The Collatz Entropy Project: A Computational Proof of Inevitable Collapse

"Mathematics may not be ready for such problems." — Paul Erdős

"Computer Science is." — This Repository / Swen Kalski

1. Overview

The Collatz Conjecture ( $3n+1$ ) has baffled mathematicians for decades. Traditional approaches often get lost in the chaotic behavior of individual trajectories. This project takes a fundamentally different approach: Information Theory and Binary Mechanics.

Instead of asking "Does it reach 1?", we ask: "Does the system gain or lose information over time?"

By analyzing the binary structure of the operations, we have demonstrated that the Collatz system is dissipative. It is an entropy-destroying process with a specific, measurable negative drift. Infinite growth is not just statistically unlikely; it is structurally impossible.

2. The Three Pillars of evidence

This repository contains three distinct layers of evidence that, when combined, hopefully, close the lid on the conjecture.

Phase 1: The Binary Straitjacket (Structural Proof)

See: PAPER_1.md | Code: collatz_straitjacket.cpp

My experiments suggest that a "Monster" (a number that ascends infinitely without crashing) requires a specific binary structure: it must consist of an infinite sequence of trailing ones (...11111).

• The Mechanism: The operation $3n+1$ acts as a "bit-mixer." It forces random binary cliffs.
• The Limit: To survive $k$ steps without a major crash (division by 4 or more), a number must end in $k$ ones.
• The Contradiction: Since all natural numbers have a finite bit-length, no number can sustain an infinite ascent. The "straitjacket" eventually tightens, forcing a collapse.

Phase 2: The Entropic Barrier

See: PAPER_2.md | Code: collatz_loop_breaker.cpp

I demonstrated that non-trivial closed cycles (loops) are statistically impossible because the system cannot conserve the "information energy" required to close a circle. What becomes a strong evidence.

• The Loop Condition: To form a closed loop, the information gained by multiplication ($3^k$) must exactly match the information lost by division ($2^m$). This requires a specific division ratio of $\log_2(3) \approx 1.585$ .
• The Drift Gap: Our experiments prove the system enforces a natural division ratio of $\approx 2.00$ . This creates a persistent "Energy Debt" of 0.415 bits per step.
• The Impossibility: The probability of a trajectory defying this drift to close a large loop decreases exponentially. For large $N$, the required deviation is statistically indistinguishable from zero ($> 1000\sigma$).

Phase 3: Digital Gravity (Statistical Proof)

See: PAPER_3.md | Code: collatz_entropy.cpp

I measured the "thermodynamics" of the system over millions of trajectories.

• Ascent Gain: $\log_2(3) \approx 1.585$ bits of information.
• Descent Loss: Experimentally proven average of $\approx 2.00$ bits lost per step.
• Net Drift: $-0.415$ bits per step.

This might confirms that the system has "Digital Gravity." The house edge is approx. 20%. Any upward trajectory is merely a temporary fluctuation against a relentless downward pull.

Phase 3: The Death Line (Visual Proof)

See: collatz_gravity_well.png | Code: plot_gravity.py

The image above visualizes the "Logarithmic War."

• The Y-Axis: Information content (Bits / $\log_2$ ).
• The Jagged Lines: Real numbers fighting for altitude.
• The Red Dashed Line: The calculated entropy drift ( $-0.415$ bits/step).

Conclusion of Evidence: No matter how high a number starts (even Mersenne numbers like $2^{50}-1$ ), it cannot escape the Entropy Decay Zone. The red line represents the speed of information destruction, which is mathematically faster than the speed of information creation in the $3n+1$ system.

3. Repository Structure

• PAPER_1.md
• PAPER_2.md
• PAPER_3.md
• src/collatz_bit_mechanic.cpp - Tool to analyze individual binary steps and carries.
• src/collatz_entropy.cpp - Tool to measure the 2.00 vs 1.585 drift ratio on massive datasets.
• src/collatz_straitjacket.cpp - The "Monster Hunter" algorithm.
• scripts/plot_gravity.py - Python script to generate the visualization.

4. How to Reproduce

Prerequisites

• C++ Compiler (g++ or clang)
• Python 3 + Matplotlib (for visualization)

Running the Entropy Analysis

g++ -o collatz_entropy src/collatz_entropy.cpp
./collatz_entropy
# Output will confirm the -0.415 drift

Generating the Death Line Graph

pip install matplotlib numpy
python scripts/plot_gravity.py

5. Conclusion of the experiment (what might be a Proof)

The Collatz Conjecture is not a mystery of randomness; it is a predictable consequence of binary arithmetic.

• Structure: Infinite growth requires infinite bits (impossible).
• Statistics: The descent is stronger than the ascent ( $-0.415$ drift).
• Dynamics: Variance allows temporary peaks, but gravity guarantees the final collapse.

Authored by Swen Kalski 2026