> Full-text translation from Chinese. The original remains authoritative for its research claims; this translation is not independent mathematical verification.

# NIOT-01 — Numerical Identity Units and the Projection Non-Identity Principle

## Numerical Identity Units and the Projection Non-Identity Principle

**Series:** Numerical Information Object Theory (NIOT) / Numerical Information Object Theory  
**Paper:** Paper 01  
**Author:** Neo.K  
**Institution:** EveMissLab / 一言諾科技有限公司  
**Version:** v0.1  
**Date:** 2026-09-22  
**Document type:** Foundational ontology / definition paper / formal opening paper of the series  
**Prerequisite:** NIOT-00, Numerical Information Object Theory: From Linear Floating-Point Representation to Unbounded Probe Matrices and Microscopic Numerical Worlds  
**Related existing theories:** ICNS, MNVP, CAIR, ISQL, Identity Calculus, Active Index Calculus, completion–projection bidirectional semantics  
**Status:** DRAFT / FOUNDATIONAL ONTOLOGY / FORMALIZATION-READY  
**Canonical Source:** UTF-8 Markdown  
**Mathematical source conventions:** Inline math uses only `$...$`, and display math uses only `$$...$$`

---

# Abstract

In traditional mathematics, a numerical value is usually identified by the set to which it belongs and the value itself, for example $x\in\mathbb R$. Traditional programming and numerical computation further implement numerical values as finite representations, such as floating-point formats, integer bit patterns, strings, arbitrary-precision decimals, or symbolic representations. These methods are already powerful enough for standard mathematics and engineering computation. But when numerical values are placed within AI, semantic runtimes, data-source tracing, cross-representation conversion, dynamic indexing, certified computation, and multi-projection systems, an important question emerges: **Is the scalar currently read out equal to the entire numerical computational object?**

This paper proposes the first set of ontological axioms for Numerical Information Object Theory, centrally distinguishing three different levels:

$$
\boxed{
\text{abstract value}
\neq
\text{represented value}
\neq
\text{numerical information object}.
}
$$

This paper uses $\mathfrak N$ for a numerical information object and $\pi_v$ for its scalar-value projection. For any actual numerical object:

$$
\pi_v(\mathfrak N)=x
$$

means only that $x$ is an admissible value projection of $\mathfrak N$, and does not automatically imply:

$$
\mathfrak N=x.
$$

This principle is called the **Projection Non-Identity Principle**.

Further, this paper proposes a rigorous version of "every number is one." Here "one" is not the numerical value $1$, nor a claim that all numbers are equal to each other; it means that every identified numerical information object whose provenance and operational continuity can be preserved can be treated as an **identity unit**:

$$
\boxed{
\mathbf 1_{\mathfrak N}
}
$$

Here $\mathbf 1_{\mathfrak N}$ denotes the identity of "this object as a complete unit of study," rather than its scalar value. Indexing, slicing, probes, representation conversions, and local projections can change the information we observe without necessarily changing the numerical object's canonical identity under admissible operations.

This paper further introduces the value-equivalence fiber. If two different numerical information objects have the same scalar projection:

$$
\pi_v(\mathfrak N_1)
=
\pi_v(\mathfrak N_2)
=
x,
$$

they are equivalent only under value projection, and need not be identical in provenance, history, semantics, precision, certificates, indices, relations, or representation structure. Thus scalar projection may naturally compress a collection of richer information structures to the same value:

$$
\boxed{
\pi_v^{-1}(x)
=
\{
\mathfrak N:\pi_v(\mathfrak N)=x
\}.
}
$$

This paper treats this fiber as the first mathematical entry point into "numerical informational distinctions."

Ultimately, this paper establishes the ontological foundation of NIOT: a numerical information object does not negate traditional mathematical numerical values; rather, it adds computable structures such as identity, representation, provenance, index, relation, history, certificate, and bridge beyond the existing value layer. Classical numerical values remain in their original mathematical domains. The new theory addresses **how the same value becomes an information object that can be probed, sliced, addressed, projected, and verified in the actual computational world**.

**Keywords:** Numerical Identity, Projection Non-Identity, Numerical Information Object, Identity Unit, value fiber, Informational Distinction, canonical identity, scalar projection, Identity Calculus, ISQL, CAIR, ICNS

---

# 1. Making the Problem Precise

## 1.1 At Which Layer Is a Number an Object?

Consider:

$$
x=0.73125.
$$

This expression has at least four different readings:

1. $x$ is an abstract real number;
2. $x$ is some finite decimal representation;
3. $x$ is the displayed result of a machine floating-point value;
4. $x$ is the scalar projection of a more complete computational state.

In traditional elementary computation, these four readings can often be safely compressed together. In the following settings, however, they can no longer be assumed identical:

- Interconversion of exact real numbers and floating-point numbers;
- Different sources with identical displayed values;
- The same value with different units;
- The same value with different precision;
- The same value with different uncertainty;
- The same value at different temporal states;
- The same value with different version semantics;
- The same value from different generation paths;
- The same value with different certificates;
- The same value with different indices / addresses;
- The same value in different canonical objects.

This paper therefore begins not from:

$$
x
$$

itself, but from:

$$
\boxed{
\mathfrak N
}
$$

as displayed above.

---

## 1.2 Three-Layer Distinction

This paper first establishes a minimal three-layer structure:

### First Layer: Abstract Value

$$
x\in X,
$$

where $X$ may be:

$$
\mathbb N,\mathbb Z,\mathbb Q,\mathbb R,\mathbb C
$$

or another mathematical carrier domain.

This layer addresses addition, multiplication, order, topology, limits, algebraic structures, measure, geometry, and other standard mathematical properties. This paper does not replace this layer.

### Second Layer: Representation

Let:

$$
r\in\mathcal R
$$

be a representation, such as a decimal string, binary string, float, arbitrary precision, rational pair, symbolic term, interval, stream, tensor cell, or database field.

There is a semantic map:

$$
\llbracket-\rrbracket:
\mathcal R
\rightarrow
X.
$$

### Third Layer: Numerical Information Objects

Define:

$$
\boxed{
\mathfrak N
\in
\mathfrak X
}
$$

where $\mathfrak X$ is the numerical information object space.

$\mathfrak N$ carries not only a value but also structures that allow it to be identified, manipulated, and verified in actual systems.

Therefore:

$$
\boxed{
X
\neq
\mathcal R
\neq
\mathfrak X.
}
$$

---

# 2. Minimal Ontology of Numerical Information Objects

## 2.1 Initial Structure

Define a numerical information object:

$$
\boxed{
\mathfrak N
=
\left\langle
\iota,
v,
\sigma,
a,
r,
h,
p,
q,
c,
b
\right\rangle.
}
$$

where:

- $\iota$: canonical identity;
- $v$: value state;
- $\sigma$: semantic / representation schema;
- $a$: address / index state;
- $r$: relation state;
- $h$: history / derivation;
- $p$: provenance;
- $q$: quality / precision / uncertainty;
- $c$: context / constraint;
- $b$: bridge information.

This does not claim that every implementation must permanently store ten fields; rather, the complete theoretical space of a numerical information object at least permits these informational directions to exist.

---

## 2.2 Core and Envelope

To avoid making theoretical objects excessively large, this paper distinguishes:

$$
\boxed{
\mathfrak N
=
\operatorname{Core}(\mathfrak N)
+
\operatorname{Envelope}(\mathfrak N).
}
$$

The plus sign here is not numerical addition, but structural composition.

The Canonical Core requires at least:

$$
\boxed{
\operatorname{Core}(\mathfrak N)
=
\langle
\iota,
v,
\sigma
\rangle.
}
$$

The Envelope may dynamically contain:

$$
\boxed{
\operatorname{Envelope}(\mathfrak N)
=
\langle
a,r,h,p,q,c,b,\ldots
\rangle.
}
$$

The Core ensures that object identity and minimal semantics can be recovered; the Envelope permits information to be progressively probed and expanded.

---

# 3. Projection Non-Identity Principle

## 3.1 Definition

Let:

$$
\pi:
\mathfrak X
\rightarrow
Y
$$

be any admissible projection.

Then, in general:

$$
\boxed{
\pi(\mathfrak N)
\neq
\mathfrak N.
}
$$

Only when $\pi$ has been proved to be an information-preserving isomorphism on the specified information domain can the projection and source object be treated as identical in a restricted sense.

---

## 3.2 Scalar projection

Let:

$$
\pi_v:
\mathfrak X
\rightarrow
X
$$

be the scalar-value projection.

then:

$$
\boxed{
\pi_v(\mathfrak N)=x
}
$$

means only that $\mathfrak N$ is projected to $x$ at the value layer.

It does not mean:

$$
\boxed{
\mathfrak N=x.
}
$$

This is the first central axiom of NIOT.

---

## 3.3 Why Is This Not a Language Game?

Suppose:

$$
\mathfrak N_1
=
\langle
\iota_1,
1.00,
\sigma_1,
p_1,
q_1,\ldots
\rangle
$$

and:

$$
\mathfrak N_2
=
\langle
\iota_2,
1.00,
\sigma_2,
p_2,
q_2,\ldots
\rangle.
$$

Even if:

$$
\pi_v(\mathfrak N_1)
=
\pi_v(\mathfrak N_2)
=
1.00,
$$

one may still have:

$$
\iota_1\neq\iota_2,
$$

$$
p_1\neq p_2,
$$

$$
q_1\neq q_2.
$$

For example, one may be a precision measurement and the other a model estimate; one has precision $0.01$ and the other's uncertainty is unknown; one is projected from an exact rational and the other obtained by float rounding.

If only:

$$
1.00,
$$

is retained, all these differences disappear.

Therefore:

$$
\boxed{
\text{same scalar}
\not\Rightarrow
\text{same numerical information object}.
}
$$

---

# 4. Value Equivalence and Information Fibers

## 4.1 Value-equivalence

Define the relation:

$$
\mathfrak N_1
\sim_v
\mathfrak N_2
$$

if and only if:

$$
\boxed{
\pi_v(\mathfrak N_1)
=
\pi_v(\mathfrak N_2).
}
$$

This relation means only that the value projections are equal.

It does not mean:

$$
\mathfrak N_1
=
\mathfrak N_2.
$$

---

## 4.2 Scalar fiber

For any:

$$
x\in X,
$$

Define:

$$
\boxed{
F_x
=
\pi_v^{-1}(x)
=
\{
\mathfrak N\in\mathfrak X
:
\pi_v(\mathfrak N)=x
\}.
}
$$

$F_x$ is the set of all numerical information objects that project to the same scalar $x$.

This is the first formal entry point into "informational distinctions" in this paper.

---

## 4.3 Fiber multiplicity theorem

If:

$$
|F_x|>1,
$$

the scalar projection:

$$
\pi_v
$$

is not injective on $F_x$.

Thus there exist:

$$
\mathfrak N_1\neq\mathfrak N_2
$$

such that:

$$
\pi_v(\mathfrak N_1)
=
\pi_v(\mathfrak N_2).
$$

Thus:

$$
\boxed{
\text{scalar equality is insufficient for object identity}.
}
$$

This is not an additional assumption, but a direct consequence of a non-injective projection.

---

## 4.4 Minimal Definition of Informational Distinction

For:

$$
\mathfrak N_1,\mathfrak N_2\in F_x,
$$

if there is an admissible probe or structural function:

$$
P
$$

such that:

$$
P(\mathfrak N_1)
\neq
P(\mathfrak N_2),
$$

the two are said to have:

$$
\boxed{
\text{informational distinction under }P.
}
$$

Thus an "informational distinction" is not a mysterious new physical quantity; it is:

$$
\boxed{
\text{a structural difference erased by scalar projection but still distinguishable by other admissible observations.}
}
$$

---

# 5. A Rigorous Version of "Every Number Is One"

## 5.1 The Most Easily Misunderstood Statement

This paper uses:

> Every number is one.

But one must immediately exclude:

$$
1=2=3
$$

this erroneous interpretation.

The "one" in this paper is not:

$$
1\in\mathbb N.
$$

but:

$$
\boxed{
\text{one identity-bearing unit}.
}
$$

---

## 5.2 Identity Unit

For any numerical information object with a canonical identity:

$$
\mathfrak N,
$$

define its identity unit:

$$
\boxed{
\mathbf 1_{\mathfrak N}.
}
$$

$\mathbf 1_{\mathfrak N}$ means:

> In the present theory, $\mathfrak N$ is treated as a complete unit of study and operation whose identity can be preserved.

Therefore:

$$
\mathbf 1_{\mathfrak N_1}
\neq
\mathbf 1_{\mathfrak N_2}
$$

is not contradictory.

Both are "one unit," but not the same unit.

---

## 5.3 Identity Unit Axiom

This paper provisionally proposes:

$$
\boxed{
\forall \mathfrak N\in\mathfrak X,
\quad
\exists
\mathbf 1_{\mathfrak N}
}
$$

such that every admissible indexing, projection, and slicing operation should, unless explicitly declared an identity-changing operation, preserve:

$$
\boxed{
\iota(\mathbf 1_{\mathfrak N})
=
\iota(\mathfrak N).
}
$$

---

## 5.4 Separating "One" from "Value"

If:

$$
\pi_v(\mathfrak N)=7,
$$

then:

$$
\mathbf 1_{\mathfrak N}
$$

is not:

$$
1
$$

nor:

$$
7.
$$

It is:

$$
\boxed{
\text{the identity unit whose scalar projection happens to be }7.
}
$$

Thus a fuller formulation of "every number is one" is:

$$
\boxed{
\text{Every identified numerical information object can be treated as an identity unit.}
}
$$

---

# 6. Identity Calculus Enters the Numerical Domain

## 6.1 Slicing Is Not Splitting

The central principle of Identity Calculus is:

$$
\boxed{
\text{slicing}
\neq
\text{ontological destruction}.
}
$$

Define indexed slices of a numerical information object:

$$
\boxed{
d_I(\mathfrak N)
=
\{
(\mathfrak N,i)
:
i\in I
\}.
}
$$

These are not new independent numerical entities, but different indexed views of the same canonical object.

---

## 6.2 Identity preservation

For all admissible indices:

$$
i\in I,
$$

require:

$$
\boxed{
\iota(\mathfrak N,i)
=
\iota(\mathfrak N).
}
$$

This is called:

$$
\boxed{
\text{indexed identity preservation}.
}
$$

---

## 6.3 Forgetful integration

Define forgetting the index:

$$
\boxed{
\int_{\mathrm{id}}
(\mathfrak N,i)
=
\mathfrak N.
}
$$

then:

$$
\boxed{
\int_{\mathrm{id}}
\circ
d_I
=
\operatorname{id}_{\mathfrak N}
}
$$

This is not a general measure integral in this paper, but an identity-preserving retraction.

---

# 7. Active Indexing and Numerical Addressing

## 7.1 A Passive Index Is Not Enough

If:

$$
i
$$

is only a metadata label, it can only indicate "what this fragment is called."

Active Index Calculus instead permits:

$$
\boxed{
\widehat{\mathcal E}_{i\rightarrow j}
:
(\mathfrak N,i)
\mapsto
(\mathfrak N,j).
}
$$

Thus an index can become navigation, routing, computation, selective expansion, and local recombination.

---

## 7.2 An Address Is Not a Value

Let:

$$
a_i
$$

be an information address.

then:

$$
a_i(\mathfrak N)
$$

may point to a precision state, provenance node, history segment, spectral coordinate, relation cluster, certificate, or bridge edge.

Therefore:

$$
\boxed{
\text{address space}
\neq
\text{value space}.
}
$$

This is one necessary condition for the establishment of microscopic numerical worlds.

---

# 8. Projection family

## 8.1 An Object May Have Multiple Admissible Projections

Let:

$$
\Pi
=
\{
\pi_v,
\pi_r,
\pi_p,
\pi_q,
\pi_h,
\pi_a,
\ldots
\}.
$$

representing value projection, representation projection, provenance projection, quality projection, history projection, and address projection, respectively.

For the same:

$$
\mathfrak N,
$$

different projections may produce:

$$
\pi_i(\mathfrak N)
\neq
\pi_j(\mathfrak N)
$$

while still sharing:

$$
\iota(\mathfrak N).
$$

---

## 8.2 CAIR-Type Structure

This yields the same basic form as CAIR:

$$
\boxed{
\text{one canonical object}
\rightarrow
\text{many projections}.
}
$$

Numerical Information Object Theory brings this structure into the numerical domain.

Thus decimal, binary, float, interval, graph, and matrix may all be projections or bridge results of a canonical numerical object, rather than necessarily competing "true representations."

---

# 9. Canonical Identity and Representation Identity

## 9.1 Two Kinds of Identity Must Be Distinguished

Define:

$$
\iota_C
$$

as canonical identity,

$$
\iota_R
$$

as representation identity.

It may be that:

$$
\iota_C(\mathfrak N_1)
=
\iota_C(\mathfrak N_2)
$$

However:

$$
\iota_R(\mathfrak N_1)
\neq
\iota_R(\mathfrak N_2).
$$

For example, the same exact rational:

$$
\frac12
$$

can be represented as:

$$
0.5,
$$

$$
\frac12,
$$

$$
0.1000_2.
$$

If a bridge is proved semantics-preserving, they may share the same canonical value identity while having different representation identities.

---

## 9.2 The Reverse May Also Hold

Two representations appear identical:

$$
\texttt{"1.0"}
$$

but when their schemas differ:

$$
\sigma_1\neq\sigma_2,
$$

they may represent a real number, version field, interval endpoint, scale factor, or categorical code.

Therefore:

$$
\boxed{
\text{surface token identity}
\not\Rightarrow
\text{semantic identity}.
}
$$

This directly continues the central problem of ICNS.

---

# 10. No-Information-Ex-Nihilo Principle

## 10.1 Why Add This Principle?

If one claims that "numerical values contain additional internal information," the most dangerous misunderstanding is:

> Simply by imagining it, one can extract any information from a number.

This paper explicitly excludes that claim.

---

## 10.2 Principle

Any computable information added to:

$$
\operatorname{Envelope}(\mathfrak N)
$$

should have at least one traceable source:

$$
\boxed{
\text{source}
\vee
\text{derivation}
\vee
\text{probe rule}
\vee
\text{structural invariant}
\vee
\text{certificate}.
}
$$

In other words:

$$
\boxed{
\text{information discovery}
\neq
\text{information invention}.
}
$$

---

## 10.3 Inference and Observation Must Be Distinguished

If information results from model inference, it should be labeled:

$$
\text{INFERRED}.
$$

If it is read directly:

$$
\text{OBSERVED}.
$$

If it is formally derived:

$$
\text{DERIVED}.
$$

If it is not yet known:

$$
\text{UNKNOWN}.
$$

One must not make:

$$
\boxed{
\text{UNKNOWN}
\rightarrow
\text{TRUE}
}
$$

without intervening evidence.

---

# 11. Same-Value, Different-Object Theorem

## 11.1 Theorem

If:

$$
\pi_v(\mathfrak N_1)
=
\pi_v(\mathfrak N_2)
$$

and there exists a canonical-preserving observable:

$$
O
$$

such that:

$$
O(\mathfrak N_1)
\neq
O(\mathfrak N_2),
$$

then:

$$
\boxed{
\mathfrak N_1
\neq
\mathfrak N_2.
}
$$

---

## 11.2 Proof

If, for contradiction:

$$
\mathfrak N_1
=
\mathfrak N_2,
$$

every well-defined observable should satisfy:

$$
O(\mathfrak N_1)
=
O(\mathfrak N_2).
$$

and:

$$
O(\mathfrak N_1)
\neq
O(\mathfrak N_2)
$$

a contradiction.

Hence:

$$
\mathfrak N_1
\neq
\mathfrak N_2.
$$

Q.E.D.

---

# 12. One Object, Multiple Projections Theorem

If:

$$
\pi_i,
\pi_j
$$

are both admissible projections of the same canonical object $\mathfrak N$, then even if:

$$
\pi_i(\mathfrak N)
\neq
\pi_j(\mathfrak N),
$$

this does not imply a change in object identity.

Therefore:

$$
\boxed{
\text{different views}
\not\Rightarrow
\text{different object}.
}
$$

This supplies a common identity foundation for MNVP multi-perceptual projections, CAIR multi-program projections, ICNS multiple representations, AIC multiple indices, and cross-domain floating-point / exact-number representations.

---

# 13. Scalar collapse

## 13.1 Definition

If the complete information:

$$
\mathfrak N
$$

is reduced to retaining only:

$$
\pi_v(\mathfrak N),
$$

this is called:

$$
\boxed{
\text{scalar collapse}.
}
$$

---

## 13.2 Scalar Collapse Need Not Be an Error

If the task requires only:

$$
x+y,
$$

and the other information is irrelevant to the task, scalar collapse may be a reasonable compression.

NIOT therefore does not argue that every computation must permanently retain the entire envelope; rather, it argues:

$$
\boxed{
\text{collapse should be treated as a projection decision, not an ontological truth.}
}
$$

---

## 13.3 Task-relative sufficiency

For a task:

$$
\mathcal T,
$$

If:

$$
\pi_v(\mathfrak N)
$$

suffices to complete the task, then:

$$
\boxed{
\operatorname{Sufficient}
(
\pi_v,
\mathcal T
)
=
1.
}
$$

Only if it is insufficient must other informational directions be expanded.

This provides an ontological foundation for subsequent "information-gap-driven probe selection."

---

# 14. Numerical Information Objects Are Not Merely Renamed Metadata

Traditional metadata often takes the form:

$$
x
+
\{
m_1,m_2,\ldots
\}.
$$

This design assumes:

$$
x
$$

is still the subject, and everything else is merely attached to it.

NIOT instead permits:

$$
v
$$

to be only a projection coordinate of:

$$
\mathfrak N
$$

as displayed above.

Thus relation, history, address, and certificate may, for particular tasks, become first-class computational objects alongside scalar value.

In some tasks $\pi_v$ is most important; in others, provenance, certificates, or history may be more important.

Thus:

$$
\boxed{
\text{value-first}
}
$$

is not the only acceptable ontology.

---

# 15. Numerical Local Worlds

## 15.1 Definitional Intuition

A numerical information object may serve as the center:

$$
\mathfrak N_x.
$$

By establishing representation neighbors, provenance neighbors, derivation neighbors, relation neighbors, spectral neighbors, bridge neighbors, and certificate neighbors around it, one obtains:

$$
\boxed{
\mathcal W(\mathfrak N_x).
}
$$

called a numerical local information world.

---

## 15.2 "Microscopic" Does Not Mean More Decimal Places

For example:

$$
0.73125
$$

Computing another hundred thousand digits after this is not necessarily what this paper means by "microscopic."

The microscopic perspective in this paper is:

$$
\boxed{
\text{increasing the distinguishable informational directions.}
}
$$

Thus precision depth is only one form of making a view microscopic.

---

# 16. Discrete and Continuous Numbers Can Both Use This Framework

For:

$$
n\in\mathbb Z,
$$

one can establish:

$$
\mathfrak N_n
$$

and distinguish whether it is an index, count, version, graph-node ID, cardinality, or time step.

For:

$$
x\in\mathbb R,
$$

one can add exact / approximate status, completion provenance, interval enclosures, decimal expansions, machine projections, and measurement semantics.

For:

$$
z\in\mathbb C,
$$

one can distinguish Cartesian projection, polar projection, phase structure, and branch information.

NIOT is therefore not restricted to floating-point real numbers.

---

# 17. Precise Interface with ICNS

ICNS addresses:

$$
\boxed{
\text{How should a numeral token be fully interpreted?}
}
$$

NIOT addresses:

$$
\boxed{
\text{What informational structure can an interpreted numerical value have once it enters the computational world?}
}
$$

Therefore:

$$
\boxed{
\text{token}
\xrightarrow{\mathrm{ICNS}}
\text{interpreted numeral}
\xrightarrow{\mathrm{NIOT}}
\mathfrak N.
}
$$

ICNS is an important entry point to NIOT, but NIOT does not require every numerical object to originate from a textual numeral. Sensors, model outputs, graph nodes, tensor cells, and symbolic solvers can also establish $\mathfrak N$ directly.

---

# 18. Precise Interface with CAIR

CAIR has already proposed:

$$
\text{canonical object}
\neq
\text{projection}.
$$

NIOT narrows this principle to the numerical domain:

$$
\boxed{
\mathfrak N
\neq
\pi_v(\mathfrak N).
}
$$

Thus one can consider in the future:

$$
\boxed{
\text{Canonical Numerical Object}
}
$$

as a numerical specialization of CAIR.

But this paper does not currently merge NIOT directly into CAIR: NIOT requires mathematical ontological research, while CAIR primarily addresses computational authority ontology and governance; the two should first remain independently verifiable.

---

# 19. Precise Interface with ISQL

The value of ISQL lies in:

$$
\boxed{
\text{difference}
\rightarrow
\text{anchor}
\rightarrow
\text{address}
\rightarrow
\text{relation}
\rightarrow
\text{computable state}.
}
$$

The NIOT Envelope can be implemented through ISQL-type structures, particularly:

$$
a,
r,
h.
$$

Therefore:

$$
\boxed{
\mathrm{NIOT}
+
\mathrm{ISQL}
}
$$

may form a semantic runtime for microscopic numerical information.

---

# 20. Precise Interface with MNVP

MNVP addresses:

$$
\text{Semantic Numerical Object}
\rightarrow
\text{Perceptual Representation}.
$$

NIOT supplies the richer source:

$$
\mathfrak N.
$$

MNVP may select:

$$
\pi_v,
\pi_q,
\pi_p,
\pi_r
$$

and other informational directions to generate views.

Therefore:

$$
\boxed{
\mathrm{NIOT}
\rightarrow
\mathrm{MNVP}
}
$$

is a natural downstream relation.

---

# 21. Precise Interface with Floating-Point Theory

If:

$$
f\in\mathbb F
$$

is a machine floating-point value, one can establish:

$$
\mathfrak N_f.
$$

At minimum, it can record bit representation, precision, rounding mode, exception state, source domain, and bridge certificate.

More generally:

$$
\boxed{
\pi_F:
\mathfrak N
\rightarrow
\mathbb F.
}
$$

A float is therefore:

$$
\boxed{
\text{finite machine projection}
}
$$

rather than the entire mathematical / informational object.

---

# 22. Minimal Axiom Set of This Paper

## Axiom I — Layer Separation

$$
\boxed{
\text{abstract value}
\neq
\text{representation}
\neq
\text{numerical information object}.
}
$$

## Axiom II — Projection Non-Identity

$$
\boxed{
\pi(\mathfrak N)
\neq
\mathfrak N
}
$$

unless an information isomorphism has been proved on the specified domain.

## Axiom III — Identity Unit

$$
\boxed{
\forall\mathfrak N,
\quad
\exists\mathbf 1_{\mathfrak N}.
}
$$

## Axiom IV — Indexed Identity Preservation

$$
\boxed{
\iota(\mathfrak N,i)
=
\iota(\mathfrak N).
}
$$

holds for every admissible identity-preserving slice.

## Axiom V — Value Equivalence Is Weaker Than Object Identity

$$
\boxed{
\mathfrak N_1\sim_v\mathfrak N_2
\not\Rightarrow
\mathfrak N_1=\mathfrak N_2.
}
$$

## Axiom VI — No Information Ex Nihilo

New computable information must have a traceable source, derivation, probe rule, structural invariant, or certificate.

## Axiom VII — Unknown Is Not True

$$
\boxed{
\text{UNKNOWN}
\neq
\text{TRUE}.
}
$$

---

# 23. First Set of Provable Propositions

## Proposition 1 — Scalar collapse creates equivalence classes

From:

$$
\pi_v:
\mathfrak X
\rightarrow
X
$$

one can establish an equivalence relation on $\mathfrak X$:

$$
\mathfrak N_1\sim_v\mathfrak N_2
\iff
\pi_v(\mathfrak N_1)=\pi_v(\mathfrak N_2).
$$

Thus a scalar view can be understood as:

$$
\boxed{
\mathfrak X/\sim_v.
}
$$

In other words:

> Traditional scalar numerical space can be regarded as a quotient projection of a richer numerical-information space under value equivalence.

---

## Proposition 2 — More probes refine the quotient

If a new observable is added:

$$
P,
$$

Define:

$$
\mathfrak N_1
\sim_{v,P}
\mathfrak N_2
$$

if and only if:

$$
\pi_v(\mathfrak N_1)=\pi_v(\mathfrak N_2)
$$

and:

$$
P(\mathfrak N_1)=P(\mathfrak N_2).
$$

then:

$$
\boxed{
\sim_{v,P}
\subseteq
\sim_v.
}
$$

Thus the essence of adding a probe can be understood as:

$$
\boxed{
\text{refine equivalence classes}.
}
$$

This directly establishes a mathematical entry point for the Probe Calculus of Paper 02.

---

## Proposition 3 — Identity can survive quotient refinement

Even if:

$$
\sim_{v,P_1}
\supset
\sim_{v,P_1,P_2}
\supset
\cdots
$$

is continually refined, the identity of an individual canonical object:

$$
\mathfrak N
$$

namely:

$$
\iota(\mathfrak N)
$$

need not change as a result.

Thus:

$$
\boxed{
\text{increasing observational resolution}
\neq
\text{changing object identity}.
}
$$

---

# 24. From Equivalence-Class Refinement to "Microscopic Numerical Worlds"

Here one can explain more rigorously for the first time:

> What entering a microscopic numerical world actually means.

Not:

$$
x
\rightarrow
\text{more digits},
$$

but:

$$
\boxed{
[\mathfrak N]_{\sim_v}
\supseteq
[\mathfrak N]_{\sim_{v,P_1}}
\supseteq
[\mathfrak N]_{\sim_{v,P_1,P_2}}
\supseteq
\cdots
}
$$

That is:

$$
\boxed{
\text{continually refining the distinguishable equivalence classes of the same numerical object.}
}
$$

This is a more accurate mathematical description than "magnifying a number."

---

# 25. Distinguishability Is Not the Entire Ontology

Even if probes continue to be added, one cannot directly claim:

$$
\boxed{
\text{we have exhausted the ontology}.
}
$$

Because:

1. There may be probes not yet defined;
2. There may be unobservable structures;
3. There may be task-relative irrelevance;
4. There may be computational infeasibility;
5. There may be indistinguishable equivalence classes.

NIOT therefore adopts:

$$
\boxed{
\text{open-ended epistemic expansion}
}
$$

rather than:

$$
\boxed{
\text{claim of total hidden information}.
}
$$

---

# 26. Local Completeness Rather than Absolute Completeness

For a task $\mathcal T$, if the probe family:

$$
\mathcal P_{\mathcal T}
$$

already suffices to determine the required invariants, say:

$$
\boxed{
\mathfrak N
\text{ is task-complete under }
\mathcal P_{\mathcal T}.
}
$$

Formally:

$$
\operatorname{Complete}
(
\mathfrak N,
\mathcal P_{\mathcal T},
\mathcal T
)
=
1.
$$

This does not mean that $\mathfrak N$ has been universally fully described.

This distinction underlies the subsequent runtime stopping conditions.

---

# 27. Position within the Subsequent Series

The NIOT series is provisionally fixed at ten papers:

| Paper | Provisional title | Central task |
|---|---|---|
| **00** | Numerical Information Object Theory: Overview | Establish the parent framework |
| **01** | Numerical Identity Units and the Projection Non-Identity Principle | Establish ontology and axioms |
| **02** | Information Probe Calculus | Define probes, composition, fibers, and refinement |
| **03** | Unbounded Expansion, Finite Matrices | Dynamic finite dimension, sparsification, and closure |
| **04** | Spectral Discrimination and Multi-Probe Coupling | Multiple scales, spectra, coupling, and informational distinctions |
| **05** | Numerical Index Geometry | AIC, addresses, reindexing, and local information worlds |
| **06** | Floating-Point Projection, Completion, and Cross-Domain Bridges | float, exact, completion, and rounding |
| **07** | Bidirectional Structural Localization and Responsibility Closure | BCODR generalization, overlap, and spine |
| **08** | Canonical Numerical Architecture | Integration of ICNS, ISQL, CAIR, and MNVP |
| **09** | AI-Native Numerical Information Runtime | Adaptive probes, task closure, verification, and MVP |

This paper is:

$$
\boxed{
\text{NIOT-01}.
}
$$

---

# 28. Directions for Formal Verification

Most central axioms and propositions in this paper are suitable for Lean / Coq formalization.

The first stage can establish the conceptual structure:

```text
structure NumericalInfoObject where
  id        : Identity
  value     : Value
  schema    : Schema
  envelope  : Envelope
```

and define:

```text
scalarProj : NumericalInfoObject → Value
```

and:

```text
ValueEquivalent A B :=
  scalarProj A = scalarProj B
```

Then prove:

1. $\sim_v$ is an equivalence relation;
2. Probe refinement produces a finer equivalence relation;
3. Object equality implies equality of all admissible observables;
4. Different observables imply different objects;
5. Indexed views preserve canonical identity.

This paper does not currently claim that these have been machine-verified; they are explicit proof obligations for the next stage.

---

# 29. Minimal Counterexample Test

Establish two objects:

$$
\mathfrak N_A,
\mathfrak N_B
$$

such that:

$$
\pi_v(\mathfrak N_A)
=
\pi_v(\mathfrak N_B)
=
0.5.
$$

but let:

$$
P_{\mathrm{source}}(\mathfrak N_A)
\neq
P_{\mathrm{source}}(\mathfrak N_B).
$$

If a runtime uses only the scalar:

$$
0.5,
$$

it cannot distinguish them.

If a NIOT runtime can, without changing the scalar:

1. preserve the two canonical identities;
2. probe the source distinction;
3. address the source nodes;
4. feed the source distinction into subsequent computation;
5. still project back to the same $0.5$;

this at least proves:

$$
\boxed{
\text{NIOT object}
\neq
\text{scalar + inert metadata}.
}
$$

because the Envelope has entered computational semantics.

---

# 30. Conclusion

This paper establishes the first formal ontological foundation for Numerical Information Object Theory.

Its core is not overturning:

$$
x\in\mathbb R.
$$

Nor is it claiming that a purely abstract number mysteriously carries all information.

What this paper actually does is distinguish:

$$
\boxed{
\text{value}
\neq
\text{representation}
\neq
\text{identity-bearing numerical object}.
}
$$

Once this distinction holds, the scalar need no longer be the sole center of a numerical system.

For any numerical information object:

$$
\mathfrak N,
$$

we can obtain:

$$
\pi_v(\mathfrak N)=x,
$$

while retaining identity, schema, address, relation, history, provenance, quality, context, and bridge.

Therefore:

$$
\boxed{
x
}
$$

is only a value projection of:

$$
\boxed{
\mathfrak N
}
$$

as displayed above.

And "every number is one" is formally rewritten as:

$$
\boxed{
\forall\mathfrak N,
\quad
\mathfrak N
\text{ can be treated as one identity-preserving informational unit}.
}
$$

This "one" is not the numerical value $1$.

It is:

$$
\boxed{
\text{a local information world that can be sliced, probed, indexed, converted, and projected,
while still being tracked as the same object.}
}
$$

Finally, scalar projection produces:

$$
\mathfrak X/\sim_v,
$$

while adding new probes continually refines these equivalence classes.

Thus "entering microscopic numerical worlds" receives an operational mathematical description for the first time:

$$
\boxed{
\text{starting from coarse-grained value equivalence,
and gradually entering finer information-distinguishability structures.}
}
$$

The next paper will proceed directly from this result to:

$$
\boxed{
\textbf{NIOT-02 — Information Probe Calculus.}
}
$$

The central question will no longer be "what other information does a number have," but:

$$
\boxed{
\text{How can we admissibly probe, compose, compare, refine, and stop probing?}
}
$$

---

# Canonical Source Declaration

This file is the UTF-8 Markdown canonical source for `NIOT-01-v0.1`.

The only mathematical delimiters are `$...$` and `$$...$$`.

This paper distinguishes:
- abstract numerical values;
- representations;
- numerical information objects;
- canonical identity;
- projections;
- observables;
- inferred / observed / derived / unknown states.

This paper does not claim that:
- all abstract numbers inherently contain arbitrary external information;
- every projection is invertible;
- every probe reveals the entire ontology;
- the current axioms have been formally machine-verified.

Subsequent formalization, runtimes, and counterexample tests should use the definitions in this file as one of their reference bases.

===END===