Remarkable Cardinal¶

A virtual large cardinal κ for which arbitrarily high ground-model rank or hereditary-size segments admit, in set-forcing extensions, elementary small embeddings whose critical point is mapped to κ.

Version: v1 · 2026-08-30 · History
Domain-specific #: 2654
Origin domain: set theory
Subdomain: virtual large cardinal theory
Aliases: Virtually supercompact cardinal

Core Idea¶

A remarkable cardinal is a large cardinal whose supercompact-like elementary embeddings need not exist in the ground universe but do exist in its set-generic multiverse. In the modern rank formulation, a cardinal \(\kappa\) is remarkable when, for every \(\lambda>\kappa\), there is some \(\bar\lambda<\kappa\) such that a set-forcing extension contains an elementary embedding

\[ j:V_{\bar\lambda}^{V}\longrightarrow V_{\lambda}^{V} \quad\text{with}\quad j(\operatorname{crit}(j))=\kappa. \]

The superscript \(V\) is load-bearing: both rank segments are sets from the ground model. The forcing extension may add the embedding \(j\), but it does not replace the source and target with arbitrary structures created by forcing. Gitman and Schindler use this as the first example of a virtual large cardinal and identify remarkable cardinals as virtually supercompact.^[1]

An equivalent hereditary-size formulation says that for every regular \(\lambda>\kappa\), some set-forcing extension has a \(V\)-regular \(\bar\lambda<\kappa\) and an elementary embedding

\[ j:H_{\bar\lambda}^{V}\longrightarrow H_{\lambda}^{V} \]

with critical point \(\gamma\) and \(j(\gamma)=\kappa\).^[2] The critical point is therefore below \(\kappa\); the embedding maps it to \(\kappa\). This “small embedding” shape differs from the familiar large-embedding form whose critical point itself is the named large cardinal.

Schindler introduced remarkability to calibrate the consistency strength of forcing-absoluteness phenomena for \(L(\mathbb R)\). His original paper shows that the notion is not merely a decorative variant of supercompactness: it is a reusable bridge among elementary embeddings, proper forcing, generic absoluteness, and lower regions of the large-cardinal hierarchy.^[3]

Structural Signature¶

The property has these mandatory roles:

candidate cardinal \(\kappa\) — the image singled out by every witnessing embedding;
arbitrary target height \(\lambda>\kappa\) — quantification over all sufficiently high rank segments, or all regular hereditary-size targets in the \(H_\lambda\) version;
smaller source height \(\bar\lambda<\kappa\) — a ground-model rank or hereditary-size segment below the candidate;
ground-model source and target — \(V_{\bar\lambda}^{V},V_{\lambda}^{V}\) or \(H_{\bar\lambda}^{V},H_{\lambda}^{V}\);
set-forcing extension — an outer model in which the otherwise absent embedding exists;
elementary embedding \(j\) — preservation of every first-order formula between source and target;
critical point \(\gamma\) — the least ordinal moved by \(j\);
image equation \(j(\gamma)=\kappa\) — the defining placement of the candidate cardinal;
uniform recurrence — a witness exists for every required target height, not just one convenient \(\lambda\).

The locked recognition form is

\[ \forall \lambda>\kappa\;\exists\bar\lambda<\kappa\;\exists\mathbb P\; \Vdash_{\mathbb P} \exists j\bigl(j:V_{\bar\lambda}^{V}\prec V_{\lambda}^{V} \wedge j(\operatorname{crit}(j))=\kappa\bigr). \]

This notation abbreviates “some set-forcing extension”; it does not assert one fixed forcing works uniformly unless an equivalent characterization specifically supplies one. The \(H_\lambda\) form can be witnessed after collapsing below \(\kappa\), using absoluteness of embeddings between countable structures.^[2]^[1]

What It Is Not¶

Remarkability is not cardinality in general. Cardinality classifies set size by bijections. Remarkability adds a specific large-cardinal property involving arbitrary target heights, ground-model segments, forcing extensions, elementarity, a critical point, and an image equation.

It is not supercompactness. Magidor's small-embedding characterization of a supercompact \(\kappa\) has embeddings of the same broad form already in \(V\). Remarkability virtualizes that characterization by requiring the embeddings only in set-forcing extensions.^[2]

It is not a generic large cardinal in the older whole-universe sense. Generic large-cardinal embeddings commonly have the form \(j:V\to M\) in an extension, with \(M\) an inner model of that extension. Virtual large cardinals instead use set-sized source and target structures from \(V\); only the embedding is generic.^[1]

It is not weak remarkability. Wilson studies a weakening that can fail the relevant \(\Sigma_2\)-reflection condition. He proves that the \(\Sigma_2\)-reflecting weakly remarkable cardinals are exactly the remarkable cardinals, while a non-reflecting weakly remarkable cardinal has a different, higher equiconsistency profile.^[4]

It is not the existence of one accidental elementary embedding. The definition quantifies over every target height above \(\kappa\). Nor is any embedding between countable transitive structures sufficient: source, target, critical point, image, regularity where required, and ground-model provenance must all match.

It is not a proof in ZFC that a remarkable cardinal exists. As with large-cardinal axioms generally, the entry states a conditional property and its consequences. Equiconsistency results compare theories; they do not establish absolute consistency.

Scope of Application¶

Remarkable cardinals are used in set theory where forcing extensions expose elementary embeddings between fixed ground-model structures. Their primary setting is virtual large-cardinal theory, but their applications reach forcing absoluteness, structural reflection, forcing axioms, preservation and indestructibility, and descriptive-set-theoretic models involving \(L(\mathbb R)\).

The notion originated in Schindler's analysis of how proper forcing can change the theory of \(L(\mathbb R)\). The existence of a remarkable cardinal is equiconsistent with the assertion that the theory of \(L(\mathbb R)\) cannot be changed by proper forcing.^[3]^[1] “Equiconsistent” means each theory proves the consistency of the other relative to the usual background transformations; it is not an identity of statements.

Gitman and Schindler place remarkability inside the virtual hierarchy. Unlike generic whole-universe notions that can be borne by a small cardinal in an extension, their virtual cardinals are actual large cardinals. The virtual approach is compatible with \(V=L\), because the source and target are ground-model sets and the witnessing embedding can exist only after forcing.^[1]

Later work uses remarkable embeddings to define Laver-like anticipation functions and prepare indestructibility. Cheng and Gitman prove that every remarkable cardinal carries a remarkable Laver function and that remarkability can be made indestructible under specified forcing classes; they also show preservation by canonical forcing for GCH.^[5] These are applications and refinements, not extra clauses in the base definition.

The scope excludes ordinary cardinal arithmetic, finite cardinals, and metaphorical uses of “remarkable.” It also excludes every virtual large cardinal notion other than virtual supercompactness unless equivalence is proved.

Clarity¶

A reliable verification sequence is:

Fix a cardinal \(\kappa\).
Quantify over every \(\lambda>\kappa\), or every regular \(\lambda>\kappa\) in the \(H_\lambda\) presentation.
Find a source height \(\bar\lambda<\kappa\), with the required ground-model regularity in the \(H_\lambda\) form.
Identify a set-forcing extension containing an elementary embedding between the indicated ground-model segments.
Let \(\gamma=\operatorname{crit}(j)\), verify \(\gamma<\bar\lambda\), and check \(j(\gamma)=\kappa\).
Repeat the condition uniformly for arbitrary target heights.

Three common errors are immediately diagnostic. Writing \(\operatorname{crit}(j)=\kappa\) changes the small-embedding definition. Omitting the superscript \(V\) can blur virtuality by allowing the structures themselves to come from the extension. Replacing “for every \(\lambda\)” with “for some \(\lambda\)” destroys the large-cardinal recurrence.

The \(H_\lambda\) and \(V_\lambda\) forms are equivalent presentations with slightly different quantifiers. A draft should state one precisely and explain the other rather than splice their regularity conditions into an unverified hybrid.

Manages Complexity¶

Remarkability compresses a web of forcing and embedding data into one stable property. Without the concept, each absoluteness or preservation argument would separately track collapses, countable structures, source and target segments, critical points, and elementarity. The remarkable-cardinal package says that supercompact-like small embeddings recur throughout the generic multiverse at every height.

The virtual viewpoint also separates two questions that are easily conflated:

which set-sized structures already belong to the ground universe; and
in which outer universe an elementary map between them exists.

That separation permits strong embedding reasoning at a consistency level far below actual supercompactness. It explains why forcing can reveal a map without retroactively making the ground model contain it and why the target's closure properties cannot simply be imported from the actual supercompact setting.

The countable-embedding absoluteness lemma supplies technical compression. Once a source structure is countable in a suitable transitive model and a target is present, an elementary embedding existing somewhere can be reproduced with prescribed finite behavior, including the critical point. This underwrites equivalent collapse formulations and makes the generic location of the witness less arbitrary.^[1]

Abstract Reasoning¶

Several inferences follow directly from the quantifier pattern.

One witness cannot establish remarkability. If an embedding exists only for a fixed \(\lambda\), it witnesses a local virtual reflection phenomenon. Remarkability requires unbounded recurrence across every higher target.

The critical point is not \(\kappa\). If \(\gamma=\operatorname{crit}(j)\) and \(j(\gamma)=\kappa\), then \(\gamma<\kappa\). The candidate is the image of a smaller critical point. This is characteristic of the small-embedding formulation.

Virtual existence is weaker than actual existence. If the required embeddings already exist in \(V\), the stronger small-embedding property associated with supercompactness is present. An embedding found only in \(V[G]\) does not imply that \(V\) contains it.

Ground-model provenance limits the weakening. Allowing the source and target to be newly created in the extension would no longer express the same virtual large-cardinal template. The property compares fixed structures under an expanded supply of maps.

Collapse formulations are not mere conveniences. Countability permits tree and well-foundedness absoluteness arguments to relocate embeddings while preserving the critical point. This makes a canonical \(\operatorname{Coll}(\omega,<\kappa)\) presentation available in the hereditary-size formulation.^[2]

Consistency comparisons are not implication chains. The equiconsistency with proper-forcing absoluteness for \(L(\mathbb R)\) measures proof-theoretic strength. It does not say every universe with one statement literally contains the same objects as every universe with the other.

Knowledge Transfer¶

Within virtual large-cardinal theory, the method transfers by starting with a large-cardinal notion characterized through set-sized elementary embeddings and asking for those maps in forcing extensions while retaining ground-model source and target structures. Remarkability is the virtual form of supercompactness under the relevant small-embedding characterization.^[1]

Within forcing theory, the collapse-and-absoluteness apparatus transfers to preservation, Laver functions, indestructibility, and generic structural reflection. Analysts retain the same roles—ground-model structures, outer-model map, critical point, image target, and height quantification—while altering the forcing class or additional properties.

Transfer has limits. Virtualizing a different large-cardinal notion can produce a distinct hierarchy, and familiar equivalences can split when Kunen-style arguments no longer apply to virtual embeddings. Gitman and collaborators explicitly warn that conditions redundant in actual large-cardinal settings can yield non-equivalent virtual notions.^[6]

There is no literal transfer outside set theory. A “remarkable” large number, a surprising database cardinality, or an embedding in machine learning does not instantiate this property. Removing cumulative-hierarchy levels, forcing extensions, elementary embeddings, critical points, and cardinal quantifiers leaves only a metaphor.

Examples¶

A correct local witness pattern. Fix regular \(\lambda>\kappa\). In some forcing extension, let \(j:H_{\bar\lambda}^{V}\to H_\lambda^{V}\) be elementary, with \(\bar\lambda<\kappa\) regular in \(V\), \(\gamma=\operatorname{crit}(j)\), and \(j(\gamma)=\kappa\). This is one required witness. It establishes remarkability only when such a pattern exists for every regular target \(\lambda\).

The collapse presentation. In a \(\operatorname{Coll}(\omega,<\kappa)\)-generic extension, each relevant smaller \(H_{\bar\lambda}^{V}\) becomes countable. The absoluteness lemma can reproduce the elementary map with the same critical point behavior, yielding a uniform environment for studying remarkable embeddings.^[2]

Comparison with supercompactness. Magidor's characterization supplies small embeddings in the universe itself for a supercompact cardinal. Moving only the embedding into a forcing extension while keeping both segments from \(V\) gives the remarkable/virtually supercompact pattern. This comparison explains the name's position without equating the strengths.

A failure by wrong critical point. Suppose a forcing extension has \(j:H^V_{\bar\lambda}\to H^V_\lambda\) with \(\operatorname{crit}(j)=\kappa\). Since \(\bar\lambda<\kappa\) in the remarkable pattern, this is impossible as written; \(\kappa\) is outside the source height. The intended condition is \(j(\gamma)=\kappa\) for a smaller \(\gamma\).

A failure by one target. A virtual elementary embedding at one \(\lambda\) may be mathematically interesting, but it does not satisfy the universal target-height quantifier.

Forcing-absoluteness calibration. Schindler's equiconsistency result uses the existence of a remarkable cardinal as the exact large-cardinal strength for an assertion that proper forcing does not change the theory of \(L(\mathbb R)\). This is a principal application rather than a redefinition.^[3]

Remarkable Laver function. Cheng and Gitman construct a partial \(\ell:\kappa\to V_\kappa\) whose values can be anticipated through suitable remarkable embeddings, enabling preparation and indestructibility arguments.^[5]

Structural Tensions¶

Ground model versus forcing extension. The structures must remain from \(V\), while the map may appear only in \(V[G]\). Relaxing the first side changes the notion; requiring the second side to collapse back into \(V\) approaches a stronger actual property.

Small source versus high target. Every source height lies below \(\kappa\), yet targets range arbitrarily above it. The embedding's elementarity transports a smaller critical point to the candidate cardinal and reflects high structure through a small domain.

Virtual strength versus actual closure. The pattern resembles supercompactness, but the generic target need not have all closure behavior used in actual supercompact arguments. Proofs must not silently import it.

Forcing flexibility versus identity stability. Witnessing extensions may vary with \(\lambda\), but the ground-model segments and image equation stabilize the property. Allowing arbitrary new source and target models would make it too permissive.

Hierarchy label versus consistency theorem. “Virtually supercompact” locates the definition structurally; it does not assign supercompact consistency strength. Results about weakly remarkable, \(\omega\)-Erdős, or \(0^\#\) boundaries require separate theorems and exact hypotheses.

Existence axiom versus conditional mathematics. Rich consequences follow from assuming a remarkable cardinal, while ZFC does not supply a known proof of its existence or consistency. Reference prose must preserve that conditional stance.

Structural–Framed Character¶

Remarkable Cardinal is highly structural. Its recognition depends on an exact quantifier prefix, cumulative-hierarchy or hereditary-size segments, elementarity, forcing location, critical point, and image equation. Small changes create different large-cardinal notions or invalid statements.

It is nonetheless wholly framed by set theory. “Ground model,” “set-forcing extension,” \(V_\alpha\), \(H_\lambda\), elementary embedding, regular cardinal, and critical point are constitutive. The structure recurs across proofs and applications inside forcing and large-cardinal theory, not across unrelated material substrates.

The node is therefore domain-specific rather than prime-like. Its autonomy comes from a named property with equivalent definitions, a clear recognition test, a research literature, and distinct applications—not from broad substrate independence.

Structural Core vs. Domain Accent¶

The apparent portable core is: fixed structures may acquire a structure-preserving map in an extended environment. That skeleton is too weak to recover remarkability. It could describe algebraic extensions, model completions, or software migration.

The domain accent supplies nearly all discriminating content: rank-initial or hereditary-size ground-model segments; a candidate cardinal; arbitrary target heights; set forcing; first-order elementarity; the least moved ordinal; and the equation mapping that critical point to \(\kappa\). Those features make the property recognizable.

Removing the domain yields generic extension-and-mapping ideas already covered by broader abstractions. Preserving the identity requires set-theoretic vocabulary and axioms. This is the exact pattern of a strong domain-specific abstraction with no prime promotion case.

Remarkable Cardinal is strictly a kind of Cardinality/Cardinal Number in the catalog's broad sense: \(\kappa\) is a transfinite cardinal carrying an additional property. This is the proposed minimal DAG parent.

It is related to Infinity because every remarkable cardinal is an infinite large cardinal, but Infinity does not supply the embedding criterion. It is related to elementary embedding, forcing, reflection, and possible-world reasoning; those relations belong in prose unless separately supported by exact catalog nodes and nonredundant edges.

Cardinality does not compositionally close the candidate. Bijection-based size, even combined with Infinity and forcing, does not entail the universal small-embedding scheme or image equation.

Relationships to Other Abstractions¶

Current abstraction Remarkable Cardinal Domain-specific

Parents (1) — more general patterns this builds on

Remarkable Cardinal is a kind of Cardinality Prime

Remarkable Cardinal is strictly a kind of Cardinality/Cardinal Number in the catalog's broad sense: \(\kappa\) is a transfinite cardinal carrying an additional property.

Hierarchy paths (5) — routes to 3 parentless roots

Remarkable Cardinal → Cardinality → Bijectivity → Function (Mapping)

Show alternative paths (4)

Neighborhood in Abstraction Space¶

Remarkable Cardinal sits in a sparse region of the domain-specific corpus (87^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Ordered Models & Definability Properties (5 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-09-08

Not to Be Confused With¶

Remarkable cardinal: the accepted specialist name introduced by Schindler.
Virtually supercompact cardinal: modern equivalent name under the set-sized embedding template.
Supercompact cardinal: requires the relevant embedding strength in the universe, not only generic set-sized witnesses.
Generic supercompactness: commonly uses an embedding of \(V\) into an inner model of an extension and can behave differently from virtual supercompactness.
Weakly remarkable cardinal: a weakened property; only its \(\Sigma_2\)-reflecting instances coincide with remarkable cardinals.^[4]
Remarkable Laver function: a function associated with a remarkable cardinal, not the cardinal property itself.
Critical point \(\kappa\): wrong for the small-embedding definition; a smaller \(\gamma\) is critical and maps to \(\kappa\).
One virtual embedding: insufficient without witnesses for all required targets.
Large cardinal: the genus, encompassing many inequivalent properties.
Remarkable number: ordinary-language praise or recreational terminology has no relation to this set-theoretic notion.

References¶

[1] Victoria Gitman and Ralf Schindler, “Virtual Large Cardinals,” Annals of Pure and Applied Logic 169, no. 12 (2018): 1317–1334; author-hosted final manuscript. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g

[2] Victoria Gitman, “Remarkable Laver Functions,”, CUNY Set Theory Seminar exposition, 2015. It states the hereditary-size embedding characterization and collapse/absoluteness formulation. registry ↩a ↩b ↩c ↩d ↩e

[3] Ralf-Dieter Schindler, “Proper Forcing and Remarkable Cardinals,” Bulletin of Symbolic Logic 6, no. 2 (2000): 176–184. registry ↩a ↩b ↩c

[4] Trevor M. Wilson, “Weakly Remarkable Cardinals, Erdős Cardinals, and the Generic Vopěnka Principle,” Journal of Symbolic Logic 84, no. 4 (2019): 1711–1721. registry ↩a ↩b

[5] Yong Cheng and Victoria Gitman, “Indestructibility Properties of Remarkable Cardinals,” Archive for Mathematical Logic 54 (2015): 961–984; preprint. registry ↩a ↩b

[6] Stamatis Dimopoulos, Victoria Gitman, and Dan Saattrup Nielsen, “The Virtual Large Cardinal Hierarchy,”, Fundamenta Mathematicae 2024, author publication page and manuscript. registry ↩