← Library EQUILIBRIUM equilibrium-system.com

CONVEYOR OF REASON

CONVEYOR OF ROOM "CRYSTAL SOULS"

Definitions, Graph Diffusion, and the Limit Transition Theorem 

(graph laplasian → operator Laplace-Beltrami / diffusion generator on variety).

1) Laplacean on graph

Let the weighted undirected graph G be given=(V,E), |V|=n. Weight matrix:

W=(w_{ij})_{i,j=1}^n,\quad w_{ij}=w_{ji}\ge 0,\quad w_{ii}=0.

Degrees (weighted):

d_i=\sum_{j=1}^n w_{ij},\qquad D=\mathrm{diag}(d_1,\dots,d_n).

1.1. Irregular Laplasian

L = D - W.

1.2. Normalized laplasian (symmetrical)

L_{\mathrm{sym}} = I - D^{-1/2} W D^{-1/2}.

1.3. Normalized laplasian (accidental/Markovian)

Define the Markov matrix of transitions:

P = D^{-1W, then L\mathrm{rw}} = I - P.

This is a generator of discrete "heat transfer" on the graph.

2) Graph diffusion

2.1. Discrete time (random wandering)

Let ut\in\mathbb{R}^n be the distribution/field at vertices at time t.

u_{t+1} = P\,u_t. 

Equivalently:

u_{t+1}-u_t = -(I-P)u_t = -L_{\mathrm{rw}}u_t.

2.2. Continuous time (graph “heat equation”)

\frac{d u(t)}{dt} = -L\,u(t)\quad \text{or}\quad \frac{d u(t)}{dt} = -L_{\mathrm{rw}}u(t).

Decision:

u(t)=e^{-tL}u(0)\quad \text{or}\quad u(t)=e^{-tL_{\mathrm{rw}}}u(0).

Interpretation: L (or L_-\mathrm{rw} ) is a discrete analogue of the Laplace operator: it measures "how much the value in the node differs from the average in neighbors".

3) Setup for limit theorem (points on manifold)

Let:

Typical weight design: w_{ij} = k_\varepsilon\!\left(\|x_i-x_j\|^2\right), \quad k_\varepsilon(s)=\exp\!\left(-\frac{s}{4\varepsilon}\right) (can also be with a radius cut-off; the main thing is the locality with \varepsilon\to 0).

4) Limit theorem (graph generator → Diffusion Operator)

Below is the standard result form: local graph operator at n\to\infty, \varepsilon\to 0 converges to the diffusion generator at M. Depending on the standard, it will either \Delta_M, or \Delta_M plus drift from density p.

4.1. Random Wandering Operator and its Generator

Define (empirically) P=D^{-1W. For f:M function\to\mathbb{R} The vector f_n\in\mathbb{R}^n, (f_n)_i=f(x_i).


Consider a scaled operator:

\mathcal{G}_{n,\varepsilon} f (x_i)\;:=\;\frac{(P f_n)_i - f(x_i)}{\varepsilon}.

This is a discrete analogue of the “time derivative” in diffusion, because P-I \approx \varepsilon\,\mathcal{L}.

4.2. Theorem (limit generator without “density correction”). Theorem (similarity to the operator with density drift).

Let M be compact, p\in C^2(M), p>0, f\in C^3(M). 

Let \varepsilon=\varepsilon_n\to 0 and n\varepsilon^{m/2+2}\to\infty (The condition is “enough points in the local ball”. Then with probability \to 1 at n\to\infty, For all i:

\mathcal{G}_{n,\varepsilon} f (x_i) \;\longrightarrow\; c_1\,\Delta_M f(x_i)\;+\;c_2\,\langle \nabla \log p(x_i),\,\nabla f(x_i)\rangle.

Where:

Meaning: Ordinary graph random wandering restores diffusion distorted by the density of the sample: there is a drift towards large p(x).

4.3. Theorem (density correction → clean \Delta_M)

To remove the influence of p, use \alpha-normation (often called “diffusion maps normalization”):

Define the generator:

\tilde{\mathcal{G}}_{n,\varepsilon}^{(\alpha)} f (x_i) :=\frac{(\tilde P f_n)_i-f(x_i)}{\varepsilon}.

Theorem (choice \alpha=1 Laplace - Beltrami).

Under the same conditions on M,p,f,n,\varepsilon, for \alpha=1:

\tilde{\mathcal{G}}_{n,\varepsilon}^{(1)} f (x_i) \;\longrightarrow\; c\,\Delta_M f(x_i).

Meaning: the correct rationing “separates geometry from data distribution” and in the limit gives a pure geometric operator.

5) How to read as "the limit of diffusion"

If graph diffusion is continuous:

\frac{d u}{dt} = \mathcal{G}_{n,\varepsilon} u

then at the limit it becomes PDE on the variety:

6) Minimal engineering conclusions for EQUILIBRIUM-741

The next message I will give is the same as the document section (Definition -Proposition -Theorem -Proof sketch) and add error estimates: O offset(\varepsilon) and variance O((n\varepsilon^{m/2+1})^{-1/2typical picture for nuclear approximations).

First, the Laplace-Beltrami operator.

Then - how to interpret it correctly in the model of "Crystal Soul" (as a geometric carrier of states).

I. Laplace-Beltrami

Let: (M, g) be a smooth compact Riemannian manifold of dimension m, where g is a metric.

1. Definition through gradient divergence

\Delta_M f = \mathrm{div}(\nabla f) where:

This is an invariant definition - it does not depend on the coordinates.


2. In Local Coordinates

If g_{ij} — Components of metrics, g=\det(g_{ij}), Toto:

\Delta_M f = \frac{1}{\sqrt{|g|}} \sum_{i,j} \partial_i \left( \sqrt{|g|} \, g^{ij} \, \partial_j f \right)

where g^{ij} is the inverse matrix to g_{ij}.

3. Physical meaning \Delta_M f measures:

how much the value of a function at a point differs from the average for an infinitely small neighborhood.


If: \Delta_M f > 0 — point “below average” (inflow),

if: \Delta_M f < 0 — “above average” (outflow).

4. Spectral structure

Consider the problem: \Delta_M \phi_k = -\lambda_k \phi_k

We receive: 0 = \lambda_1 < \lambda_2 \le \lambda_3 \le \dots

Own functions \phi_k:

This is spectral geometry.

II. Now it's "CRYSTALL OF SOULS" (strict model). Let's take poetry away. Give me a model.

1. Model Postulate

The "soul" is not an object, but a state space having:

Marking: \mathcal{S} = (M, g), where:

This is the Crystal — not solid, but geometrically structured.

2. Why “Crystal”

Crystal is characterized by:

In the language of geometry:

crystal = space with a discrete spectrum of laplasian.

If M is compact, the spectrum is discrete. This is “crystalline.”

3. State dynamics

Let: u(t,x) be the internal state distribution.

Then natural evolution: \frac-\partial u-\partial t- = \Delta_M u

This means:

The system strives for harmonious balance.


4. Harmonic states

Solutions:\Delta_M f = 0 — harmonic functions.

They minimize Dirichlet energy: E(f) = \int_M |\nabla f|^2 \, dV

Interpretation:

harmonic state = minimum internal voltage.


III. Spectrum as a structure of "crystal"

Each eigenvalue: \lambda_k

Determines the extent of the “internal rhythm”.

IV. Connection with the EQUILIBRIUM-741 graph model

If the 741 node graph approximates M, then:

L_{graph} \to c\,\Delta_M

And then:

V. Strict wording of the transition

If:

\frac{P-I}{\varepsilon} \longrightarrow c\,\Delta_M

in the operator norm (under suitable conditions). This is the bridge:

Discrete crystal → Continuous crystal.

VI. Key idea

If you want "Crystal Soul" to be a mathematically correct concept, it should be:

It is not a metaphor, it is a geometric model.


Let's make "Crystal Soul" mathematically richer: from Laplace-Beltrami (on functions) → to the curvature of Ricci (as internal coherence / voltage) → to the Hodge operator (on forms: flows, contours, cavities).

Below is a frame that can really be put in the document.

1) Ricci's curvature in the model "Crystal Soul"

Let the Crystal of the Soul be given as Riemannian manifold (M,g).

1.1. What is Ricci

Ricci's curvature is a tensor \mathrm{Ric}, which aggregates the sectional curvature and answers the question:

“How do volumes and geodesy behave in the middle direction?”

If we take the unit vector v\in T_xM \mathrm{Ric}(v,v) measures the mean curvature of planes containing v.

1.2. Why Ricci is “Connectivity/Sustainability”

Geometric meaning through the divergence of geodesy:

In your language: Ricci = "internal law of attraction/disagreement" in sensitivity metric g.

2) Hoxha operator: transition from functions to “structures”

Laplace-Beltrami \Delta acts on functions. The Hoxha operator extends this to differential forms, that is, to:

This is the perfect language for a “crystal” because a crystal is not only values, but also contours/cycles/flows.

2.1. d, δ and Hox Laplacean

There are two basic operators:

Then Hodg-Laplasian:\Delta_H = d\delta + \delta d

It works on k-forms. On 0-forms (functions), it coincides with Laplace-Beltres (to the accuracy of the accepted sign).

3) Hodge decomposition: “crystal mechanics” in one formula

For a compact M, any k-shape unfolds orthogonally:

\Omega^k(M) = \underbrace{\mathrm{im}(d)}_{\text{exact (gradient part)}} \;\oplus\; \underbrace{\mathrm{im}(\delta)}_{\text{coexact (rotational part)}} \;\oplus\; \underbrace{\mathcal{H}^k}_{\text{harmonic forms}}, where

\mathcal{H}^k=\{\omega\in\Omega^k: \Delta_H\omega=0\}.

The key fact of Hodge:

\mathcal{H}^k \cong H^k_{\mathrm{dR}}(M)

harmonic k-forms are isomorphic to de Rama cohomologies.

Translated into "Crystal Soul"

This is no longer a metaphor: it is an exact decomposition.

4) Ricci Communication Node ↔ Hodge: Weitzenbeck's formula

The most important bond for you is what curvature does to forms.

There is a formula of the Weitzenbek type (schematic):

\Delta_H = \nabla^\*\nabla + \mathcal{R}, where:

For 1-forms \omega (very important special case) there is a classical formula of Bohner:

\frac{1}{2}\Delta |\omega|^2 = |\nabla \omega|^2 + \langle \mathrm{Ric}(\omega), \omega\rangle + \langle \Delta_H \omega, \omega\rangle

The meaning (in one line) of Ricci controls whether stable “contours” can exist (harmonic 1-forms) and how much they are suppressed / supported.

5) "Theorems of Meaning" for the Soul Crystal (in strict form)

5.1. If Ricci is strictly positive - "contours disappear" (at the level of 1-forms)

If \mathrm{Ric} \ge c\,g with c>0, then (under standard compactness conditions) the following often holds:

\mathcal{H}^1 = 0 \quad\Rightarrow\quad H^1_{\mathrm{dR}}(M)=0

Translation: “the crystal has no stable 1-circuits”, there are no inexhaustible “petals of flow” of the first order.

It is a mathematical image of “inner cohesion”: anything that looks like a stream is either gradient (reduced to potential) or fading.

5.2. If Ricci is negative in some places - contours and fashions are easier to fix

Negative curvature zones statistically support:

6) How to apply this to EQUILIBRIUM-741 (discrete prototype)

If M is sampled by a graph:

Then the Soul Crystal becomes computable:

7) Minimum integration formula in "Crystal Soul"

Definition (working): The Soul Crystal is (M,g) plus field/flow \omega\in\Omega^1(M) with dynamics

\frac{\partial \omega}{\partial t} = -\Delta_H \omega \;-\;\nabla \Phi(\omega)

where \Phi is the potential (nonlinearity: values/will/choice), and the geometry g and Ricci define the “resistance/assembly structure”.

If you continue to “build”, the next logical step is to choose what we fix on:


Let's make "Crystal Soul" mathematically richer: from Laplace-Beltrami (on functions) → to the curvature of Ricci (as internal coherence / voltage) → to the Hodge operator (on forms: flows, contours, cavities).

Below is a frame that can really be put in the document.

1) Ricci's curvature in the model "Crystal Soul"

Let the Crystal of the Soul be given as Riemannian manifold (M,g).

1.1. What is Ricci

Ricci's curvature is a tensor \mathrm{Ric}, which aggregates the sectional curvature and answers the question:

“How do volumes and geodesy behave in the middle direction?”

If we take the unit vector v\in T_xM \mathrm{Ric}(v,v) measures the mean curvature of planes containing v.

1.2. Why Ricci is “Connectivity/Sustainability”

Geometric meaning through the divergence of geodesy:

In your language: Ricci = "internal law of attraction/disagreement" in sensitivity metric g.

2) Hoxha operator: transition from functions to “structures”

Laplace-Beltrami \Delta acts on functions.

The Hoxha operator extends this to differential forms, that is, to:

This is the perfect language for a “crystal” because a crystal is not only values, but also contours/cycles/flows.

2.1. d, δ and Hox Laplacean

There are two basic operators:

Then Hodge-laplasian: \Delta_H = d\delta + \delta d,acts on k forms.

On 0-forms (functions), it coincides with Laplace-Beltres (to the accuracy of the accepted sign).

3) Hodge decomposition: “crystal mechanics” in one formula

For a compact M, any k-shape unfolds orthogonally:

\Omega^k(M) = \underbrace{\mathrm{im}(d)}_{\text{exact (gradient part)}} \;\oplus\; \underbrace{\mathrm{im}(\delta)}_{\text{coexact (rotational part)}} \;\oplus\; \underbrace{\mathcal{H}^k}_{\text{harmonic forms}}

where

\mathcal{H}^k=\{\omega\in\Omega^k: \Delta_H\omega=0\}.

The key fact of Hoxha: \mathcal{H}^k \cong H^k_{\mathrm{dR}}(M)

harmonic k-forms are isomorphic to de Rama cohomologies.

Translated into "Crystal Soul"



This is no longer a metaphor: it is an exact decomposition.

4) Ricci Communication Node ↔ Hodge: Weitzenbeck's formula

The most important bond for you is what curvature does to forms.

There is a formula of the Weitzenbek type (schematic):

\Delta_H = \nabla^\*\nabla + \mathcal{R}, where:

For 1-forms \omega (very important special case) there is a classical formula of Bohner:

\frac{1}{2}\Delta |\omega|^2 = |\nabla \omega|^2 + \langle \mathrm{Ric}(\omega), \omega\rangle + \langle \Delta_H \omega, \omega\rangle

Meaning (in one line)

Ricci controls whether stable "contours" (harmonic 1-forms) can exist and how much they are suppressed/supported.

5) "Theorems of Meaning" for the Soul Crystal (in strict form)

5.1. If Ricci is strictly positive - "contours disappear" (at the level of 1-forms)

If \mathrm{Ric} \ge c\,g with c>0, then (under standard compactness conditions) the following often holds:

\mathcal{H}^1 = 0 \quad\Rightarrow\quad H^1_{\mathrm{dR}}(M)=0

Translation: “the crystal has no stable 1-circuits”, there are no inexhaustible “petals of flow” of the first order.

It is a mathematical image of “inner cohesion”: anything that looks like a stream is either gradient (reduced to potential) or fading.

5.2. If Ricci is negative in some places - contours and fashions are easier to fix

Negative curvature zones statistically support:

6) How to apply this to EQUILIBRIUM-741 (discrete prototype)

If M is sampled by a graph:

Then the Soul Crystal becomes computable:

7) Minimum integration formula in "Crystal Soul"

Definition (Working):

The Soul Crystal is (M,g) plus field/flow \omega\in\Omega^1(M) with dynamics

\frac{\partial \omega}{\partial t} = -\Delta_H \omega \;-\;\nabla \Phi(\omega)

where \Phi is the potential (nonlinearity: values/will/choice), and the geometry g and Ricci define the “resistance/assembly structure”.

If you continue to “build”, the next logical step is to choose what we fix on:

Take k=1 is the most “live” and controlled level: flows, directions, transport, internal circulations.

Below is the finished section for the document: 1-forms + Hodge Decomposition + Hodge Laplacean + Contact with Ricci + Dynamics.

Section: Crystal Soul on 1-forms

1. Basic structure

Let the "Crystal of the Soul" be Riemannian diversity

\mathcal{S}=(M,g), where M is a compact smooth manifold, g is a Riemannian metric.

Condition/voltage/transport are described by 1-form

\omega \in \Omega^1(M).

Intuitively, \omega is a “dimension along the direction”: how much “flow” passes along the tangent vector.

2. Operators d, \delta and Hodge Laplacian on 1 forms

2.1. External derivative d:\Omega^1(M)\to \Omega^2(M)

measures the “vortex/circulation” component (rotor generalization).

2.2. Codifferential

\delta:\Omega^1(M)\to \Omega^0(M)

— adjoined to d operator with respect to L^2-scalar product, divergence generalization.

2.3. Hodge Laplacean

\Delta_H \omega = (d\delta+\delta d)\omega,\qquad \omega\in\Omega^1(M).

It is a key “operator of internal alignment” of flows in the crystal.

3. Hodge decomposition for 1-forms (main mechanics)

For the compact M:

\Omega^1(M)=\mathrm{im}(d)\ \oplus\ \mathrm{im}(\delta)\ \oplus\ \mathcal{H}^1,

Where \mathcal{H}^1=\{\omega:\Delta_H\omega=0\}.

So any 1 form decomposes in a single way:

\omega = d\varphi \;+\; \delta \psi \;+\; h, where:

3.1. Interpretation (Strictly Linked)

The key fact of Hoxha: \mathcal{H}^1 \cong H^1_{\mathrm{dR}}(M).

That is, h is the topological memory of the crystal at the level of 1-circuits.

4. Energy 1-forms and “crystal tension”

Determine the energy (Dirichlet-type):

E(\omega)=\int_M \big(|d\omega|^2 + |\delta\omega|^2\big)\,dV.

Minimums in a fixed class of cohomology are exactly the harmonic 1-forms of h.

That is, “sustainable contours” are energetically optimal representatives of topological classes.

5. Relationship with Ricci: Weitzenbeck/Bochner formula on 1 forms

There is a decomposition (operating):

\Delta_H \omega = \nabla^\*\nabla\,\omega \;+\; \mathrm{Ric}(\omega),

where:

Hence the “energy” formula of Bohner:

\int_M \langle \Delta_H\omega,\omega\rangle\,dV = \int_M \Big(|\nabla\omega|^2 + \langle \mathrm{Ric}(\omega),\omega\rangle\Big)\,dV.

5.1. Hard consequence (crystal collection)

If \mathrm{Ric}\ge c\,g \quad (c>0), the Harmonic 1-no form:

\mathcal{H}^1=\{0\}, \quad\Rightarrow\quad H^1_{\mathrm{dR}}(M)=0.

That is, with a strictly positive Ricci curve, “structural loops” 1-of order" is suppressed: the crystal becomes "topologically single-fold" at k level=1.

6. Dynamics 1-forms: graph diffusion → continuous diffusion

Determine the evolution of the flow \omega(t):

6.1. Linear relaxation (flow diffusion)

\frac{\partial \omega}{\partial t} = -\Delta_H \omega.

Property: at t\to\infty, \omega(t)\to h,

That is, the system “washes” the exact and vortex and leaves only a harmonic (topological) component.

This is the most important meaning: diffusion reveals memory.

6.2. Managed Evolution (Will/External Impact)

\frac{\partial \omega}{\partial t} = -\Delta_H \omega + \mathcal{U}(t),

where \mathcal{U} is the source/control (can be local).

7. A discrete prototype for EQUILIBRIUM-741 at the level of 1-forms

To make it calculated:

(In a “pure graph” approximation, the first part is often taken as the basis; for a complete \Delta_1, another layer of 2-simplexes is needed - triangles / clicks).

Next you can:

Ready-made semantic conclusion (strictly)

Crystal Soul (k=1) — This is (M,G) + 1-form \omega, where:


“Algorithm 1: Discrete Hodge Decomposition on EQUILIBRIUM-741 (k=1)” in a rigorous engineering form.

Two options:

Symbols (for both options)

Let G=(V,E), |V|=n=741, |E|=m. Select the orientation of each edge e=(i\to j) The incident node-rebre matrix:

B \in \mathbb{R}^{n\times m},\quad B_{ve}= \begin{cases} -1,& e\text{ leaves }v\\ +1,& e\text{ enters }v\\ 0,& \text{otherwise} \end{cases}

1-form (flow on the ribs):\omega \in \mathbb{R}^{m}

(value \omega_e on the orientated rib).

Rib weight (conductivity) w_e>0. Let's collect the diagonal:

W_1 = \mathrm{diag}(w_e)\in\mathbb{R}^{m\times m}.

the Scalar Product 1-forms\langle a,b\rangle_{1} = a^\top W_1\ b.

A) Algorithm on one graph (without triangles)

Goal: find the decomposition \omega = d\varphi \;+\; \omega_{\mathrm{cyc}} \;+\; h, where:

On a pure graph, the “vortex” part corresponds to cycles (cycle space). Full \delta d via 2-forms will appear in option B.

A1. Operators d_0 and \delta_1

Discrete differential in 0Forms (potentials \varphi\in\mathbb{R}^n): d_0 \varphi \;=\; B^\top \varphi \in \mathbb{R}^m.

Codifferential on 1-forms (including weights):

\delta_1 \omega \;=\; B W_1\,\omega \in \mathbb{R}^n.

A2. Gradient part d\varphi (Poisson solution on graph)

We are looking for \varphi at least energy:

\varphi = \arg\min_{\varphi} \|\omega - B^\top\varphi\|_{W_1}^2.

Normal Equations: (B W_1 B^\top)\,\varphi = B W_1\,\omega.

Marking: L_0 := B W_1 B^\top — Weighed Laplasian on the tops.

Important: L_0 degenerate (constants in the nucleus), so we fix "calibration": for example, \varphi_{v_0}=0 (a) a single reference unit, or \sum_v \varphi_v=0. After the decision: \omega_{\mathrm{grad}} := B^\top \varphi.

A3. Balance after gradient

r := \omega - \omega_{\mathrm{grad}}.

Then automatically: \delta_1 r = B W_1 r = 0 (The remainder is divergence-free).

A4. Harmonic part h and cyclic part \omega_QQQ\mathrm{cyc}

On the pure graph, the divergence-free flows r are decomposed into:

Convenient calculated method:

A4.1. Build a Cycle Base

For example, through a tree.

Here \beta_1 = m-n+\#components is the cycle rank.

A4.2. Projection of the remainder per cycle space

We are looking for coefficients a: a = \arg\min_a \|r - C a\|_{W_1}^2

Normal Equations: (C^\top W_1 C)\,a = C^\top W_1 r.

Then:

\omega_{\mathrm{cyc}} := C a,\qquad h := r - \omega_{\mathrm{cyc}}

In this graph variant, h is that which is orthogonal to the selected cyclic subspace (depending on choice C). To obtain a canonical harmonic part, a better variant is B (with 2-simplexes) or spectral projection onto the nucleus 1-laplasian (below).

A4.3. Canonical variant through 1-laplasian (spectral)

Define graph 1-laplasian (without triangles):

L_1 := W_1^{-1} B^\top B W_1 \quad \text{(or the symmetric form } B^\top B \text{ with a suitable metric)}

Then the harmonic part is the projection of r to \ker L_1 (in practice: find your own vectors with \lambda\approx 0, assemble base H, project).

B) Full Simplicial Algorithm (nodes+rebre+triangles)

This is the most “correct” option, because it gives an exact analogue:

\omega = d_0\varphi + \delta_2 \psi + h,

where \psi is a 2 form on triangles.

B0. Need 2-simplexes

Construct a set of triangles F (clicks of size 3) - for example, according to the condition "if three edges are present, there is a triangle" (Vietoris-Rips on the graph) or only "trusted" triangles.

Let |F|=p.

Edge operator edge-triangle (incident):

B_2 \in \mathbb{R}^{m\times p}

where the column corresponds to the oriented triangle and the rows to the edges; elements +1/-1/0 depending on the alignment of orientations.

Weight of triangles:

W_2=\mathrm{diag}(w_f)\in\mathbb{R}^{p\times p}.

B1. Full co-differentials

d_0 = B^\top : \mathbb{R}^n\to \mathbb{R}^m, \qquad d_1 = B_2^\top : \mathbb{R}^m\to \mathbb{R}^p.

Related (weighted):

\delta_1 = B W_1 : \mathbb{R}^m\to \mathbb{R}^n, \qquad \delta_2 = W_1^{-1} B_2 W_2 : \mathbb{R}^p\to \mathbb{R}^m.

B2. 1-Laplasian Hodge (canonical)

\Delta_1 = d_0\delta_1 + \delta_2 d_1 = B^\top B W_1 \;+\; W_1^{-1} B_2 W_2 B_2^\top.

(Equivalent symmetric shapes are possible through “root weights.”)

B3. Step 1 - find \varphi (gradient part)

Same as in A2:

(B W_1 B^\top)\,\varphi = B W_1\,\omega, \quad \text{(gauge fixing is required)}

\omega_{\mathrm{grad}} = B^\top \varphi, \quad r_1=\omega-\omega_{\mathrm{grad}}.

B4. Step 2 — find \psi (the vortex part through 2-forms)

Looking for \psi\in\mathbb{R}^p to approximate r_1 How \delta_2\psi:

\psi = \arg\min_\psi \|r_1 - \delta_2 \psi\|_{W_1}^2.

Normal Equations:

(d_1 W_1^{-1} d_1^\top)\,\psi = d_1 r_1

in a more transparent form through B_2:

(B_2^\top W_1^{-1} B_2 W_2)\,\psi = B_2^\top r_1.

After the decision:

\omega_{\mathrm{curl}} := \delta_2 \psi, \qquad h := r_1 - \omega_{\mathrm{curl}}.

B5. Checking the harmony

In the full complex, the harmonic 1-form satisfies:

\delta_1 h = 0 \quad\text{and}\quad d_1 h = 0,

That is, both divergence-free and vortex-free.

Algorithm 1 (summary): Discrete Hodge Decomposition on EQUILIBRIUM-741 (k=1)

Login:

oriented graph G, rib weights W_1, stream \omega\in\mathbb{R}^m.

(Optional: triangles F, B_2, weights W_2.)

Output:

\omega_{\mathrm{grad}}, \omega_{\mathrm{curl}or \omega_{\mathrm{cyc}h such that

\omega=\omega_{\mathrm{grad}}+\omega_{\mathrm{curl}}+h.

Steps:

Practical binding to "Crystal Soul"

On k=1 you have a pure typology:

Diffusion on 1 forms:

\frac{d\omega}{dt}=-\Delta_1 \omega

washes away the gradient and vortex and leaves the harmonic.

We fix the default so that it is:

• computable at 741 node,

• did not explode the number of triangles,

• gave correct B_2, \Delta_1 and decomposition.

Below is the EQUILIBRIUM-741 Default Triangles v0.1 standard.


Default: how to build triangles F (2-simplexes)


0) The input we consider to be already specified

• Count G=(V,E), |V|=741

• rib weight w_{ij}>0 (cost) or conductivity c_{ij}>0

• kNN-graph is already built (typically k=10..16)


1) Make the graph undirected "for clicks"

We search for triangles on an undirected frame:

An edge \{i,j\} exists if there is i\to j or j\to i.

(The orientation will be needed later for signs in B_2.)

2) Construct triangles as clicks of size 3


We include the triangle \{i,j,k\} if all three edges are present:

\{i,j\},\{j,k\},\{k,i\}\in E.

It is Vietoris-Rips at \epsilon, but is implemented as “all 3-clicks of the graph”.


3) Limitation of complexity (required)


So that p=|F| does not become huge, enter Local Cutoff.


Default-limitation


For each vertex i:

1. take her neighbors N(i)

2. Consider pairs (j,k)\subset N(i) that are connected by an edge \{j,k\}

3. add each triangle found, \{i,j,k\}, as a candidate

Then we use the cap:

• cap_tri_per_node = 60 (The hard limit of “how many triangles a vertex can produce”)

• if there are more candidates - leave the "strongest" in terms of weight speed (see below)

This keeps the p within reasonable limits.

4) Sort by "strongest triangles"

We need a quick “quality triangle” to cut the extra.

If you have weight, it is worth w (less than = is better)

s(i,j,k) = w_{ij}+w_{jk}+w_{ki}

We leave triangles with a minimum s.

If you have a weight - conductivity c (more than = better)

s(i,j,k) = c_{ij}\,c_{jk}\,c_{ki}

Let's leave the triangles with a maximum s.

5) Triangle orientation (for signs)

We define canonical orientation:

For any \{i,j,k\}, order i<j<k and define the oriented 2-simplex:[i,j,k]

Its oriented border: \partial[i,j,k] = [j,k] - [i,k] + [i,j]

This rule specifies the characters in the matrix B_2.

6) Construction of matrix B_2 (rib-triangle)

Let each edge in E have a fixed orientation (as you originally chose): e=(u\to v).

For triangle t=[i,j,k] with i<j<k its boundary contains oriented edges:

• [i,j] with sign +1

• [i,k] with sign -1

• [j,k] with sign +1

Further for each such rib:

• if the orientation of the edge in the graph coincides with [a,b] (i.e. a\to b) then B_2(e,t)=\pm 1 As above

• If the graph is oriented the opposite of b\to a, the sign changes: B_2(e,t)=\mp 1

Everything. This is a standard simplicial orientation.


7) Triangle weights W_2 (default)


We take a steady default: If you have a “value” w

First we translate in conductivity c_{ab}=\exp(-w_{ab}/\tau), and then:

w_t = \frac{1}{3}\left(c_{ij}+c_{ik}+c_{jk}\right)

That is, a triangle is “strong” if the connections within it are strong.

If you already have a C 

w_t = \frac{1}{3}\left(c_{ij}+c_{ik}+c_{jk}\right)

Conclusion: What to Expect and What to Consider Next

After that, you have:

• B (node-rib)

• W_1 (rib weights)

• B_2 (reverse triangle)

• W_2 (triangle weights)

And you're building a Hodge Laplacian on 1 forms:

\Delta_1 = d_0\delta_1 + \delta_2 d_1

in matrix form (working default implementation):

\Delta_1 = B^\top B W_1 +  W_1^{-1} B_2 W_2 B_2^\top

(Equivalent symmetric variants are acceptable; the main thing is consistency with the selected scalar works.)

Then you start decomposing:

1. decide on \varphi (gradient)

2. decide on \psi (Vortex through triangles)

3. get h (harmonic)

Default parameters (to fix the standard)

• kNN: k=12

• cap_tri_per_node: 60

• \tau: Take the median w_{ij} by the edges (as scale), or fixate

• Calibration of capacity: \sum_i \varphi_i=0 (or fix one node)