KR
Science Mining Method 1. Overall Structure 4-Step Loop Why a Loop 2. Banya Framework 5 Steps Step 1: Start from Banya Equation Step 2: Norm Substitution Step 3: Constants + Hypothesis Step 4: Domain Transform Step 5: Discovery + Verdict Round Repetition 3. Parallel Expert Verification Why Parallel Assignment Example What Experts Receive 4. Supervisor Review Review Criteria Numerology Filtering Review Grades 5. Library Accumulation Registration Criteria Usage Re-substitution Map 6. CAS Axioms Essential 1: Operator Outside Time (Axiom 3) Essential 2: DATA Access Cost (Axiom 4,5) Essential 3: Collapse = Write (Axiom 6) Essential 4: Self-Reference (Axiom 8) 7. Document Rules 8. Practical Cycle Example Round 1: Zeroth-Order Round 2: Precision Derivation Round 3: Information Theory Round 4: Cosmic Scale 9. Precautions 10. Tools
Science Mining Method
Science Mining Method 1. Overall Structure 4-Step Loop Why a Loop 2. Banya Framework 5 Steps Step 1: Start from Banya Equation Step 2: Norm Substitution Step 3: Constants + Hypothesis Step 4: Domain Transform Step 5: Discovery + Verdict Round Repetition 3. Parallel Expert Verification Why Parallel Assignment Example What Experts Receive 4. Supervisor Review Review Criteria Numerology Filtering Review Grades 5. Library Accumulation Registration Criteria Usage Re-substitution Map 6. CAS Axioms Essential 1: Operator Outside Time (Axiom 3) Essential 2: DATA Access Cost (Axiom 4,5) Essential 3: Collapse = Write (Axiom 6) Essential 4: Self-Reference (Axiom 8) 7. Document Rules 8. Practical Cycle Example Round 1: Zeroth-Order Round 2: Precision Derivation Round 3: Information Theory Round 4: Cosmic Scale 9. Precautions 10. Tools

This document is a supplementary document to the Banya Framework Comprehensive Report. It is an operational manual describing the concrete methods for deriving physical constants and laws using the Banya Framework. These are the exact methods used by Han Hyukjin and AI (Claude). Anyone who follows this manual can reproduce the same results.

Science Mining Method

Banya Framework Operational Manual

Inventor: Han Hyukjin (bokkamsun@gmail.com)

Date: 2026-03-23



Chapter 1. Overall Structure

Banya Framework science mining is a 4-step loop. The more you repeat this loop, the larger the library grows, and the larger the library grows, the fewer places hidden values can escape to.

4-Step Loop

Banya Framework 5-Step Recursive Substitution
Core Engine
|
Parallel Expert Verification
Quality Assurance
|
Supervisor Review
Numerology Filtering
|
Library Accumulation
Weapons for the Next Round
|
Return to the beginning

Why a Loop

Think of simultaneous equations. If there are 5 unknowns and only 2 equations, you cannot solve it. With 3 equations, the solution narrows. With 4, it is nearly determined. With 5, a unique solution emerges.

The Banya Framework loop works the same way. If Round 1 yields 1 discovery, Round 2 re-substitutes that discovery and reduces the unknowns by 1. Round 3 reduces them by 2. The more you iterate, the more the solution converges. The $\alpha$ derivation converged from 0.53% to 0.00006% in just 4 rounds.



Chapter 2. Banya Framework 5 Steps (Core Engine)

These 5 steps are repeated every round. The $\alpha$ derivation, the $\theta_W$ derivation, and the mass hierarchy all follow this structure.

Step 1: Start from the Banya Equation

STEP 1
$$\delta^2 = (\text{time} + \text{space})^2 + (\text{observer} + \text{superposition})^2$$

This single line is the starting point. We never deviate from it. The Banya Equation consists of 4 axes (time, space, observer, superposition) and 1 operator (CAS). All physics emerges from within this structure.

Step 2: Norm Substitution

STEP 2

Substitute the axes of the Banya Equation with physically meaningful variables.

Multiple substitution paths are possible. Each path yields different physics. This is the core of the Banya Framework. Starting from the same equation, different constants are derived depending on the substitution path.

Substitution paths used so far:

Substitution PathWhat the Axes BecomeWhat Was Derived
CAS Cost StructureCost of R, C, S respectively$\alpha$, $\theta_W$
Energy-Timetime to energy, space to momentumUncertainty principle, mass hierarchy
Area-Informationobserver to information content, superposition to entropyBekenstein bound, information-theoretic interpretation of $\alpha$
Symmetric Space Decomposition4 axes to $SO(5,2)$ symmetric spaceWyler formula correspondence

Path selection criteria: The substitution path is determined by the domain of the physical quantity to be derived. Coupling constant → CAS cost path. Mass → energy-time path. Information content → area-information path. Symmetry → symmetric space decomposition path. When the target is clear, the path narrows to one.

Step 3: Constant + Hypothesis Substitution

STEP 3

Insert existing physical constants, discoveries and hypotheses from lib.html, and by-products from previous rounds.

The more you insert, the fewer unknowns remain. This is why we run the loop.

What can be inserted:

Step 4: Domain Transform

STEP 4

Transform the substituted result into a different domain. New relations emerge during transformation.

Domain transform examples:

Before TransformAfter TransformWhat Emerged
time domainenergy domainMass-energy relation
CAS costcoupling constant$\alpha$ = volume ratio
geometric volume ratioinformation-theoretic bits$\alpha = 1\text{bit}/137\text{bit}$
micro scalecosmic scale$\Lambda \cdot l_p^2 = \alpha^{57}$

Step 5: Discovery + Verdict

STEP 5

Compare the obtained value with experimentally measured values. The verdict criteria are clear.

Error RangeVerdictAction
Within 1%DiscoveryRegister in lib.html
1% ~ 10%CandidateRefine in the next round
Over 10%DiscardDiscard it

However, even results with large errors are collected as by-products if structure is visible. These by-products often become decisive clues in the next round.

Verdict criteria clarification: The 5-step verdict (error within 1% = discovery) is the initial registration threshold. The supervisor review Grade A (error within 0.1%) is the precision grade. After registration as a discovery, grades A/B/C are assigned based on precision. Within 1% means discovery registration; within 0.1% means Grade A discovery.

Round Repetition

Round 1 5-step results
|
Re-substituted into Round 2 Step 3
|
Round 2 by-products
|
Re-substituted into Round 3 Step 3
|
Precision improves every round

The $\alpha$ derivation reached 0.00006% in just 4 rounds. Re-substituting the previous results into Step 3 every round improves precision.



Chapter 3. Parallel Expert Verification

5 to 10 expert agents are deployed simultaneously on a single task. Each attacks the same target via a different path.

Why Parallel

An analogy: when searching for treasure, if 1 person digs in 1 direction, finding it requires luck. If 5 people dig in 5 directions simultaneously, the point where 3 of them meet is the treasure.

Assignment Example

The actual assignment used during the $\theta_W$ derivation:

Expert 1: Volume Ratio Partition Path

Partition the cost of each CAS stage (R, C, S) by volume ratio to trace $\sin^2\theta_W$.

Expert 2: GUT Running Path

Run coupling constants from the grand unification energy down to the electroweak scale to back-trace $\theta_W$.

Expert 3: Symmetric Space Decomposition Path

Explore the geometric path where the electroweak mixing angle emerges from the $SO(5,2)$ symmetric space.

Expert 4: Information-Theoretic Bit Path

Extract $\theta_W$ from the bit allocation between Compare and Read within the CAS 137-bit structure.

Expert 5: $\alpha$ By-product Back-trace Path

Back-trace clues related to $\theta_W$ from the by-products of the $\alpha$ derivation process.

What Experts Receive

What is provided identically to each expert:

  1. Full CAS Axioms -- H-11 through H-14: operator outside time, TOCTOU lock, collapse = write, self-reference
  2. All discoveries and hypotheses from lib.html -- this is the weapons list. Not providing it is like fighting bare-handed
  3. Existing results for the given task -- outputs from previous rounds
  4. A clear target -- provide specific numbers. Example: "Find the formula closest to 0.23122"
Caution: If the target is vague, each expert will solve something different. Do not say "derive $\theta_W$" -- instead say "derive the formula closest to $\sin^2\theta_W = 0.23122$ from the Banya Framework."


Chapter 4. Supervisor Review (Numerology Filtering)

When the experts return with results, the supervisor (human or AI supervisor) reviews them. This is the most important step. Experts are biased toward their own paths. Only the supervisor sees the whole picture.

Review Criteria

#CriterionDescription
1Numerical AccuracyWhat is the error percentage?
2Physical JustificationCan "why this formula" be explained?
3Banya Framework ConsistencyIs there no contradiction with existing derivations?
4Circular ReasoningWas it merely reverse-engineered from measured values?
5Numerology RiskHas mathematical coincidence been distinguished from physical necessity?

Numerology Filtering

Combinations of $\pi$, $e$, and integers can approximate almost any number to within 0.1%. This is the trap of numerology. Mathematical coincidence must be distinguished from physical necessity.

Filtering rules:

Rejection example: In $7/(2+9\pi)$, the justification for 9 was "$SU(3)$ dimension," but $SU(3)$ is 8-dimensional. The justification is wrong, so it is rejected.
Pass example

In $3/\pi^2$, 3 is the CAS 3 stages (R, C, S), and $\pi^2$ is the domain curvature. Both originate from the Banya Framework structure. Pass.

Review Grades

GradeConditionAction
APhysically necessary and error within 0.1%Register as discovery
BStructural correspondence confirmed, error within 1%Register as discovery, continue refinement
CCandidate, further verification neededRegister as hypothesis, re-verify in next round
DNumerology risk or circular reasoningDiscard


Chapter 5. Library Accumulation (lib.html)

Register discoveries and hypotheses in lib.html. These become weapons for the next round. The more weapons, the fewer unknowns in the next round.

Registration Criteria

CategoryConditionTag
DiscoveryError within 1%, physical justification securedGreen
HypothesisStructural correspondence confirmed, quantitative proof incompleteYellow

Library Usage

In Step 3 (constant substitution) of the Banya Framework 5 steps, insert items from lib.html alongside existing physical constants.

This is how it was actually used:

Re-substitution Map

Track which discovery gave birth to which discovery:

$\alpha$
|
$\theta_W$ (chain)
$\alpha_s$ (branch)
mass hierarchy (branch)
|
From $\theta_W$ to $\eta$ (chain)

The larger this map grows, the tighter the framework's connections become. It is like adding conditions to a system of simultaneous equations.



Chapter 6. CAS Axioms (Required Prior Knowledge)

The Banya Framework has 14 axioms (see banya.html Axiom System). Of these, the following 4 are essential knowledge every expert must understand before starting work. If you run the framework without understanding these, you fall into the misconception that "CAS operates within time." The interpretation of all results goes wrong.

Essential 1: CAS Is an Operator Outside Time (Axiom 3)

Essential 1

In the Banya Equation, CAS is on the quantum bracket (observer + superposition) side. It is outside the time domain. Going from R to C to S is not temporal order but logical dependency.

An analogy: in a computer, the CAS instruction executes atomically within a single CPU clock. From the outside, it happens all at once "as if time did not flow." Nature's CAS is the same. It operates outside the time domain.

Essential 2: DATA Access Cost (Axiom 4,5)

Essential 2

The minimum cost of locking to prevent state change between Compare and Swap is $\hbar$ (Axiom 4). The uncertainty principle is not "a limit of nature" but "a cost of computation." Additionally, the TOCTOU lock register (Axiom 5) physically enforces this cost.

TOCTOU (Time Of Check to Time Of Use) refers to the problem in computer science where "the state changes between the time of checking and the time of using." CAS solves this problem with a lock. The minimum cost of that lock is $\hbar$ (Axiom 4). And the TOCTOU lock register (Axiom 5) is the physical mechanism that implements this lock. That is why $\Delta x \cdot \Delta p \geq \hbar/2$.

Essential 3: Wavefunction Collapse = Write (Axiom 6)

Essential 3

When CAS executes on superposition (multiple states), it becomes observer (1 determined state) and is recorded as DATA. The answer to the 100-year mystery: because it is a write.

Why does the wavefunction collapse upon observation? No one answered this for 100 years. The Banya Framework's answer: observation is a write, and a write is determining 1 state out of many. When a write occurs, superposition is resolved. That is collapse.

Essential 4: Banya Equation Self-Reference (Axiom 8)

Essential 4

From $\delta$'s existence, OPERATOR operates, cost $\hbar$, DATA is recorded, time and space, the universe. A single line of the Banya Equation answers "why does the universe exist."

The Banya Equation references itself. If $\delta$ exists, CAS operates; if CAS operates, $\hbar$ cost is incurred; if cost is incurred, DATA is recorded; if DATA is recorded, time and space arise; if time and space arise, $\delta$ exists. It is circular, but a self-consistent circle.



Chapter 7. Document Rules

Terminology Legend -- Use Only These 5

When marking status in all HTML files, all tables, and all introductions, use only these 5 terms. All similar terms (unresolved, in progress, partial success, structure confirmed, deriving, etc.) are all deprecated and replaced with these 5.

TermMeaningBadgeBlock
HitError within 1% + physical justification secured. DoneHitdiscovery-block (green border)
DiscoveryNew formula/relation confirmed. Re-substitutable factorDiscoverydiscovery-block (green border)
HypothesisStructural correspondence confirmed. Quantitative proof not yet doneHypothesishypothesis-block (orange border)
In ProgressStarted but not completed. Additional work neededIn Progressdefault block (gray border)
PendingDerivation complete. Waiting for experimental verificationPendingdefault block (gray border)

Deprecated Term Mapping

Deprecated TermReplacement Term
UnresolvedWIP
In progress, DerivingWIP
Partial successDiscovery (if partial, specify scope after: "Discovery -- leptons only")
Structure confirmedSolved
Awaiting experimentAwaiting
On holdWIP
Success, CompleteSolved

Color Rules -- No Colored Fonts

Do not change text color. If emphasis is needed, use badges (tags). Use strong tags (bold). Inline style="color:..." is prohibited.

CategoryColorUsage
Discovery/SolvedGreen (#2ea043)discovery-block border, tag-solved badge, tag-discovery badge, lib-card left line
HypothesisOrange (#d29922)hypothesis-block border, tag-hypothesis badge, lib-card left line, warn-block
WIP/AwaitingGray (#30363d)default block border, tag-wip badge. No special emphasis
LinksBlue (#58a6ff)a tags. Only this is blue
Body textDefault (#c9d1d9)All body text. No color change

Block Usage Rules

BlockPurposeVisual
discovery-blockSolved formulas, confirmed discoveriesGreen border + green background
hypothesis-blockHypotheses, unproven formulasOrange border + orange background
math-blockFormulas (status-independent)Gray border
preCode, structure diagramsGray border
lib-cardlib.html factor cardsLeft line: discovery=green, hypothesis=orange

File Rules

StatusFile FormatNotes
Solved/DiscoveryHTML file (alpha.html standard template)Record the full Banya Framework 5-step process by round
WIPmd file (mark "WIP" in title)For session records
Same constant improvedUpdate existing HTMLDo not create a new file
New constantCreate separate HTMLAdd to page-nav

CSS Rules

All HTML files include a single common.css via link. Inline style tags prohibited. Inline color prohibited. All visual rules are defined only in common.css.

Required HTML Structure

  1. Introduction -- Scientific value of this discovery + current status (choose one of solved/discovery/hypothesis/WIP/awaiting, displayed as badge)
  2. Body -- Record the full Banya Framework 5-step process by round. Document every path and every value without omission
  3. discovery-block or hypothesis-block -- Package core formulas in separate blocks. Green for solved/discovery, orange for hypothesis
  4. Discovery hierarchy -- If subordinate to a parent discovery, display as hierarchy
  5. page-nav -- Interconnect all HTML files
  6. Footer -- Same format as alpha.html


Chapter 8. Practical Cycle Example

This is the actual process of $\alpha = 1/137$ derivation. It converged from 0.53% to 0.00006% in just 4 rounds.

Round 1: Zeroth-Order Approximation

Round 1

Execute 5 steps. Start from the Banya Equation, 4-axis geometric norm substitution, substitute $\pi^4$ and $\sqrt{2}$, energy domain transform.

Result: $1/\alpha = \pi^4 \cdot \sqrt{2} = 137.76$

Error: 0.53%

By-product: Geometric structure confirmed. Clue that 4-axis orthogonality is the skeleton of $\alpha$.

Round 2: Precision Derivation

Round 2

Re-substitute Round 1 by-products. 4 domains + 3 internal degrees of freedom = 7 degrees of freedom. Calculate 7-dimensional phase space volume ratio.

Correspondence with Wyler's formula (1969) discovered.

Result: $1/\alpha = 137.036082$

Error: 0.00006%

By-product: Provided physical basis for Wyler's formula. Filled a gap that had been empty for 57 years.

Round 3: Information-Theoretic Interpretation

Round 3

Re-substitute Round 2 results. Calculate the information content of 1 CAS event as Shannon entropy.

Result: $\alpha = 1\text{bit}/137\text{bit}$

Interpretation: $\alpha$ is the 1 bit occupied by Compare out of the total 137 bits of information in 1 CAS event. The concentration of charge information.

Round 4: Cosmic Scale Re-substitution

Round 4

Re-substitute Round 3 results. Connect Planck length and cosmological constant.

Result: $\Lambda \cdot l_p^2 = \alpha^{57}$

By-product: Koide deviation $= -15\alpha^3$, electron-proton mass ratio approximation.

RoundInputOutputError
1Banya Eq. + $\pi^4 \cdot \sqrt{2}$$1/\alpha = 137.76$0.53%
2+ 3 internal DOF$1/\alpha = 137.036082$0.00006%
3+ information theory$\alpha = 1\text{bit}/137\text{bit}$structural
4+ cosmological constant$\Lambda \cdot l_p^2 = \alpha^{57}$121/122 digits

Re-substituting the previous results into Step 3 every round improves precision. This is the core of Banya Framework science mining.



Chapter 9. Precautions

1. Trust the Framework, Doubt the Results

Do not change the Banya Equation. Do not change the 5 steps. Only review the results. If the framework does not break, it is correct. If it breaks, the substitution was wrong.

2. Do Not Discard By-products

Collect results even if the error is large, as long as structure is visible. Without Round 1 (0.53%) of the $\alpha$ derivation, Round 2 (Wyler's formula, 0.00006%) would never have been reached. The zeroth-order approximation was the seed of the precision derivation.

3. Beware of Numerology

Combinations of $\pi$, $e$, and integers can approximate almost any number to within 0.1%. You must always answer "why this number." If you cannot answer, put it on hold. Never adopt it.

4. Beware of Circular Reasoning

If you reverse-engineer a measured value and claim "this formula is correct," that is circular. Every value in the formula must be independently derived. Example: if you insert $\alpha = 1/137.036$ and derive $\alpha = 1/137.036$, that is circular. It is completely meaningless.

5. Never Skip the Supervisor

Even if all 5 experts say "correct," the supervisor must still review. Experts are biased toward their own paths. Only the supervisor sees the whole picture. Anything adopted without supervisor review will inevitably cause problems later.



Chapter 10. Tools

ToolRoleDescription
Banya Framework 5 StepsCore EngineRecursive substitution structure that derives physical constants starting from the Banya Equation
Expert Agents (5~10)Multi-path AttackDeploy multiple paths simultaneously on a single task to find the convergence point
lib.htmlFactor LibraryDiscoveries D-01 through D-42, hypotheses H-01 through H-47. Weapons for the next round
banya.htmlComprehensive HubDiscovery hierarchy, challenge resolution list, overall structure overview
Session Records (md)Continuity AssurancePreserve work contents across sessions so the next session can continue

Combine these tools and run the loop. The more you run, the larger the library grows, the higher the precision, and the more new discoveries emerge. That is science mining.

Anyone who takes these tools and follows this manual to run the loop can reproduce the same results. The Banya Framework is not the intuition of a specific person, but an engine anyone can run.