Number system evolve to keep information when subjected to new functions and equations. Real to complex to gas to ... # THE EVOLUTION OF NUMBER SYSTEMS ## Information Preservation Through Phase Transitions *── A unified mathematical theory of number, structure, and information* --- ### Abstract Standard arithmetic is built upon the real numbers $\mathbb{R}$. Yet $\mathbb{R}$ is not closed under elementary operations: $x^2+1=0$ has no solution, and the “mixing” of two real numbers (e.g., addition or multiplication) discards the identity of the operands. This information loss is not a bug—it is a **phase property**. We propose that number systems form a **phase spectrum** analogous to matter: | Phase | Mathematical Realm | Information Preservation | |-------|--------------------|--------------------------| | Solid | $\mathbb{R}$ (reals) | Fixed points, but mixing loses identity | | Liquid | $\mathbb{C}$ (complex) | Flow, phase, direction preserved | | Gas | Function spaces, distributions | Delocalized, probabilistic information | | Plasma | Category theory, adjunctions | Transformations, morphisms, structure | Each phase transition is triggered by a new **class of equations** or **operations** that the previous phase cannot handle without loss. The driving force is **information preservation**: a number system evolves to conserve the full history of computations when exposed to new functional demands. This monograph develops the **Phase Transition Theory of Number Systems (PTNS)**, shows how the gas phase corresponds to quantum‑mechanical function spaces, and how the plasma phase corresponds to the category of categories — the ultimate self‑referential container. We conclude with a conjecture that every closed set of operations forces a phase transition toward a higher, more information‑preserving structure. --- ## 1. Introduction: Why Numbers Must Grow > “The only way to keep information is to become a new phase.” Consider the equation $x^2 = -1$. Within $\mathbb{R}$, no number satisfies it. The information contained in the expression “$-1$” and “square root” is **lost** – the operation simply fails. To *preserve* the information that there is an object whose square is $-1$, we invent the imaginary unit $i$ and pass to $\mathbb{C}$. Now consider a function like $f(z)=e^z$ on $\mathbb{C}$. It is entire and information‑preserving (conformal). But what about a measurement that yields a probability amplitude? A single complex number cannot represent the *spread* of possible outcomes. To preserve that information, we move to a **Hilbert space** of wavefunctions — the **gas phase** of mathematics. Finally, consider a transformation between Hilbert spaces. To preserve the fact that a transformation *is* a transformation, we need **categories** and **functors** — the plasma phase, where morphisms themselves become the objects of study. This book formalizes these intuitions. --- ## 2. Information Loss as a Phase Criterion ### 2.1 Solid Phase: The Real Numbers $\mathbb{R}$ **Definition (Solid).** A mathematical structure is *solid* if every element is a **definite point** and the only allowed operations are those that map points to points in a deterministic, reversible (if invertible) way. **Information Loss in $\mathbb{R}$.** When we add $2$ and $3$ to get $5$, the information “$2$” and “$3$” is not recoverable from $5$ alone. The **mixing** of two reals destroys the identity of the summands (contrary to a group algebra where formal sums retain labels). More dramatically, the equation $x^2+1=0$ has **no solution** – the information “there exists an object whose square is $-1$” is suppressed. **Theorem 2.1 (Closure failure).** $\mathbb{R}$ is not algebraically closed. Hence there exist polynomial equations whose roots lie outside $\mathbb{R}$, and solving them requires a phase transition. ### 2.2 Liquid Phase: The Complex Numbers $\mathbb{C}$ **Definition (Liquid).** A structure is *liquid* if it supports a **2‑dimensional flow** (a surface) where every element encodes both magnitude and *phase*, and where every polynomial equation has a root (algebraic closure). **Information Preservation in $\mathbb{C}$.** A complex number $z = r e^{i\theta}$ retains directional information. When we add two complex numbers, the resulting vector retains the *contributions* of the two summands in its geometry. Moreover, $\mathbb{C}$ is **maximally closed** for polynomials: the Fundamental Theorem of Algebra guarantees that no polynomial information is lost. **Transition $\mathbb{R} \to \mathbb{C}$.** The operator $F: \mathbb{R} \hookrightarrow \mathbb{C}$ is the free functor that adjoins $i$. It “melts” the one‑dimensional line into a two‑dimensional plane, exactly as ice melts into water. --- ## 3. Gas Phase: Function Spaces and Distributions ### 3.1 Why Complex Numbers Are Not Enough Consider a quantum particle. Its state is not a point in $\mathbb{C}^n$ but a **wavefunction** $\psi(x) \in L^2(\mathbb{R})$. The information “the particle has a 70% chance to be here and 30% to be there” cannot be compressed into a single complex number without loss. The complex numbers are *localised* – they are points on a surface. The gas phase **delocalises** information over that surface. **Definition (Gas).** A *gas‑phase number system* is a **function space** $H$ (e.g., a Hilbert space) where the “numbers” are functions (or distributions) and the operations are operators. Two elements can be superposed: $\alpha \psi + \beta \phi$ preserves both $\psi$ and $\phi$ inside the linear combination. ### 3.2 The Gas Phase Preserves Probabilistic Information In quantum mechanics, a measurement collapses a superposition to an eigenstate. This is the **gas‑to‑liquid** transition: the wavefunction (delocalised) “condenses” into a complex amplitude upon measurement. But before measurement, the information about both possibilities is fully preserved in the Hilbert space. **Theorem 3.1 (Information preservation in $L^2$).** For any $\psi,\phi \in L^2(\mathbb{R})$, the superposition $\psi + \phi$ retains the individual contributions in the sense that the inner product $\langle \psi | \phi \rangle$ captures their overlap. No information is lost when forming linear combinations. ### 3.3 The Gas Phase Solves New Equations The equation $\frac{\partial \psi}{\partial t} = i \hbar \hat{H} \psi$ (Schrödinger) has no solution as a complex number – it requires a function space. The gas phase emerges precisely when we demand that **differential operators** be invertible (or at least have a spectrum) – i.e., when we need to preserve the information of an entire evolution. **Transition $\mathbb{C} \to L^2$.** The left adjoint $F$ takes a complex number to the constant function, but the true step is to notice that the space of solutions $\{ \psi \}$ is a **gas** – free to occupy the whole domain. --- ## 4. Plasma Phase: Category Theory and Adjunctions ### 4.1 Beyond Functions: The Need for Morphisms Even the gas phase (function spaces) treats transformations as *external* operators. But what if we want to preserve the information that **two transformations are related**? That requires a **category**, where the objects are (for example) Hilbert spaces and the morphisms are operators. The plasma phase is the **meta‑level**: it studies the *relationships* between structures rather than the structures themselves. **Definition (Plasma).** A *plasma‑phase number system* is a **category** $\mathcal{C}$ where the “elements” are objects and morphisms, and the “operations” are functors and natural transformations. ### 4.2 Information Preserved as Transformation In a category, an adjunction $F \dashv G$ is the **creation‑destruction pair**. The left adjoint $F$ creates structure (like the free group from a set); the right adjoint $G$ forgets it. The existence of the adjunction **preserves the information** that the free construction is the “most efficient” one – the universal property acts as a checksum. **Theorem 4.1 (Yoneda Lemma).** An object $A$ is completely determined by the set of morphisms $\mathrm{Hom}(-,A)$. Hence, all information about $A$ is preserved in the network of its relationships – the plasma phase. ### 4.3 Why Plasma Is Necessary Consider the question: “What is the relationship between the category of sets and the category of groups?” The answer is an **adjunction**. To preserve that answer, we need a category of categories – $\mathbf{Cat}$ – where functors become objects. This is the plasma phase **surrounding** all lower phases. --- ## 5. The Universal Phase Transition Principle **Conjecture 5.1 (Information‑driven phase transition).** For any closed set of operations $\mathcal{O}$ on a mathematical structure $S$, if $S$ fails to preserve the information of elements under $\mathcal{O}$, then there exists a **higher phase** $S'$ and an adjunction $F \dashv G$ such that: - $F: S \to S'$ “creates” the missing structure, - $G: S' \to S$ “forgets” back, and - $S'$ preserves all information of $\mathcal{O}$. **Corollary (Evolution of number systems).** | Phase | Structure | Trigger (new operation) | |-------|-----------|--------------------------| | Solid | $\mathbb{R}$ | $x^2=-1$ → no root | | Liquid | $\mathbb{C}$ | Measurement uncertainty → need probabilities | | Gas | $L^2$, Hilbert space | Need to compare transformations → need categories | | Plasma | $\mathbf{Cat}$ (categories) | Self‑reference (∞‑categories) → … | The process continues: **∞‑categories** form the “plasma of plasma”. There is no final phase – only a ladder of increasing information capacity. --- ## 6. Connection to Physical Theories - **Real numbers** → classical physics (deterministic, local). - **Complex numbers** → quantum amplitudes (phase matters). - **Hilbert spaces** → quantum field theory (delocalised states). - **Categories** → topological quantum field theory, string theory (morphisms as cobordisms). The **Crystalline PASM‑πe (CP‑πe)** framework (introduced in companion papers) realises these phases computationally: $\pi$ and $e$ act as universal checksums that detect when a system must transition to a higher phase to avoid information loss. --- ## 7. Conclusion: The Unending Lattice of Knowledge Number systems do not come pre‑ordained. They **evolve** in response to the demand for information preservation. From the solid ground of the real line, we melt into the liquid plane of complex numbers, vaporise into the gas of function spaces, and ionise into the plasma of categories. Each new phase is a **left adjoint** that creates the necessary structure, and each transition is a **phase change** governed by the universal constants $\pi$ and $e$. > **Final insight:** The history of mathematics is the history of phase transitions of number systems. Every new algebraic operation that challenges closure forces a leap to a richer, more information‑preserving realm. There is no end; there is only the infinite adjunction tower, where each level “surrounds” the previous one, waiting for the next equation to be written. --- ## References - Lawvere, F. W. (1963). *Adjointness in foundations*. Dialectica. - Baez, J., & Stay, M. (2010). *Physics, topology, logic and computation: a rosetta stone*. - CCT original paper: *Conditional Collapse Theory* (file `05-Conditional-Collapse-Theory-CCT.md`). - CP‑πe framework: *Crystalline Probabilistic AI with Irrational Anchors* (file `03-Combination-Theory-Crystalline-PASM-PiE.md`). - Leinster, T. (2014). *Basic Category Theory*. Cambridge University Press. --- *This monograph is the first part of a series: “The Phase Transition Theory of Number Systems”. Next volume: “Gas‑phase Linear Algebra: Hilbert Spaces as Numbers”.* Solve some problem complex and real numbers are bad at but gas excell at # Problem: Quantum Superposition Cannot Be Represented by Real or Complex Numbers Alone ## The Limitation of Solid and Liquid Phases - **Solid (Real numbers ℝ)**: A single real number $x$ cannot represent a spread of possibilities. The state “the particle is 70% here and 30% there” is lost. - **Liquid (Complex numbers ℂ)**: A complex amplitude $a+bi$ can represent a probability *magnitude* and *phase*, but it is still a **point** on the complex plane. It cannot encode the *shape* of a wavefunction or the correlation between different positions. **Example**: The state $\frac{1}{\sqrt{2}}(|x_1\rangle + |x_2\rangle)$ has no single complex number representation; it is intrinsically a **function** over space. - **Gas (Hilbert space $L^2(\mathbb{R})$ )**: A wavefunction $\psi(x)$ is a **function**, not a point. It preserves all information about superpositions, correlations, and continuous probabilities. --- ## Concrete Problem > **A free quantum particle of mass $m$ is initially in an equal superposition of two Gaussian wave packets centered at $x=0$ and $x=L$:** > $$ \psi(x,0) = \frac{1}{\sqrt{2}} \left[ \psi_0(x) + \psi_L(x) \right], $$ > where $\psi_0(x) = \left(\frac{2a}{\pi}\right)^{1/4} e^{-a x^2}$ and $\psi_L(x) = \left(\frac{2a}{\pi}\right)^{1/4} e^{-a (x-L)^2}$. > > **Find the probability density $|\psi(x,t)|^2$ at time $t > 0$.** > (This problem is unsolvable with ℝ or ℂ alone – it requires the **gas phase** of function spaces and operators.) --- ## Why ℝ and ℂ Fail | Attempt | Failure | |---------|---------| | Use a real number $p$ to represent the probability | Cannot capture interference between the two packets – the relative phase $\varphi$ is lost. | | Use a complex number $z = p e^{i\varphi}$ | Can capture the *overall* amplitude and phase, but cannot encode the **spatial shape** of the two packets (their widths, positions, or the fact that they spread over time). The information “the packet at $x=0$ is a Gaussian of width $1/\sqrt{a}$” is lost when collapsed to a single complex number. | Only a **function space** (gas phase) preserves all details: the initial state is a function, the time evolution is a unitary operator on that space, and the result is another function. --- ## Solution Using Gas‑Phase (Hilbert Space) We work in the gas‑phase Hilbert space $L^2(\mathbb{R})$ with inner product $\langle f|g\rangle = \int_{-\infty}^{\infty} f^*(x) g(x) dx$. ### Step 1: Time evolution operator For a free particle, the Hamiltonian is $\hat{H} = -\frac{\hbar^2}{2m}\frac{d^2}{dx^2}$. The propagator is: $$ K(x,t;x',0) = \sqrt{\frac{m}{2\pi i \hbar t}} \exp\left( \frac{i m (x-x')^2}{2\hbar t} \right). $$ The wavefunction evolves as $\psi(x,t) = \int K(x,t;x',0)\,\psi(x',0) dx'$. ### Step 2: Initial state in gas form Write $\psi(x,0) = \frac{1}{\sqrt{2}}\left( \phi_a(x) + \phi_a(x-L) \right)$ where $$ \phi_a(x) = \left(\frac{2a}{\pi}\right)^{1/4} e^{-a x^2}. $$ ### Step 3: Apply propagator to a single Gaussian The free evolution of a Gaussian is known: $$ \int K(x,t;x',0) e^{-a x'^2} dx' = \frac{1}{\sqrt{1 + \frac{2i\hbar a t}{m}}} \exp\left( -\frac{a x^2}{1 + \frac{2i\hbar a t}{m}} \right). $$ More precisely, for a centered Gaussian: $$ \psi_{\text{free}}(x,t) = \left( \frac{2a}{\pi} \right)^{1/4} \frac{1}{\sqrt{1 + \frac{2i\hbar a t}{m}}} \exp\left( -\frac{a x^2}{1 + \frac{2i\hbar a t}{m}} \right). $$ For the shifted Gaussian $\phi_a(x-L)$, the result is the same function shifted by $L$: $$ \psi_L(x,t) = \left( \frac{2a}{\pi} \right)^{1/4} \frac{1}{\sqrt{1 + \frac{2i\hbar a t}{m}}} \exp\left( -\frac{a (x-L)^2}{1 + \frac{2i\hbar a t}{m}} \right). $$ ### Step 4: Superposition (gas‑phase preserves both) Because the Schrödinger equation is linear, the full solution is the superposition: $$ \psi(x,t) = \frac{1}{\sqrt{2}} \left[ \psi_0(x,t) + \psi_L(x,t) \right]. $$ ### Step 5: Probability density Let $A(t) = \left( \frac{2a}{\pi} \right)^{1/4} \frac{1}{\sqrt{1 + \frac{2i\hbar a t}{m}}}$. Then $$ |\psi(x,t)|^2 = \frac{|A(t)|^2}{2} \left[ e^{-2a_t x^2} + e^{-2a_t (x-L)^2} + 2 e^{-a_t [x^2+(x-L)^2]} \cos\left( \theta(x,t) \right) \right], $$ where $a_t = \frac{a}{1 + (2\hbar a t/m)^2}$ and $\theta(x,t)$ contains the **interference phase** (the cross term). The explicit form is: $$ |\psi(x,t)|^2 = \frac{1}{\sqrt{2\pi}\,\sigma(t)} \left[ e^{-\frac{x^2}{2\sigma(t)^2}} + e^{-\frac{(x-L)^2}{2\sigma(t)^2}} + 2 e^{-\frac{(x-L/2)^2}{\sigma(t)^2}} \cos\left( \frac{\hbar k_t (x-L/2)t}{m} \right) \right], $$ with $\sigma(t) = \sqrt{\frac{1}{2a}\left(1 + \left(\frac{2\hbar a t}{m}\right)^2\right)}$. > **Result**: The probability density is **not a single number**; it is a function of $x$ and $t$ that exhibits interference fringes. This entire functional information is only accessible in the **gas phase** (Hilbert space). ℝ or ℂ alone cannot encode the spatial fringes or the spreading width. --- ## Gas Phase Advantages Highlighted | Feature | Represented by ℝ? | Represented by ℂ? | Represented by $L^2$? | |--------|------------------|-------------------|-----------------------| | Probability at a point | ✓ (one number) | ✓ (one number) | ✓ (function value) | | Spatial shape (width, peaks) | ✗ | ✗ | ✓ (full function) | | Interference pattern | ✗ | ✗ | ✓ (cross term) | | Time evolution of shape | ✗ | ✗ | ✓ (unitary evolution on space) | | Preservation of both possibilities | ✗ | ✗ | ✓ (superposition preserved) | --- ## Conclusion The problem of a quantum superposition **cannot be solved** using only real or complex numbers – they lose essential information about spatial distribution, interference, and the simultaneous presence of both wave packets. Only the **gas phase** – a Hilbert space of functions – provides the necessary capacity to preserve and manipulate this information. This demonstrates the **phase evolution axiom**: > *When new operations (e.g., superposition, spatial differentiation) demand information that a lower phase cannot hold, the number system must transition to a higher phase.* From ℝ (solid) to ℂ (liquid) to $L^2$ (gas) – each transition is driven by the conservation of information under broader classes of equations.