Why is the division closure defined with a witness t ∈ T rather than an arbitrary ring element?

Because the conclusion has to be sign information. From at ∈ T and t ∈ T one may cancel t in any ordering above T, since orderings satisfy P = P. With an arbitrary multiplier of unknown sign, nothing follows: a (-1) ∈ T says a is *negative*.

Division-Closed Preorderings | KEVOS® Mathematics

Q: Does (17.13) require the ring to be formally real?

Implicitly, yes: the statement presupposes a preordering T, and (17.11) then makes R formally real. If no preordering exists there are also no orderings and both sides of the formula are undefined rather than empty.

Q: Is T the smallest division-closed preordering containing T?

Yes. It is division-closed by (17.14), being an intersection of orderings, and any division-closed preordering S T satisfies T ⊆ S = S. So T T is a genuine closure operator, idempotent and monotone.

Q: How is this related to the Positivstellensatz?

Both answer "what is forced by all orderings?", but with different data. (17.13) is the abstract statement for a preordering in an arbitrary ring; the Positivstellensatz is the commutative refinement that additionally describes the multipliers explicitly in terms of the generators of a semialgebraic set. The Motzkin example shows why the multipliers cannot be dispensed with.

Q: Why does the field case collapse?

Because one can divide. Given ab^2 ∈ T(F), multiply by (b^-1)^2 ∈ T(F) to get a ∈ T(F). Hence T(F) = T(F) and total positivity coincides with being a sum of squares. The same argument works in any division ring, as in (17.15a).

Q: Can a preordering have exactly one ordering above it?

Yes, and then T is that ordering — the intersection of a one-element family. T(Q) is the standard example: it is already total, hence maximal, hence equal to the unique ordering of Q.

Executive Summary

A preordering $T$ records positivity information; the orderings above it are its total completions. Which information is forced by all of them? The answer is $(17.13)$ : the intersection of all orderings containing $T$ is the division closure

\bar{T} = {a \in R : a t \in T for some t \in T} = {a \in R : a b^{2} \in T for some b \neq 0} .

(17.12)

So a preordering is an intersection of orderings exactly when it is closed under dividing out a positive factor $(17.14)$ , and an element is positive in every ordering exactly when some $a b^{2}$ is a sum of permuted doubled products $(17.15)$ . Over a field this is Artin's theorem — totally positive equals sum of squares — the tool he used on Hilbert's 17th problem.

$\bar{T}$ Intersection of orderings

4Equivalent descriptions

$a b^{2} \in T$ Total positivity test

1927Artin, Hilbert 17

Overview

Two operations turn a preordering into something bigger. Zorn's Lemma produces a maximal preordering, an ordering, but destroys canonicity — there are usually many. Intersecting all of those orderings restores canonicity and lands back inside the preorderings, since intersections of orderings satisfy $(17.5)$ and $(17.6)$ . The question this page answers is which preorderings are fixed by the round trip.

The answer is purely algebraic and involves no orderings at all. In an ordering $P$ , if $a t \in P$ with $t \in P$ then $a \in P$ : otherwise $- a \in P$ , hence $- a t \in P$ , contradicting $P \cap (- P) = \emptyset$ . So every intersection of orderings is division-closed. $(17.13)$ says the converse holds, and does so by producing an ordering avoiding any prescribed $a \notin \bar{T}$ .

The proof needs exactly one new idea beyond Preorderings in Rings: applying the auxiliary construction $T_{b}$ to $b = - a$ . Failure of $T_{- a}$ to be a preordering is, by $(17.8)$ , the identity $t' + (- a) t = 0$ , that is $a t = t' \in T$ , that is $a \in \bar{T}$ . So the division closure is precisely the set of elements whose negatives cannot be made positive.

Learning Objectives

Recall the four descriptions of $\bar{T}$ from $(17.12)$ and use whichever is convenient.
Prove $\bar{P} = P$ for an ordering $P$ , and deduce $\bar{T} \subseteq ⋂ P$ .
Run the $T_{- a}$ argument to produce an ordering omitting a prescribed $a \notin \bar{T}$ .
State $(17.14)$ and use it to decide whether a given preordering is an intersection of orderings.
Apply $(17.15)$ to test total positivity by a single membership $a b^{2} \in T (R)$ .
Show that $T (ℝ [x, y])$ is not division-closed using the Motzkin polynomial.

Definitions

Definition(17.14)Division-closed preordering

A preordering $T \subseteq R ∖ {0}$ is division-closed if, for all $a \in R$ and $t \in T$ , $a t \in T$ implies $a \in T$ . Equivalently $\bar{T} = T$ , since $T \subseteq \bar{T}$ always holds.

By $(17.12)$ the one-sided condition is not restrictive: $a t \in T$ , $t' a \in T$ , $a b^{2} \in T$ and $b^{2} a \in T$ all define the same set $\bar{T}$ , so division-closedness may be tested on whichever side is convenient.

$\bar{T}$: The division closure; a preordering containing $T$ , by (17.13).
Totally positive: $a \neq 0$ lies in every ordering of $R$ . Equivalently $a \in \bar{T (R)}$ .
$T (R)$: The weak preordering: sums of permuted doubled products. Contained in every preordering, so its orderings are all the orderings of $R$ .
Saturation: The name used in commutative real algebra for the same operation $T \mapsto \bar{T}$ .

Division-closedness is a statement about cancelling a positive factor, not about the existence of inverses. In a division ring every preordering is automatically division-closed; in a polynomial ring in two variables the weak preordering is not.

Core Concepts

Orderings are their own division closures

Let $P$ be an ordering and suppose $a p \in P$ with $p \in P$ . Then $a \neq 0$ , since $0 \cdot p = 0 \notin P$ . If $a \notin P$ , totality gives $- a \in P$ , so $(- a) p \in P \cdot P \subseteq P$ ; but $(- a) p = - (a p)$ , so both $a p$ and $- (a p)$ lie in $P$ , contradicting $P \cap (- P) = \emptyset$ . Hence $\bar{P} = P$ .

This one-line computation supplies the easy inclusion of $(17.13)$ : for any ordering $P \supseteq T$ we get $\bar{T} \subseteq \bar{P} = P$ , so $\bar{T}$ is inside the intersection.

The hard inclusion, and where $T_{- a}$ enters

To show the intersection is no bigger than $\bar{T}$ , one must construct, for each $a \notin \bar{T}$ , an ordering containing $T$ but not $a$ . The natural candidate is an ordering containing $- a$ . The auxiliary set $T_{- a}$ is the smallest candidate preordering containing $T$ and $- a$ , and $(17.8)$ says it fails only when there is an identity

t' + (- a) t = 0 with t, t' \in T, that is a t = t' \in T, that is a \in \bar{T} .

(D.1)

The failure condition for $T_{- a}$ is literally the membership condition for $\bar{T}$ . That coincidence is the theorem.

a \notin \bar{T}

T_{- a}

is a preorderingan ordering

P \supseteq T

with

- a \in P

a \notin P

Why the closure is needed at all

In a field one may divide an inequality by a positive square, so the weak preordering is already division-closed and no correction is required. In a general commutative ring the correction is real: a polynomial can be forced positive by every ordering without itself being a sum of squares. The Motzkin polynomial in the worked example is the standard witness.

Key Results

Theorem(17.13)Division closure equals intersection of orderings

Let $R$ be a ring and $T \subseteq R ∖ {0}$ a preordering. Then

\bar{T} = ⋂ {P : P is an ordering of R with P \supseteq T} .

(17.13)

In particular $\bar{T}$ is itself a preordering, being an intersection of a non-empty family of orderings.

Proof

Write $T'$ for the right-hand intersection; it is non-empty as a family, since $T$ extends to an ordering by $(17.10)$ and Zorn's Lemma.

** $\bar{T} \subseteq T'$ .** Let $P \supseteq T$ be an ordering. Then $\bar{T} \subseteq \bar{P}$ , because a witness $t \in T$ for $a \in \bar{T}$ is also a witness in $P$ . And $\bar{P} = P$ by the computation above. Hence $\bar{T} \subseteq P$ for every such $P$ , so $\bar{T} \subseteq T'$ .

** $T' \subseteq \bar{T}$ .** Let $a \neq 0$ with $a \notin \bar{T}$ ; we produce an ordering $P \supseteq T$ with $a \notin P$ . Consider $T_{- a}$ , the auxiliary set of $(17.8)$ for the element $- a$ . If $T_{- a}$ were not a preordering, $(17.8)$ would give $t, t' \in T$ with $t' + (- a) t = 0$ , i.e. $a t = t' \in T$ , i.e. $a \in \bar{T}$ — contrary to assumption. So $T_{- a}$ is a preordering. It contains $T$ and $- a$ , and by $(17.10)$ together with Zorn's Lemma it extends to an ordering $P \supseteq T_{- a} \supseteq T$ . Since $- a \in P$ and $P \cap (- P) = \emptyset$ , we get $a \notin P$ , hence $a \notin T'$ .

Finally $0 \notin T'$ , so the two sets agree on all of $R$ .

Corollary(17.14)Which preorderings are intersections of orderings

A preordering $T \subseteq R ∖ {0}$ is an intersection of orderings of $R$ if and only if $T$ is division-closed, i.e. $a t \in T$ with $t \in T$ implies $a \in T$ .

Proof. If $T = ⋂_{i} P_{i}$ and $a t \in T$ with $t \in T$ , then for each $i$ we have $a t \in P_{i}$ and $t \in P_{i}$ , so $a \in \bar{P_{i}} = P_{i}$ ; hence $a \in T$ . Conversely if $T$ is division-closed then $T = \bar{T}$ , and $(17.13)$ exhibits $\bar{T}$ as an intersection of orderings.

Corollary(17.15)Test for total positivity

Let $R$ be a formally real ring and $a \in R ∖ {0}$ . Then $a$ is positive in every ordering of $R$ if and only if there exists $b \in R ∖ {0}$ with $a b^{2} \in T (R)$ .

Proof. Every ordering of $R$ contains the weak preordering $T (R)$ , so the orderings containing $T (R)$ are all of them. By $(17.13)$ their intersection is $\bar{T (R)}$ , and by $(17.12)$ membership in $\bar{T (R)}$ is exactly the condition $a b^{2} \in T (R)$ for some $b \neq 0$ .

Corollary(17.15a)Fields and division rings

Let $D$ be a division ring and $T$ a preordering of $D$ . Then $T$ is division-closed, so $T$ is the intersection of the orderings of $D$ containing it. Consequently, in a formally real field $F$ , an element is totally positive if and only if it is a sum of squares.

Proof. For $t \in T$ the arrangement $t^{- 1} t t^{- 1}$ is an instance of $(17.6)$ with the doubled element $t^{- 1}$ and the inserted element $t$ , so $t^{- 1} = t^{- 1} t t^{- 1} \in T$ . If now $a t \in T$ with $t \in T$ , then $a = (a t) t^{- 1} \in T \cdot T \subseteq T$ . Division-closedness gives $T = \bar{T}$ , and $(17.14)$ applies. Taking $T = T (F)$ , whose elements are the nonzero sums of squares, gives Artin's statement.

Proof Techniques and Method

How these proofs work, and which move to reuse.

Fix the element you want excludedTo prove an intersection is no larger than a candidate set, take

a

outside the candidate and aim to build one ordering missing

a

Force the opposite signBuild

T_{- a}

: the smallest closed set containing

T

and

- a

. It automatically satisfies both preordering axioms.

Read off properness from (17.8)The only failure is

0 \in T_{- a}

, which is the identity

a t = t' \in T

— precisely the membership you assumed false.

Zorn, then concludeEnlarge

T_{- a}

to an ordering. It contains

T

and

- a

, hence omits

a

, so

a

is not in the intersection.

The same three-step template — assume outside, build auxiliary cone, extend by Zorn — proves $(17.16)$ for extension to a larger ring. Recognising it saves rereading each proof.

Worked Example

The Motzkin polynomial: $T (R)$ is not division-closed

Work in the commutative ring $R = ℝ [x, y]$ , where $T (R)$ is the set of nonzero sums of squares of polynomials. Let

M (x, y) = x^{4} y^{2} + x^{2} y^{4} - 3 x^{2} y^{2} + 1 .

(E.1)

Motzkin's polynomial, 1967.

$M$ is nonnegative on $ℝ^{2}$ by the arithmetic–geometric mean inequality applied to $x^{4} y^{2}$ , $x^{2} y^{4}$ and $1$ , whose geometric mean is $x^{2} y^{2}$ . But $M$ is not a sum of squares of polynomials — the classical degree and coefficient analysis rules it out — so $M \notin T (R)$ .

On the other hand $(x^{2} + y^{2}) M$ is a sum of squares of polynomials, a standard identity. Since $x^{2} + y^{2} \in T (R)$ , the first description in $(17.12)$ gives $M \in \bar{T (R)}$ . Hence

M \in \bar{T (R)} ∖ T (R),

(E.2)

So $T (ℝ [x, y])$ is a preordering that is not division-closed, and by $(17.14)$ it is not an intersection of orderings.

By $(17.15)$ , $M$ is totally positive: it is positive in every ordering of $ℝ [x, y]$ despite not being a sum of squares. This is precisely the phenomenon that makes denominators unavoidable in Hilbert's 17th problem.

A totally positive element with an explicit certificate

In $F = ℚ (\sqrt{2})$ there are exactly two orderings, given by the two real embeddings. The element $3 + \sqrt{2}$ is positive under both ( $\approx 4.414$ and $\approx 1.586$ ), so it is totally positive, and $(17.15 a)$ predicts it is a sum of squares. It is:

3 + \sqrt{2} = (1 + \frac{\sqrt{2}}{2})^{2} + 1^{2} + (\frac{1}{2})^{2} + (\frac{1}{2})^{2} .

(E.3)

Check: $(1 + \sqrt{2} / 2)^{2} = 1 + \sqrt{2} + \frac{12}{=} \frac{32}{+} \sqrt{2}$ , and $\frac{32}{+} 1 + \frac{14}{+} \frac{14}{=} 3$ .

By contrast $1 + \sqrt{2}$ is positive in one ordering and negative in the other ( $1 - 1.414 < 0$ ), so it is not totally positive and, by $(17.15 a)$ , cannot be a sum of squares in $F$ .

Process and Workflow

Is a given nonzero $a$ positive in every ordering of a formally real ring $R$ ?

Find $b \neq 0$ with $a b^{2} \in T (R)$ Then

a \in \bar{T (R)}

and

a

is totally positive, by

(17.15)

. A single witness

b

suffices; no enumeration of orderings is needed.

Find $t \in T (R)$ with $a t \in T (R)$ The same conclusion by the first description in

(17.12)

. Often easier, since

t

ranges over all sums of permuted doubled products, not just squares.

Exhibit one ordering with $a$ negativeThen

a

is not totally positive. Constructively, show

T (R)_{- a}

is proper and extend it by Zorn.

Neither worksThe question is genuinely hard: for polynomial rings it is equivalent to a positivity certificate problem with no a priori degree bound.

Identify the preorderingUsually

T (R)

, or the preordering generated by a set of elements you wish to declare positive.

Compute the division closureSearch for witnesses

t

with

a t \in T

. Each success adds an element to

\bar{T}

Test division-closednessIf

\bar{T} = T

, the preordering is an intersection of orderings by

(17.14)

and carries no hidden positivity.

Otherwise pass to $\bar{T}$ Replace

T

\bar{T}

, which is a preordering with the same orderings above it and is now division-closed.

Comparison and Classification

Division-closedness in standard settings
Ring $R$	Preordering	Division-closed?	Reason
Any division ring $D$	any preordering $T$	yes	$t^{- 1} \in T$ , so positives cancel, $(17.15 a)$
Formally real field $F$	$T (F)$	yes	divide by $b^{2}$
$ℝ [x]$	$T (ℝ [x])$	yes	nonnegative one-variable polynomials are sums of two squares
$ℝ [x, y]$	$T (ℝ [x, y])$	no	Motzkin's $M$ lies in $\bar{T} ∖ T$
General formally real ring	$T (R)$	not in general	Lam, §17 Exercise 8, gives an explicit example
Any $R$	$\bar{T}$	yes	by construction, $\bar{\bar{T}} = \bar{T}$

What each object controls
	Determines the orderings above it	Closed under sums	Total	Canonical
Preordering $T$	yes	yes	no	no
Division closure $\bar{T}$	yes	yes	no	yes
Ordering $P$	yes	yes	yes	no
Weak preordering $T (R)$	yes	yes	no	yes

What each object controls

$T$ and $\bar{T}$ have exactly the same orderings above them, which is why replacing $T$ by $\bar{T}$ costs nothing and buys $(17.14)$ .

Relationship Map

$R ∖ {0}$ all nonzero elements

Ordering $P$ total; maximal preordering

Division closure $\bar{T}$ intersection of all orderings above

T

Preordering $T$ the data you started with

Weak preordering $T (R)$ smallest of all

Each band is contained in the one outside it. The two inner bands coincide exactly when $T$ is division-closed; the two outer bands coincide exactly when $T$ has a unique ordering above it.

T (R) \subseteq T

\bar{T} = ⋂ P

T

division-closed

\Leftrightarrow

T = ⋂ P

Applications and Industry Use

Applications here means where this structure is used — inside mathematics and in the engineering and computing disciplines that consume it.

Real algebra

Hilbert's 17th problem

Artin's solution runs exactly along $(17.13)$ : a positive semidefinite rational function is positive in every ordering of $ℝ (x_{1}, \dots, x_{n})$ , hence lies in the weak preordering of that field, hence is a sum of squares of rational functions.

Optimisation

Why denominators appear

The Motzkin example is the reason sums-of-squares relaxations of polynomial optimisation need multipliers. Practical solvers search for a certificate $σ M = τ$ with $σ, τ$ sums of squares — the division-closure condition, made numerical.

Control engineering

Lyapunov certificates

Proving a polynomial vector field stable means certifying positivity of a candidate Lyapunov function. Certificates are searched for in a preordering generated by the constraint polynomials, with multipliers exactly as above.

Ring theory

Recovering a preordering from its orderings

$(17.14)$ says the map from preorderings to families of orderings loses nothing precisely on the division-closed ones. This is the order-theoretic analogue of a radical or closure operator elsewhere in ring theory.

Honestly stated: within noncommutative ring theory the division closure is infrastructure — it is what makes "totally positive" computable in principle. The visible downstream engineering use is in the commutative shadow, where the same condition becomes a semidefinite program.

Computational Notes

Computational notes cover algorithms, cost and library behaviour rather than manufacturing process.

For $R = ℝ [x_{1}, \dots, x_{n}]$ and fixed degree $d$ , deciding whether a polynomial is a sum of squares of polynomials of degree $\leq d / 2$ is a semidefinite feasibility problem of size polynomial in $(\binom{n + d / 2}{n})$ — solvable in practice by SOSTOOLS, YALMIP or SumsOfSquares.jl.
Deciding total positivity — membership in $\bar{T (R)}$ — is deciding nonnegativity on $ℝ^{n}$ , which is decidable by real quantifier elimination but doubly exponential in the number of variables.
Searching for a division-closure witness means searching for a multiplier: find sums of squares $σ, τ$ with $σ a = τ$ . Fixing the degree of $σ$ makes this a semidefinite program; no a priori bound on that degree is known in general.
The passage from $T$ to $\bar{T}$ is not effective for arbitrary rings, and the orderings supplied by $(17.13)$ come from Zorn's Lemma, so they are not computable objects.
For noncommutative polynomial rings there is a genuine positive result: a noncommutative polynomial nonnegative on all tuples of symmetric matrices is a sum of hermitian squares, and this is again testable by semidefinite programming — the free analogue of the commutative theory.

Failure Modes and Common Mistakes

Do not assume $\bar{T}$ is bigger than $T$ ; for division rings and for one-variable polynomial rings over formally real fields it is not.
Do not conclude from $(17.13)$ that any particular ordering is computable; the theorem is an intersection formula, not a construction.
Do not apply $(17.15)$ to a ring that is not formally real. If $0 \in T (R)$ there are no orderings, and every statement about total positivity is vacuous or false.
Do not confuse the commutative notion of preordering, which contains $0$ and all squares, with Lam's, which excludes $0$ ; membership statements differ at the boundary.

Quick Reference

Division closure

\bar{T} = {a : a t \in T

for some

t \in T}

Equivalent forms

t' a \in T

a b^{2} \in T

b^{2} a \in T

— all the same set,

(17.12)

(17.13)

\bar{T} = ⋂ {P

ordering

, P \supseteq T}

(17.14)

T

is an intersection of orderings

\Leftrightarrow

T

is division-closed

(17.15)

a

totally positive

\Leftrightarrow

a b^{2} \in T (R)

for some

b \neq 0

Division ringsevery preordering is division-closed

Failure

T (ℝ [x, y])

: Motzkin's

M \in \bar{T} ∖ T

Idempotence

\bar{\bar{T}} = \bar{T}

, and

T

\bar{T}

have the same orderings above them

Choosing the right statement
Question	Use
What do all orderings above $T$ agree on?	$(17.13)$ : exactly $\bar{T}$
Is $T$ recoverable from its orderings?	$(17.14)$ : yes iff division-closed
Is $a$ positive in every ordering?	$(17.15)$ : test $a b^{2} \in T (R)$
Is a sum of squares needed, or only a multiplier?	Compare $T (R)$ with $\bar{T (R)}$
Does the ring have any ordering at all?	$(17.11)$ : formal reality

Frequently Asked Questions

Why is the division closure defined with a witness

t \in T

rather than an arbitrary ring element?

Because the conclusion has to be sign information. From $a t \in T$ and $t \in T$ one may cancel $t$ in any ordering above $T$ , since orderings satisfy $\bar{P} = P$ . With an arbitrary multiplier of unknown sign, nothing follows: $a \cdot (- 1) \in T$ says $a$ is negative.

Does

(17.13)

require the ring to be formally real?

Implicitly, yes: the statement presupposes a preordering $T$ , and $(17.11)$ then makes $R$ formally real. If no preordering exists there are also no orderings and both sides of the formula are undefined rather than empty.

\bar{T}

the smallest division-closed preordering containing

T

Yes. It is division-closed by $(17.14)$ , being an intersection of orderings, and any division-closed preordering $S \supseteq T$ satisfies $\bar{T} \subseteq \bar{S} = S$ . So $T \mapsto \bar{T}$ is a genuine closure operator, idempotent and monotone.

How is this related to the Positivstellensatz?

Both answer "what is forced by all orderings?", but with different data. $(17.13)$ is the abstract statement for a preordering in an arbitrary ring; the Positivstellensatz is the commutative refinement that additionally describes the multipliers explicitly in terms of the generators of a semialgebraic set. The Motzkin example shows why the multipliers cannot be dispensed with.

Why does the field case collapse?

Because one can divide. Given $a b^{2} \in T (F)$ , multiply by $(b^{- 1})^{2} \in T (F)$ to get $a \in T (F)$ . Hence $T (F) = \bar{T (F)}$ and total positivity coincides with being a sum of squares. The same argument works in any division ring, as in $(17.15 a)$ .

Can a preordering have exactly one ordering above it?

Yes, and then $\bar{T}$ is that ordering — the intersection of a one-element family. $T (ℚ)$ is the standard example: it is already total, hence maximal, hence equal to the unique ordering of $ℚ$ .

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §17, results (17.13)–(17.15), pp. 281–282, with Exercises 8 and 9.
E. Artin, Über die Zerlegung definiter Funktionen in Quadrate, Abhandlungen aus dem Mathematischen Seminar der Universität Hamburg 5 (1927). The solution of Hilbert's 17th problem by the totally-positive criterion.
T. S. Motzkin, The arithmetic-geometric inequality, in Inequalities (O. Shisha, ed.), Academic Press, 1967. The first explicit positive semidefinite polynomial that is not a sum of squares.
T. Y. Lam, Orderings, Valuations and Quadratic Forms, CBMS Regional Conference Series in Mathematics 52, American Mathematical Society, 1983.
M. Marshall, Positive Polynomials and Sums of Squares, Mathematical Surveys and Monographs 146, American Mathematical Society, 2008. Saturation of preorderings and Positivstellensatz certificates.
A. Prestel and C. N. Delzell, Positive Polynomials: From Hilbert's 17th Problem to Real Algebra, Springer-Verlag, 2001.

AI Suggested Questions

What degree bounds are known for the multiplier in a representation certifying membership in the division closure?
How does the division closure behave under Ore localisation and under passing to a ring of quotients?
Is there a noncommutative Positivstellensatz that refines $(17.13)$ with explicit multipliers?
For which finitely generated commutative rings is the weak preordering division-closed?
How do sums of hermitian squares in free algebras relate to Lam's permuted doubled products?
What is the structure of the lattice of division-closed preorderings of a formally real ring?
Can the Motzkin phenomenon occur in a noncommutative ring with no commutative quotient of dimension two?

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

Orderings are their own division closures

The hard inclusion, and where T−a enters

Why the closure is needed at all

06Key Results

07Proof Techniques and Method

08Worked Example

The Motzkin polynomial: T(R) is not division-closed

A totally positive element with an explicit certificate

09Process and Workflow

10Comparison and Classification

11Relationship Map

12Applications and Industry Use

Hilbert's 17th problem

Why denominators appear

Lyapunov certificates

Recovering a preordering from its orderings

13Computational Notes

14Failure Modes and Common Mistakes

15Quick Reference

16Frequently Asked Questions

17Related KEVOS Topics

18References

19AI Suggested Questions

Continue learning

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

The hard inclusion, and where $T_{- a}$ enters

Key Results

Proof Techniques and Method

Worked Example

The Motzkin polynomial: $T (R)$ is not division-closed

Process and Workflow

Comparison and Classification

Relationship Map

Applications and Industry Use

Computational Notes

Failure Modes and Common Mistakes

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions