z, u)} δ_{i} (s))^{k}

we have

^{(z^{'})} α = i d

^{(z^{'}, u)} δ_{i} (s) \in \bar{F}

, i.e.

u (^{(z^{'}, u)} δ_{i} (s)) \in Z (D)

. Since

s^{p} \in Z (D)

, by lemma 8 we must have

d (u, s) = \infty

. In the proof of lemma 8 , (i) was shown that

d (u, s) = d^{'} (u, z, s)

for some parameter

z

, and the construction of this element uses lemma 3 , (ii), so it preserves the initial values of

^{(z^{'})} α

^{(z^{'}, u)} δ_{i}

. So,

_{- i}^{(z, u)} δ_{j} = 0

for

j > i

and the lemma is proved.

□

Proposition 4 Let

D

be a splittable division algebra. Then we have

D \sim = D_{1} \otimes_{F} D_{2}

, where

D_{1}, D_{2}

are splittable division algebras such that

D_{1}

is an inertially split algebra.

D

is a good splittable division algebra, then

Z ({\bar{D}}_{2}) / \bar{F}

is a purely inseparable extension and

D_{2}

is a good splittable algebra (

D_{1}

D_{2}

may be trivial).

So,

D \sim A \otimes_{F} B \otimes_{F} D_{2}

, where

A

is a cyclic division algebra and

B

is an unramified division algebra.

Proof. If

c h a r D = 0

, the proposition is obvious, so we assume

c h a r D > 0

By [9] , p.261,

D \sim = D_{1} \otimes_{F} \dots \otimes_{F} D_{k}

, where

[D : F] = p_{1}^{r_{1}} \dots p_{k}^{r_{k}}

and

[D_{i} : F] = p_{i}^{r_{i}}

Let

p_{2} = p

. Since

D_{i}

are defectless over

F

D_{1}, D_{3}, \dots D_{k}

are inertially split. Therefore, by theorem 1 the algebra

B = D_{1} \otimes D_{3} \otimes \dots \otimes D_{k}

is good splittable.

Assume first that

D

is good splittable. By proposition 1.7. in [6] , if

s \in Z (\bar{D})

is an element such that

α (s) = s

, then this element is a purely inseparable element over

\bar{F}

So, if

D

is a good splittable division algebra, then by lemma 6

D_{2}

is either inertially split or

Z ({\bar{D}}_{2}) / \bar{F}

is a purely inseparable extension. For, otherwise there exists an element

s \in Z ({\bar{D}}_{2}) \subset Z (\bar{D})

as above and by proposition 3

p | i (u, s)

for any embedding

u

. If

u

is a good embedding, then

s^{p^{k}} \in Z (D)

for some

k

, a contradiction.

So, we assume below

Z ({\bar{D}}_{2}) / \bar{F}

is a purely inseparable extension. Now, we have (see, e.g. th.1 in [8] )

\bar{D} \sim = {\bar{D}}_{2} \otimes_{\bar{F}} \bar{B}

and so

u (\bar{D}) \sim = u ({\bar{D}}_{2}) \otimes_{u (\bar{F})} u (\bar{B})

, where

u

is a good embedding. So,

E = u (Z ({\bar{D}}_{2}))

is a purely inseparable field over

u (\bar{F}) \subset Z (D)

. Consider the field

E^{'} = u (K) \otimes_{u (\bar{F})} F

, where

K

is a maximal separable subfield in

\bar{B}

This is an inertial lift of

K

D

. Consider the centralizer

C = C_{D} (E^{'}) \sim = D_{2} \otimes_{F} E^{'}

. Let

M

be a maximal subfield in

{\bar{D}}_{2}

. Note that

u ({\bar{D}}_{2}) \subset C

, so

L \subset C

, where

L = u (M) F

is the composit of

u (M)

and

F

, and

E \subset L

. Note that

[L : F] = i n d D_{2} = i n d C

. The field

L

splits

C

by dimension arguments. So, it must split

D_{2}

, since

([E^{'} : F], p) = 1

, and

D_{2}

is a

p

-algebra. Therefore,

L

is isomorphic to a maximal subfield in

D_{2}

, so

D_{2}

contain a copy of purely inseparable ”unramified” subfield, whose residue field is isomorphic to

Z ({\bar{D}}_{2})

. Therefore,

D_{2}

is a god splittable algebra. For, the centralizer of this field is an unramified division algebra, so by theorem 1 is splittable. So,

D_{2}

is good splittable if the purely inseparable field is good splittable. But it is good splittable since it contains a subfield isomorphic to

u (Z ({\bar{D}}_{2}))

by the construction. (Another way to see it is to use arguments from lemma 9 to show that there exists an appropriate

p

-basis).

Let

D

be a splittable algebra. Then the same arguments as in the previous paragraph show that

L

is isomorphic to a maximal subfield in

D_{2}

(it is not important that

Z ({\bar{D}}_{2}) / \bar{F}

may be not a purely inseparable extension). Now, the composit

E F \subset L

E F \neq L

, since every element from

E

commute with

u ({\bar{D}}_{2})

, where

u

is some fixed embedding. So we must have

\bar{C_{D_{2}} (E F) = \overset{2}{\bar{D}}}

and

C_{D_{2}} (E F)

is an unramified division algebra. Therefore,

D_{2}

is splittable division algebra.

Decomposition theorems [6] , Thm. 5.6-5.15 complete the proof.

□

This proposition shows that the study of splittable division algebras can be reduced to the study of splittable

p

-algebras. So, below in this section and in the next section we will deal with

p

-algebras only.

Proposition 5 Let

D

be a good splittable division algebra such that

Z (\bar{D}) / \bar{Z (D)}

is a purely inseparable extension. Then

D \sim = D_{1} \otimes_{Z (D)} D_{2}

, where

D_{1}

is an unramified division algebra and

D_{2}

is a good splittable division algebra such that

{\bar{D}}_{2}

is a field,

{\bar{D}}_{2} / \bar{Z (D)}

is a purely inseparable extension,

[{\bar{D}}_{2} : \bar{Z (D)}] = [Γ_{D_{2}} : Γ_{Z (D)}]

Proof. The proof is by induction on the degree

[Z (\bar{D}) : \bar{Z (D)}]

Assume

[Z (\bar{D}) : \bar{Z (D)}] = p

. Let

^{(z, u)} δ_{i}

be the map from lemma 11 . Then

^{(z, u)} δ_{i}^{p}

is a derivation trivial on the centre

Z (D)

, hence by Scolem-Noether theorem it is an inner derivation.

We claim that

z^{p} \in Z (D)

. We have

z^{- i} a z^{i} = a +_{- i} δ_{i} (a) z^{i}, a \in u (\bar{D})

Therefore,

z^{- p i} a z^{p i} = a +_{- i} δ_{i}^{p} (a) z^{p i}, a \in u (\bar{D})

and

z^{p i} a z^{- p i} = a + δ_{1}^{'} (a) z^{p i} + {δ_{1}^{'}}^{2} (a) z^{2 p i} + \dots,

where

δ_{1}^{'} = (- 1)_{- i} δ_{i}^{p} = i^{p} δ_{i}^{p}

. So,

z^{p} a z^{- p} = a + \frac{1}{i} δ_{1}^{'} (a) z^{p i} + c_{2} \frac{1}{i^{2}} {δ_{1}^{'}}^{2} (a) z^{2 p i} + \dots,

where

c_{k}

are given by ( 4 ) in lemma 7 . So,

z^{p} \in Z (D)

iff

δ_{i}^{p} = 0

. Suppose

δ_{i}^{p} \neq 0

Consider an element

Y \in Z (D)

w (Y) > 0

. Let

Y = a_{1} z^{p} + \dots, a_{1} \in u (\bar{D}) .

First note that

Y = a_{1} z^{p} + a_{2} z^{2 p} + a_{3} z^{3 p} + \dots, a_{i} \in u (\bar{D})

Indeed,

Y

must satisfy

[Y, s] = 0

, where

s

is a generator of

u (Z (\bar{D}))

over

u (\bar{F})

. Since

s \in u (Z (\bar{D}))

and

w ([z^{k}, s]) = k + i

(k, p) = 1

and

w ([z^{k}, s]) = \infty

otherwise, we then have

[z^{i_{k}}, s] = 0

for every

k

, where

Y = \sum_{k = 1}^{\infty} a_{k} z^{i_{k}}

Therefore,

p | i_{k}

Then,

Y

must satisfy

Y a = a Y

for any

a \in u (\bar{D})

. Therefore,

a_{1}, \dots a_{i} \in u (Z (\bar{D}))

and we must have

a a_{i + 1} - a_{i + 1} a = a_{1} δ_{1}^{'} (a) / i

and

a a_{2 i + 1} - a_{2 i + 1} a = a_{i} δ_{1}^{'} (a) + a_{1} c_{2} {δ_{1}^{'}}^{2} (a) .

Since

Δ (a) = a a_{2 i + 1} - a_{2 i + 1} a

is an inner derivation, we get

{δ_{1}^{'}}^{2} = δ

, where

δ

is a derivation, which is a contradiction if

δ \neq 0

and

c h a r D \neq 2

. In the last case we can use the same arguments with

a_{3 i + 1}

. Therefore,

{δ_{1}^{'}}^{2} = δ = 0

and

δ_{1}^{'} = 0

, and

z^{p} \in Z (D)

Consider the algebra

W = u (Z (\bar{D})) ((z))

. Since

z^{p} \in Z (D)

and

u (\bar{F}) \subset Z (D)

, we have

Z (W) = u (\bar{F}) ((z^{p})) = F

. So,

D \sim = W \otimes_{F} C_{D} (W)

by Double Centralizer theorem.

It is clear that

C_{D} (W)

is an unramified division algebra. Now suppose the proposition is proved for

[Z (\bar{D}) : \bar{Z (D)}] = p^{k - 1}

. By Albert's theorem (th.13 in [1] )

D_{2}

then is a cyclic algebra as a product of cyclic subalgebras

D_{i}

, where

{\bar{D}}_{i} / \bar{F}

is a simple purely inseparable extension and

D_{i}

is a good splittable algebra.

Assume

[Z (\bar{D}) : \bar{Z (D)}] = p^{k}

. For a good embedding there exists a lift

\tilde{K}

of a subfield

\bar{Z (D)} \subset K \subset Z (\bar{D})

such that the extension

K / \bar{Z (D)}

has degree

p

, i.e.

\bar{\tilde{K}} = K

Γ_{\tilde{K}} = Γ_{Z (D_{2})}

u (K) \subset \tilde{K}

\tilde{K} / Z (D)

is a purely inseparable extension of degree

p

. By the induction hypothesis the centralizer

C_{D} (\tilde{K}) \sim = A_{1} \otimes_{\tilde{K}} A_{2}

, where

A_{2}

is a cyclic division algebra and

{\bar{A}}_{2}

is a field. Note that

{\bar{A}}_{2} = Z (\bar{D})

By theorem 6 in [1] we can assume

A_{2} = (L / \tilde{K}, σ, a)

, where

a

generate

\tilde{K}

over

Z (D)

. So,

A_{2}

contains a maximal purely inseparable Kummer subfield

E = \tilde{K} (y)

with

y^{p^{k - 1}} = a

, so

E = Z (D) (y)

. By theorem 3 in [1]

L = L_{0} \times \tilde{K}

, where

L_{0}

is cyclic of degree

p^{k - 1}

over

Z (D)

and

y x_{0} = σ (x_{0}) y

, where

x_{0} \in L_{0}

Consider the centralizer

B = C_{D} (L_{0})

. We claim

B \sim = B_{1} \otimes_{L_{0}} B_{2}

, where

B_{2}

is a cyclic division algebra of degree

p

and

B_{2}

contains

\tilde{K}

Note that

B

contains

Z (D) (a) = \tilde{K}

and

A_{1}

. If

\tilde{K} L_{0} = L

is ”unramified” over

L_{0}

, then we apply the arguments for the first step of our induction to the algebra

B

By construction,

B_{2}

then will contain

L

, so

\tilde{K}

. Suppose

L

is totally ramified over

L_{0}

and let

z

be a parameter of

L

, i.e. an element with the least possible positive mean of valuation on

L

. Since

L

is purely inseparable over

L_{0}

z^{p}

is a parameter of

L_{0}

We have

W : = C_{B} (L) = C_{D} (L) \sim = A_{1} \otimes_{\tilde{K}} L

is an unramified division algebra.

Consider an embedding

u^{'} : \bar{L} = {\bar{L}}_{0} ↪ L_{0}

. As it was shown in the proof of theorem 1 there is a lift

{\tilde{u}}^{'}

u^{'}

{\tilde{u}}^{'} : \bar{W} ↪ W

. Now consider the subalgebra

W^{'} = {\tilde{u}}^{'} (\bar{W}) ((z^{p}))

. We have

Z (W^{'}) = u^{'} (\bar{L}) ((z^{p})) = L_{0}

, so

W^{'}

is an unramified subalgebra in

B

. By Double Centralizer theorem,

B \sim = W^{'} \otimes_{L_{0}} C_{B} (W^{'})

, where

C_{B} (W^{'})

is a division algebra of degree

p

and contains

L_{0} (z) = L

, so it contains

\tilde{K}

and it is cyclic by Albert's theorem (th.12 in [1] ).

Now we can word by word repeat the arguments in the proof of theorem 12 in [1] to show that there exists a cyclic Galois extension

L^{'}

L_{0}

which is cyclic Galois over

Z (D)

, and

y

acts as a Galois automorphism on

L^{'} / Z (D)

which generates

G a l (L^{'} / Z (D))

. So, there is the cyclic subalgebra

D_{2} = (L^{'} / Z (D), a d (y), y^{p^{k}})

D

. Note that

A_{2} \subset D_{2}

, and

A_{2}

is known to be a good splittable algebra with

[{\bar{A}}_{2} : \bar{Z (A_{2})}] = [Γ_{A_{2}} : Γ_{Z (A_{2})}]

Since

{\bar{A}}_{2} = {\bar{D}}_{2}

and

Z (A_{2}) = \tilde{K}

is a purely inseparable extension of

Z (D)

D_{2}

is a good splittable algebra such that

{\bar{D}}_{2}

a field and

[{\bar{D}}_{2} : \bar{Z (D)}] = [Γ_{D_{2}} : Γ_{Z (D)}]

. By Double Centralizer theorem

D \sim = D_{1} \otimes_{Z (D)} D_{2}

, where

D_{1} = C_{D} (D_{2})

must be an unramified division algebra, which completes the proof.

□

Combining all results in this section, we get the following theorem.

Theorem 3 Let

D

be a finite dimensional good splittable central division algebra over a field

F = k ((t))

c h a r (F) = p > 0

, then

D \sim = D_{1} \otimes_{F} D_{2} \otimes_{F} A_{1} \otimes_{F} \dots \otimes_{F} A_{m}

, where

A_{i}

are cyclic division algebras such that

[{\bar{A}}_{i} : \bar{Z (D)}] = [Γ_{A_{i}} : Γ_{Z (D)}]

and

{\bar{A}}_{i} / \bar{Z (D)}

are simple purely inseparable field extensions,

D_{1}

is an inertially split division algebra,

(i n d (D_{1}), p) = 1

D_{2}

is an unramified division algebra (

D_{1}, D_{2}, A_{i}

may be trivial). If

c h a r F = 0

, then

D

is an inertially split division algebra.

6 Splittability and good splittability

In this section we collect some assorted results about a relation between splittable and good splittable division algebras and about splittable division algebras. We consider here only division algebras with the following property:

Z (\bar{D}) / \bar{Z (D)}

is a simple extension.

Proposition 6 Let

D

be a central division algebra over

F

c h a r D = p > 0

such that

Z (\bar{D}) = \bar{D}

and

[Z (\bar{D}) : \bar{F}] = p

Then

D

is a splittable algebra and the local height

i = i (u, z)

(in the situatuion when it is defined, i.e. when

α = i d

) does not depend on

u

and

z

. It is a good splittable algebra if

(i, p) = 1

. If

p | i

, then there exists a parameter

z

such that

z^{p} \in Z (D)

and any ”unramified” maximal subfield is cyclic Galois. So, in both cases

D

is a cyclic division algebra of degree

p

Proof. Since

\bar{D} / \bar{F}

is a simple extension, we have

[\bar{D} : \bar{F}] = [Γ_{D} : Γ_{F}]

. Indeed, consider the fields

E = F (s)

and

E^{'} = F (z)

, where

s

is any element such that

\bar{s}

is a primitive element of the extension

\bar{D} / \bar{F}

and

z

is any parameter of

D

. Then

[\bar{D} : \bar{F}] \leq [E : F] \leq [D : F]^{1 / 2} = ([\bar{D} : \bar{F}] [Γ_{D} : Γ_{F}])^{1 / 2}

, so

[\bar{D} : \bar{F}] \leq [Γ_{D} : Γ_{F}]

. From another hand side,

[Γ_{D} : Γ_{F}] \leq [E^{'} : F] \leq ([\bar{D} : \bar{F}] [Γ_{D} : Γ_{F}])^{1 / 2}

, so

[\bar{D} : \bar{F}] = [Γ_{D} : Γ_{F}]

. So,

D

is splittable division algebra of degree

p

Z (\bar{D}) / \bar{F}

is a separable extension, then

D

is a good splittable algebra by theorem 1 . So, we assume it is a purely inseparable extension,

Z (\bar{D}) = \bar{F} (\bar{u})

. For any lift

u

of the element

\bar{u}

let

u

be an embedding constructed in lemma 9 , i.e.

^{(z, u)} δ_{j}

is defined by the values

^{(z, u)} δ_{j} (u^{k})

for any

j

. By corollary 2 the local height

i (u, z)

does not depend on

z

, and by lemma 10

i (u, z)

does not depend on

u

. For arbitrary embedding

u^{'}

, since

^{(z, u^{'})} δ_{i (u^{'}, z)}

is a derivation and

\bar{D} / \bar{F}

is a simple extension,

^{(z, u^{'})} δ_{i (u^{'}, z)}

is completely defined by a value at

\bar{u}

. Therefore,

i (u^{'}, z) = w (z u^{'} (\bar{u}) z^{- 1} - u^{'} (\bar{u}))

and

i (u^{'}, z)

is completely defined by the lift

u^{'} (\bar{u})

. But arbitrary lift of

\bar{u}

defines an embedding, on which we have proved

i

does not depend. So,

i (u, z)

does not depend on

z

and

u

Now assume

p | i

Using lemma 3 , we can assume without loss of generality that

^{(z, u)} δ_{j} = 0

j

is not divisible by

p

Indeed, if

^{(z, u)} δ_{j} \neq 0

, then we apply lemma 3 , (ii) to show that there exists a parameter

z_{j}

such that

^{(z_{j}, u)} δ_{j} (u) = 0

and

^{(z_{j}, u)} δ_{k} =^{(z, u)} δ_{k}

for

k < j

^{(z_{j})} α = i d

. Since

^{(z_{j}, u)} δ_{j}

is a derivation by proposition 2 and by induction (similar arguments was already used in the proof of proposition 3 ), and since it is defined by the values on

u^{k}

, so by the values on

u

, we have

^{(z, u)} δ_{j} = 0

. Since for

j_{1} > j_{2}

we have

w (z_{j_{1}} - z_{j_{2}}) > j_{1} - i

, the sequence

{z_{j}}

convereges to a parameter

z^{'}

, which satisfies our condition.

So, there exists the subalgebra

A = u (\bar{D}) ((z^{p}))

. Let's show that

Z (D) \subset A

. Note that every element

a \in D

can be written as

a = a_{0} + a_{1} z + \dots + a_{p - 1} z^{p - 1}

, where

a_{i} \in A

. Note that

z^{k} A z^{- k} \subset A

for every

k

. So, if

a \in Z (D)

, then

z a_{j} z^{- 1} = a_{j}

and

u a_{j} z^{j} u^{- 1} = a_{j} z^{j}

for every

j

. For

j > 0

we have

a_{j} z^{j} = \sum_{k} a_{j k} z^{k p + j}

, so by corollary 2

u a_{j} z^{j} u^{- 1} \neq a_{j} z^{j}

. Therefore,

a = a_{0} \in A

Since

A \neq D

A

must be commutative, so

z^{p} \in Z (D)

. Moreover,

A / Z (D)

is cyclic Galois. Since the arguments work for arbitrary lift

u

of the element

\bar{u}

, arbitrary ”unramified” maximal subfield in

D

must be Galois over

F

Now let

(i, p) = 1

. Using lemma 3 , (iii) we can find a parameter

z

and a primitive element

s \in \bar{D}

such that

^{(z, u)} δ_{i} (s) = s c

, where

c \in \bar{F}

. Indeed, since

(i, p) = 1

, there exists

k

such that

1 - k i

is divisible by

p

. So, by lemma 3 , (iii) for a parameter

z^{'} = u (^{(z, u)} δ_{i} (\bar{u})^{k}) z

we have

^{(z^{'}, u)} δ_{i} (\bar{u}) \in \bar{F}

, so by lemma 10 , (iii)

^{(z^{'}, u)} δ_{i} (s) = 1

, where

s = {\bar{u}}^{(z^{'}, u)} δ_{i} (\bar{u})^{- 1}

. Now, there exists

k_{1}

such that

- i k_{1} - 1

is divisible by

p

, so for

z^{''} = s^{k_{1}} z^{'}

we have

^{(z^{''}, u)} δ_{i} (s) = s c

, where

c = s^{- i k_{1} - 1} \in \bar{F}

. It is easy to see that, since

s = \bar{u} a

, where

a \in \bar{F}

, the map

^{(z, u)} δ_{j}

is uniquely defined also by

^{(z, u)} δ_{j} (s^{k})

, so by

_{m}^{(z, u)} δ_{l} (s)

for

l \leq j

. So, we assume without loss of generality that

s = \bar{u}

z = z^{''}

Using lemma 10 , (ii) we can find a converge sequence

{u_{j}}

u_{j} \in D

j \geq i

such that

u_{j + 1} = u_{j} + b_{j} z^{j + 1 - i}

u_{i} = u

b_{j} \in u_{j} (\bar{D})

(here

u_{j}

is an embedding defined by

u_{j}

, see lemma 9 ) and

_{m}^{(z, u_{j})} δ_{k} (\bar{u}) {\bar{u}}^{- 1} \in \bar{F}

for all

k \leq j

and all

m

Indeed, suppose it is true for

j \geq i

. Let

_{m}^{(z, u_{j})} δ_{j + 1} (\bar{u}) = a_{0} + \dots a_{p - 1} {\bar{u}}^{p - 1}

a_{k} \in \bar{F}

Since

_{m}^{(z, u_{j})} δ_{i} =_{m}^{(z, u)} δ_{i} = m^{(z, u)} δ_{i}

, we have

_{m}^{(z, u_{j})} δ_{i} (a_{k} {\bar{u}}^{k}) - \frac{\partial}{\partial \bar{u}} (_{m}^{(z, u_{j})} δ_{i} (\bar{u})) a_{k} {\bar{u}}^{k} = (k - 1) m c a_{k} {\bar{u}}^{k} .

So,

u_{j + 1} = u_{j} - u_{j} (\sum_{k, k \neq 1} (k - 1)^{- 1} m^{- 1} c^{- 1} a_{k} {\bar{u}}^{k}) z^{j + 1 - i}

will satisfy our condition.

We will denote by

u

now a limit of the sequence

{u_{j}}

. Using induction and proposition 2 one can easily show that

_{m}^{(z, u)} δ_{j} ({\bar{u}}^{k}) {\bar{u}}^{- k} \in \bar{F}

for any integer

k

. So, there is the subalgebra

A = u (\bar{F}) ((z))

D

. Using similar arguments as in the case

p | i

, one can show that

A

contains

Z (D)

. Since

A \neq D

, it must be commutative, so

u^{p} \in Z (D)

. Then

u

is a good embedding, which completes the proof.

□

Let

D

be a splittable division algebra and let

Z (\bar{D}) / \bar{Z (D)}

be a purely inseparable extension. As it was shown in the proof of lemma 11 , then there exists a parameter

z

D

such that

^{(z, u)} δ_{i} |_{Z (\bar{D})} \neq 0

, where

i = i (u, z)

is a local height. Though

D

may be not a good splittable algebra, the arguments from there are valid for every splittable algebra. We will call such a parameter an appropriate parameter, and the number

i (u) = {max}_{z} i (u, z) = i (u, z)

for an appropriate parameter a semilocal height. Let's prove the following simple lemma.

Lemma 12 Let

D

be a splittable central division

p

-algebra over

F

, where

p = c h a r D > 0

, and let

Z (\bar{D}) = \bar{F} (s)

be a simple extension over

\bar{F}

. Then i) there exists an embedding

u

such that

_{l}^{(z, u)} δ_{j} |_{Z (\bar{D})}

is defined by the values

_{l}^{(z, u)} δ_{j} (s^{k})

for any

j, l, z

(as in lemma 9 ); ii)

[Z (\bar{D}) : \bar{F}] = [Γ_{D} : Γ_{F}]

; iii) if

α |_{Z (\bar{D})} \neq i d

i (u)

is divisible by

p

, then there exists a subalgebra

A = u (\bar{D}) ((z))

for some appropriate parameter

z

such that

Z (D) \subset Z (A)

. Moreover,

Z (A)

is a cyclic Galois extension over

Z (D)

Proof. i) For arbitrary embedding

u

consider the field

E = u (Z (\bar{D})) F \subset D

and the centralizer

W = C_{D} (E)

. We have

\bar{W} = \bar{D}

and so

Z (\bar{W}) = \bar{E}

. Therefore,

W

must be an unramified division algebra, and by theorem 1 there exists a lift on

\bar{W}

of arbitraryembedding

\bar{E} ↪ E

. Now we can take an embedding defined by the element

s

as in lemma 9 . It's lift will be desired embedding. We will denote this embedding also by

s

ii) By proposition 1.7. in [6] the basic homomorphism

θ_{D}

(see introduction) is surjective.

So, it is sufficient to prove the assertion only for the centralizer

C_{D} (K)

, where

K

is a lift of a Galois part of the extension

Z (\bar{D}) / \bar{F}

. So, we will assume below

Z (\bar{D}) / \bar{F}

is a purely inseparable extension.

Consider a maximal separable subfield

M

\bar{D}

, and let

M^{'}

be a separable part of the extension

M / \bar{F}

. By [6] , th.2.8, th.2.9. there exists an inertial lift of

M^{'}

D

, say

\tilde{M}

Consider the centralizer

B = C_{D} (\tilde{M})

. Then

\bar{B}

is a field. Our assertion will be proved if we show it for

B

, since

[\tilde{M} : F] = i n d (\bar{D})

and

[D : F] = i n d (\bar{D})^{2} [Z (\bar{D}) : \bar{F}] [Γ_{D} : Γ_{F}]

Since

\bar{B} / \bar{Z (B)}

is a simple extension, we can repeat the arguments from the beginning of proposition 6 .

iii) If

α |_{Z (\bar{D})} \neq i d

, consider the parameter

z

from proposition 3 . Then, clearly,

A = u (\bar{D}) ((z))

will be a subalgebra with the center

K

, which is an inertial lift of a Galois part of the extension

Z (\bar{D}) / \bar{F}

Assume

α |_{Z (\bar{D})} = i d

and

i (u)

is divisible by

p

. Let

z

be an appropriate parameter.

Using lemma 3 , we can prove that

^{(z, u)} δ_{j} = 0

j

is not divisible by

p

Indeed, let

^{(z, u)} δ_{j} \neq 0

be the first map with this property for

(j, p) = 1

. If

^{(z, u)} δ_{j} |_{Z (\bar{D})} = 0

, then we apply lemma 3 , (i) to show that there exists a parameter

z_{j}

such that

^{(z_{j}, u)} δ_{j} = 0

and

^{(z_{j}, u)} δ_{k} =^{(z, u)} δ_{k}

for

k < j

^{(z_{j})} α = i d

, since

^{(z, u)} δ_{j}

is a derivation by proposition 2 and by induction (similar arguments was already used in the proof of proposition 3 ) and so it is an inner derivation by Scolem-Noether theorem.

^{(z, u)} δ_{j} |_{Z (\bar{D})} \neq 0

, then we apply lemma 3 , (ii) to show that there exists a parameter

z_{j}

such that

^{(z_{j}, u)} δ_{j} (s) = 0

and

^{(z_{j}, u)} δ_{k} =^{(z, u)} δ_{k}

for

k < j

^{(z_{j})} α = i d

. Since

^{(z_{j}, u)} δ_{j}

is a derivation and since its restriction on

Z (\bar{D})

is defined by the values on

s^{k}

, so by the values on

s

, we have

^{(z, u)} δ_{j} |_{Z (\bar{D})} = 0

, and we reduce the problem to the previous case. Since for

j_{1} > j_{2}

we have

w (z_{j_{1}} - z_{j_{2}}) > j_{1} - i

, the sequence

{z_{j}}

convereges to a parameter

z^{'}

, which satisfies our condition.

Therefore, there exists a subalgebra

A = u (\bar{D}) ((z^{'}))

D

. Using the same arguments as in proposition 6 one can show that

Z (D) \subset Z (A)

Since

z^{'}

preserves

A

, it preserves the centre

Z (A)

¿From the other hand side, it acts nontrivially on it. So,

Z (A)

is a cyclic Galois extension of degree

p

, and

a d (z^{'})

generates its Galois group.

□

This lemma shows that the study of splittable

p

-algebras over

F

can be reduced to the study of splittable

p

-algebras with a purely inseparable extension

Z (\bar{D}) / \bar{F}

and

(i (u), p) = 1

Definition 6 Let

D

be a splittable division

p

-algebra with a purely inseparable extension

Z (\bar{D}) / \bar{F}

. For any element

a \in \bar{D}

define the number

d_{D} (a) = {max}_{u, z} w (z^{- i (u, a)} u (a) z^{i (u, a)} - u (a) - u (_{- i (u, a)}^{(z, u)} δ_{i (u, a)} (a)) z^{i (u, a)}) \in N \cup \infty,

where parameters

z

are taken from the set of appropriate parameters and

i (u, a)

was defined in corollary 2 .

It seems that the number

d_{D} (a)

will play the role of a higher order level in a splittable division algebra. We will see that it codes a part of information about a division algebra.

Lemma 13 Let

D

be a splittable division

p

-algebra,

p > 2

, with a purely inseparable simple extension

Z (\bar{D}) / \bar{F}

, let

u

be some fixed embedding

u : \bar{D} ↪ D

Suppose

Z (\bar{D}) = \bar{F} (a)

and

(i (u, a), p) = 1

. Suppose

d (u, a) \leq 2 i (u, a)

Let

z

be a parameter such that

^{(z, u)} δ_{i (u, a)} (_{- i (u, a)}^{(z, u)} δ_{i (u, a)} (a)) = 0

^{(z)} α = i d

and

_{- i (u, a)}^{(z, u)} δ_{q} |_{F_{p} (a)} = 0

for

i (u, a) < q < d (u, a)

. Put

j (k) : = i (u, a^{p^{k}})

Suppose for every

k \geq 1

a parameter

z_{k}

such that

_{- j (k)}^{(z_{k}, u)} δ_{r} |_{F_{p} (a^{p^{k}})} = 0

for

j (k) < r < d (u, a^{p^{k}})

satisfy a condition

^{(z_{k}, u)} δ_{i (u, a)} =^{(z, u)} δ_{i (u, a)}

^{(z_{k})} α =^{(z)} α

Suppose for every

k \geq 1

we have

d (u, a^{p^{k}}) - j (k) = d (u, a) - j (0)

Wild division algebras over Laurent series fields

A.B. Zheglov * * Supported by the Graduiertenkolleg ”Geometrie und Nichtlineare Analisys” of the DFG † † e-mail address: azheglov@mathematik.hu-berlin.de

November 27, 2006