Closed Semialgebraic Type Sets

• Closed semialgebraic type sets are subsets of ℝⁿ defined by finitely many polynomial constraints that are closed in the Euclidean topology and serve as a basis for classification.
• They are studied using topological, spectral, and computational invariants to provide effective finite classification and reconstruction from function rings.
• Advanced C¹-triangulations and quantitative separation methods enable differential integration and optimization, linking algebraic, topological, and computational perspectives.

Closed semialgebraic sets are semialgebraic subsets of $\mathbb{R}^n$ that are closed in the Euclidean topology; in the basic closed case they are cut out by finitely many polynomial equalities and non-strict polynomial inequalities (Korda et al., 2022, Jeronimo et al., 2011). The phrase “closed semialgebraic type sets” is not introduced as a standard formal term in the papers considered here. Instead, the literature studies closed semialgebraic sets through several notions of type and classification, notably topological type, spectral type, homological invariants, and the logical structure of lattices of closed sets (Bürgisser et al., 2018, Demdah et al., 2024, Fernando et al., 2013, Tressl, 2016). In parallel, closed semialgebraic sets admit $C^1$-triangulated models that support differential-form integration and Stokes-type formulas (Ohmoto et al., 2015).

1. Foundational notions and scope

A semialgebraic set $X\subset \mathbb{R}^m$ is defined by finitely many polynomial equations, inequalities, and Boolean combinations, and a map is semialgebraic when its graph is semialgebraic. In the triangulation framework, the ambient notion is broadened to “locally semialgebraic”: $X\subset \mathbb{R}^m$ is locally semialgebraic if $X\cap D$ is semialgebraic for every compact $m$-disk $D$, and a map is locally semialgebraic if its restriction to every compact semialgebraic subset is semialgebraic. The same framework is stated to hold over any real closed field, including possibly non-archimedean real closed fields, and more generally in o-minimal, subanalytic, or $X$-categories satisfying the relevant axioms (Ohmoto et al., 2015).

For basic closed semialgebraic geometry, a standard class is

$$T=\{x\in\mathbb R^n \mid f_1(x)=\cdots=f_\ell(x)=0,\; f_{\ell+1}(x)\ge 0,\ldots,f_m(x)\ge 0\},$$

where the $f_i\in \mathbb Z[x_1,\ldots,x_n]$ have bounded degree and coefficient size. This is the setting for explicit optimization and separation theorems on connected components of closed semialgebraic sets (Jeronimo et al., 2011).

A second uniform description uses complexity bounds. A semialgebraic set $C^1$0 has complexity at most $C^1$1 if it can be described by at most $C^1$2 polynomial equations or inequalities, each of degree at most $C^1$3. Closed semialgebraic sets are included as a subclass of these bounded-complexity families (Demdah et al., 2024).

From the computational-topology side, closed semialgebraic sets also appear as sets defined by lax Boolean formulas using only the atomic predicates

$C^1$4

combined by $C^1$5 and $C^1$6, but without complements. The relevant paper explicitly states that it does not introduce a formal notion called “semialgebraic type sets” or “closed semialgebraic type sets” (Bürgisser et al., 2018).

2. Topological and spectral notions of type

One notion of type is the semialgebraic homeomorphism class under bounded complexity. For every $C^1$7, there exist integers $C^1$8 and semialgebraic subsets $C^1$9, each of complexity at most $X\subset \mathbb{R}^m$0, such that every semialgebraic set $X\subset \mathbb{R}^m$1 of complexity at most $X\subset \mathbb{R}^m$2 is semialgebraically homeomorphic to one of the $X\subset \mathbb{R}^m$3 by a homeomorphism of bounded complexity. The analogous theorem over any real closed field is stated separately. In the algebraic case, for every $X\subset \mathbb{R}^m$4 there are finitely many model algebraic sets $X\subset \mathbb{R}^m$5 of degree at most $X\subset \mathbb{R}^m$6 such that every algebraic subset of degree at most $X\subset \mathbb{R}^m$7 is semialgebraically homeomorphic to one of them by a homeomorphism of bounded complexity. The geometric engine is Hardt’s semialgebraic trivialization theorem; effectivity is extracted via Tarski–Seidenberg, the transfer principle, and Tarski decidability (Demdah et al., 2024).

A different type theory is spectral. For a semialgebraic set $X\subset \mathbb{R}^m$8, let $X\subset \mathbb{R}^m$9 be the ring of continuous semialgebraic functions and $X\subset \mathbb{R}^m$0 its bounded subring. Then

$X\subset \mathbb{R}^m$1

whereas

$X\subset \mathbb{R}^m$2

Here $X\subset \mathbb{R}^m$3 is the finite set of points of $X\subset \mathbb{R}^m$4 having a neighborhood semialgebraically homeomorphic to $X\subset \mathbb{R}^m$5. The paper also proves that $X\subset \mathbb{R}^m$6 if and only if $X\subset \mathbb{R}^m$7 is compact, and that the maximal spectra $X\subset \mathbb{R}^m$8 and $X\subset \mathbb{R}^m$9 are always homeomorphic (Fernando et al., 2013).

These two regimes isolate different invariants. The topological-type results give finite lists of representatives and bounded-complexity homeomorphisms, while spectral-type results recover semialgebraic geometry from function rings and their maximal spectra. This suggests that “type” in the closed semialgebraic setting is inherently plural rather than canonical.

3. $X\cap D$0-triangulations and differential-topological structure

Every locally closed semialgebraic set $X\cap D$1 admits a semialgebraic triangulation $X\cap D$2 such that the realization map

$X\cap D$3

is of class $X\cap D$4. The point is stronger than smoothness on simplex interiors: each closed simplex is $X\cap D$5-realized, although the differential may drop rank on the boundary or on lower-dimensional subsets. The proof uses triangulation, semialgebraic stratification, tubes around strata, the curve selection lemma in the form of the wing lemma, and an iterative “panel beating” deformation. In the stronger technical statement, given a triangulation $X\cap D$6 and a semialgebraic $X\cap D$7 map $X\cap D$8, there exists a semialgebraic homeomorphism $X\cap D$9 of $m$0 preserving each simplex such that $m$1 is $m$2. A common $m$3-refinement theorem is also established.

For a compact semialgebraic subset $m$4 of dimension $m$5 with a fundamental class, and a differential $m$6-form $m$7 on the ambient semialgebraic manifold $m$8, this yields the straightforward formula

$m$9

The integral is independent of the chosen $D$0-triangulation, and the semialgebraic Stokes formula takes the expected form

$D$1

For compact closed semialgebraic sets, these results replace more elaborate semialgebraic-chain machinery by ordinary integration over compact simplices (Ohmoto et al., 2015).

4. Quantitative optimization and separation on basic closed sets

Quantitative arithmetic information is available for minima of polynomials on connected components of basic closed semialgebraic sets. Let $D$2 be a compact connected component of

$D$3

and let $D$4 have degree $D$5. Then the minimum of $D$6 on $D$7 is an algebraic number of degree at most $D$8. If the minimum is nonzero, its absolute value has an explicit positive lower bound of doubly exponential type in $D$9. The same conclusion holds for noncompact connected components when the set of minimizers is compact. Applying the result to

$X$0

gives an explicit lower bound on the distance between two disjoint connected components of basic closed semialgebraic sets, provided at least one component is compact (Jeronimo et al., 2011).

A constructive polynomial separation theorem addresses two compact disjoint basic semialgebraic sets

$X$1

assumed contained in $X$2. The method starts from the explicit continuous separator

$X$3

which satisfies $X$4 on $X$5 and $X$6 on $X$7, then approximates $X$8 by a polynomial via a multivariate Jackson theorem and certifies the sign conditions by an effective Putinar Positivstellensatz. The output is a polynomial $X$9 satisfying

$$T=\{x\in\mathbb R^n \mid f_1(x)=\cdots=f_\ell(x)=0,\; f_{\ell+1}(x)\ge 0,\ldots,f_m(x)\ge 0\},$$0

and the computation is organized through a hierarchy of semidefinite programs. The degree bound is polynomial in $$T=\{x\in\mathbb R^n \mid f_1(x)=\cdots=f_\ell(x)=0,\; f_{\ell+1}(x)\ge 0,\ldots,f_m(x)\ge 0\},$$1 and singly exponential in $$T=\{x\in\mathbb R^n \mid f_1(x)=\cdots=f_\ell(x)=0,\; f_{\ell+1}(x)\ge 0,\ldots,f_m(x)\ge 0\},$$2 (Korda et al., 2022).

These results are not classification theorems in the homeomorphism sense. They instead endow basic closed semialgebraic sets with explicit metric, arithmetic, and optimization-theoretic separation data.

5. Computable homology of closed semialgebraic sets

Closed semialgebraic sets defined by lax formulas admit algorithmic computation of the full homology sequence, including torsion coefficients. Given a tuple of polynomials $$T=\{x\in\mathbb R^n \mid f_1(x)=\cdots=f_\ell(x)=0,\; f_{\ell+1}(x)\ge 0,\ldots,f_m(x)\ge 0\},$$3 and a lax Boolean formula $$T=\{x\in\mathbb R^n \mid f_1(x)=\cdots=f_\ell(x)=0,\; f_{\ell+1}(x)\ge 0,\ldots,f_m(x)\ge 0\},$$4, the target set is denoted $$T=\{x\in\mathbb R^n \mid f_1(x)=\cdots=f_\ell(x)=0,\; f_{\ell+1}(x)\ge 0,\ldots,f_m(x)\ge 0\},$$5, and the main output is $$T=\{x\in\mathbb R^n \mid f_1(x)=\cdots=f_\ell(x)=0,\; f_{\ell+1}(x)\ge 0,\ldots,f_m(x)\ge 0\},$$6. The algorithm is stated to work in weak exponential time: outside a subset of input data having exponentially small measure, the cost is single exponential in the size of the data.

The method first homogenizes the affine problem to a spherical one, then controls geometry through condition numbers and ill-posedness bounds, approximates the atomic pieces by finite point clouds, builds Čech complexes, combines those complexes according to the lax formula, and finally uses Mayer–Vietoris or inclusion–exclusion transfer to show that the resulting simplicial complex has the same homology as the original semialgebraic set. The final algebraic-topology step computes $$T=\{x\in\mathbb R^n \mid f_1(x)=\cdots=f_\ell(x)=0,\; f_{\ell+1}(x)\ge 0,\ldots,f_m(x)\ge 0\},$$7 using boundary matrices and Smith normal form. The paper emphasizes that it is a computational theory of closed semialgebraic sets rather than a formal theory of “semialgebraic type sets.” A plausible implication is that homology here serves as a computable surrogate for type when full semialgebraic-homeomorphism classification is not the immediate objective (Bürgisser et al., 2018).

6. Lattice-theoretic and algebraic encodings

The lattice of closed semialgebraic sets supports a further notion of type, now in the model-theoretic sense of what the inclusion structure can interpret. In one dimension, the decisive model is the lattice of finite unions of closed intervals in a dense linear order; it is bi-interpretable with the weak monadic structure $$T=\{x\in\mathbb R^n \mid f_1(x)=\cdots=f_\ell(x)=0,\; f_{\ell+1}(x)\ge 0,\ldots,f_m(x)\ge 0\},$$8, and the resulting theory is decidable. In higher dimension the behavior changes sharply. For a definably connected open set $$T=\{x\in\mathbb R^n \mid f_1(x)=\cdots=f_\ell(x)=0,\; f_{\ell+1}(x)\ge 0,\ldots,f_m(x)\ge 0\},$$9 with $f_i\in \mathbb Z[x_1,\ldots,x_n]$0, Theorem 6.2 shows that a lattice $f_i\in \mathbb Z[x_1,\ldots,x_n]$1 of definable subsets of $f_i\in \mathbb Z[x_1,\ldots,x_n]$2 that are closed in $f_i\in \mathbb Z[x_1,\ldots,x_n]$3 can define the relation

$f_i\in \mathbb Z[x_1,\ldots,x_n]$4

and, after naming a suitable interval-like element, the poset $f_i\in \mathbb Z[x_1,\ldots,x_n]$5 interprets $f_i\in \mathbb Z[x_1,\ldots,x_n]$6. In particular, higher-dimensional lattices of closed semialgebraic sets have undecidable first-order theory (Tressl, 2016).

An adjacent algebraic direction, formulated for open semi-algebraic sets rather than closed ones, characterizes rational functions by positivity and valuation data. For

$f_i\in \mathbb Z[x_1,\ldots,x_n]$7

the Ganzstellensatz states that $f_i\in \mathbb Z[x_1,\ldots,x_n]$8 admits only integral values on $f_i\in \mathbb Z[x_1,\ldots,x_n]$9 if and only if $C^1$00 lies in the integral radical of the algebra generated by

$C^1$01

and analogous boundedness criteria hold over arbitrary real closed fields (Lavi, 2011). Although this framework is not about closed semialgebraic sets, it indicates another way in which semialgebraic “type” can be encoded algebraically: through positive cones, valuation rings, and integral closure rather than through homeomorphism classes or homology.

Taken together, these results show that closed semialgebraic sets admit several non-equivalent but highly structured type theories. Topological type yields effective finite classification under bounded complexity; spectral type reconstructs the set from semialgebraic function rings up to a finite exceptional subset in the bounded case; $C^1$02-triangulations supply differential-topological models for integration; computational homology furnishes effective invariants; and lattice-theoretic analysis reveals a sharp logical dichotomy between one-dimensional and higher-dimensional closed-set geometries.