Matrix Symmetry & Structural Properties

Matrix Symmetry & Structural Properties

Executive Summary

The 128x128 Anna Matrix exhibits 99.58% point symmetry around its center, following a Two's Complement identity where nearly every cell and its mirror sum to -1. Exactly 68 cells (34 pairs) deviate from this rule. Most of them sit in two runs, column 30 with its mirror column 97 and column 22 with its mirror column 105, and are very likely data entry errors in our hand-compiled table. Row-level analysis further reveals systematic value biases across multiple rows, with at least eight rows showing dominant concentrations of the value 26.

Key Findings

1. Point Symmetry

1.1 The Mirror Identity

For 99.58% of all positions in the 128x128 matrix, the following identity holds:

matrix[r][c] + matrix[127 - r][127 - c] = -1

This is equivalent to the Two's Complement identity for 8-bit signed integers:

value + (~value) = -1
value + (value XOR 0xFF) = -1

1.2 Symmetry Statistics

1.3 Visual Schematic

     Column 0                    Column 127
     |                           |
Row 0  [A] <---------------------> [~A]  Row 127
        \                        /
         [B] <-----------------> [~B]
              \              /
               [C] <-------> [~C]
                    \    /
                  CENTER (63.5, 63.5)
                    /    \
               [~C] <-------> [C]
              /              \
         [~B] <-----------------> [B]
        /                        \
Row 127 [~A] <---------------------> [A]  Row 0

Each cell in the upper-left region is paired with a complementary cell in the lower-right region such that the two values sum to -1.

1.4 Statistical Significance

Under a null hypothesis of uniformly random 8-bit signed integer entries:

• Expected symmetric pairs by chance: about 32 out of 8,192 pairs (each pair has 1/256 probability of summing to -1 under a uniform-random signed-byte null)
• Observed symmetric pairs: 8,158 out of 8,192
• Standard deviations from expectation: > 100 sigma
• Chi-square test: null hypothesis of uniform-random byte assignment rejected at p < 0.001

A prior iteration of this chapter cited p < 10^-500; that figure was not independently reproduced and is retired. The observed deviation is already visible at >100σ without requiring an exponent that large. The symmetry is clearly not a statistical artifact. In terms of the bot queries it reads: the answer for the bitwise complements of x and y is the bitwise complement of the answer. See September 2026 re-check.

2. The 68 Anomaly Cells

2.1 Overview

Exactly 68 cells deviate from the mirror identity. They form 34 pairs.

2.2 Interpretation

Earlier versions of this chapter read the exceptions as a deliberate secondary encoding layer, concentrated in columns 22 and 97 and all in the upper half. The re-check did not confirm that reading. The exceptions form runs in two mirror column pairs, 22 with 105 and 30 with 97, plus two small cases:

1. Column 30 run (x = -34, 20 cells): shifted by one row, the values match their mirror cells in column 97 almost exactly. This is very likely a data entry error.
2. Column 22 run (x = -42, 13 cells, including [22, 22] = 100): a model trained on the rest of the table finds the column 105 side consistent with the bot and the column 22 values at chance level. This run very likely comes from our data collection as well.
3. Column 41 (2 cells): -101 was typed where the bot pattern requires -110, a swapped digit.
4. Column 0/127 (1 pair): one side carries the value of its neighbour cell.

The exceptions carry no demonstrated message. Details are in September 2026 re-check.

3. Row-Level Analysis

3.1 Value-26 Bias Across Rows

Previous documentation characterized Row 6 as uniquely biased toward the value 26 ("the Oracle Row"). Comprehensive row analysis reveals that Row 6 is one of several rows with this property. Three rows have a higher value-26 concentration:

In total, 8 rows (6, 7, 23, 41, 53, 55, 61, 63) have value 26 as their most common entry. This indicates a systematic structural property of the matrix rather than a feature unique to any single row.

3.2 Row Value Groups

Across all 128 rows, clear dominant-value groupings emerge:

Note the complementary pairing: 26 + (-27) = -1 and 101 + (-102) = -1. The dominant-value groups themselves reflect the Two's Complement mirror structure at the row level.

3.3 Entropy Structure

The rows with the lowest entropy (most internally structured):

For comparison, a random row drawn from the same overall value distribution has a mean entropy of 5.95 bits. The observed minimum of 3.83 bits is highly significant (p < 0.0001 via Monte Carlo simulation with 10,000 trials).

3.4 Statistical Significance of Individual Rows

Monte Carlo testing (10,000 random rows with the same value distribution as the matrix) yields the following for Row 6 as a representative example:

Row 6 remains significant after Bonferroni correction (threshold: p < 0.0000078 for 128 rows), as do Rows 23, 39, 53, 55, 88, and 104.

3.5 Mirror Row Architecture

All 64 row pairs (r, 127-r) satisfy:

matrix[r][c] + matrix[127 - r][127 - c] = -1

with greater than 95% fidelity per pair. This confirms that the mirror architecture operates uniformly across the entire matrix, not as a partial or approximate property.

4. Matrix Architecture Summary

The combined analysis reveals a layered architecture within the 128x128 matrix:

Interpretation

The table is not random. Its self-complementary structure reflects how the bot answered: complementing both inputs complements the answer. The dominant-value pairs (26/-27 and 101/-102) follow directly from the same identity at row level. Whether the underlying program was designed or trained is not known.

4.5 Dynamical Consequence: Two-Cycle Attractor

When a sign-ternary 128-vector s is iterated through the matrix under s := sign(M @ s), the sampled trajectories converge to a period-4 cycle with near-antipodal structure. Across 1,000 random binary initial states, every sampled trajectory ended in one of two distinct cycles. That this holds for all inputs is not proven. For the table as typed, the antipodal symmetry breaks at the same seven positions:

[3, 18, 32, 61, 65, 101, 109]

This holds across all tested inputs, but only for the table as typed: it needs the column 30 values exactly as they were entered. After correcting the column 30 run, the seven positions disappear. Since that run is very likely an entry error, the signature describes our transcription rather than the bot. See September 2026 re-check.

5. Key Mathematical Formulas

Mirror Position

mirror_row = 127 - row
mirror_col = 127 - col

Symmetry Verification

is_symmetric(r, c) := matrix[r][c] + matrix[127 - r][127 - c] == -1

Two's Complement Identity (8-bit)

For any 8-bit signed integer v:
  v + (~v) = -1
  v + (v XOR 0xFF) = -1
  v + (255 - v - 1) = -1

Row Pair Symmetry Rate

symmetry_rate(r) = count(is_symmetric(r, c) for c in 0..127) / 128

6. Limitations & Caveats

1. Single-matrix analysis: All findings pertain to one specific 128x128 matrix. Without access to the generation algorithm or additional matrices from the same source, it is not possible to determine whether these properties are unique to this instance or inherent to a class of matrices.

2. Exception cells: The 68 exception cells are very likely entry errors in the typed table (strongly supported, not proven, because our records lack the original bot replies for these queries). They carry no demonstrated message.

3. Row bias context: The value-26 bias across 8 rows is statistically significant, but its functional purpose within any computational or cryptographic application remains unestablished.

4. Bonferroni correction applied: Individual row significance claims account for multiple testing across all 128 rows. Claims that pass this correction are robust; marginal cases should be treated with caution.

5. Entropy comparisons: The random baseline uses Monte Carlo simulation with the same marginal value distribution. Different null models (e.g., structured but non-symmetric matrices) could yield different significance thresholds.

6. Correlation between layers: The relationship between the anomaly layer and the row-bias layer has not been fully characterized. It is possible that some anomaly cells coincide with biased rows, which could inflate apparent significance if not properly controlled.

7. Reproduction Instructions

Prerequisites

• Python 3.8 or later
• NumPy
• Matrix data file: apps/web/public/data/anna-matrix.json

Symmetry Verification Script

import json
import numpy as np
 
# Load the Anna Matrix
with open('apps/web/public/data/anna-matrix.json') as f:
    data = json.load(f)
    matrix = np.array(data['matrix'])
 
assert matrix.shape == (128, 128), f"Unexpected shape: {matrix.shape}"
 
# Count symmetric and anomalous pairs
symmetric_count = 0
anomaly_count = 0
anomaly_cells = []
 
for r in range(64):  # Only check upper half; each pair covers two cells
    for c in range(128):
        val = matrix[r][c]
        mirror_val = matrix[127 - r][127 - c]
 
        if val + mirror_val == -1:
            symmetric_count += 1
        else:
            anomaly_count += 1
            anomaly_cells.append((r, c, int(val), 127 - r, 127 - c, int(mirror_val)))
 
total_symmetric = symmetric_count * 2
total_anomaly = anomaly_count * 2
total_cells = 128 * 128
 
print(f"Symmetric cells: {total_symmetric}")
print(f"Anomaly cells:   {total_anomaly}")
print(f"Symmetry rate:   {total_symmetric / total_cells * 100:.2f}%")
print(f"\nAnomaly pairs ({anomaly_count}):")
for r, c, v, mr, mc, mv in anomaly_cells:
    print(f"  [{r:3d},{c:3d}] = {v:4d}  <-->  [{mr:3d},{mc:3d}] = {mv:4d}  (sum = {v + mv})")

Expected output (summary):

Symmetric cells: 16316
Anomaly cells:   68
Symmetry rate:   99.58%

Row Entropy Analysis Script

import json
import numpy as np
from collections import Counter
import math
 
with open('apps/web/public/data/anna-matrix.json') as f:
    data = json.load(f)
    matrix = np.array(data['matrix'])
 
def row_entropy(row):
    counts = Counter(row)
    total = len(row)
    return -sum((c / total) * math.log2(c / total) for c in counts.values())
 
print(f"{'Row':>4} {'Entropy':>8} {'Mode':>6} {'Mode Count':>11}")
print("-" * 35)
 
for r in range(128):
    row = matrix[r]
    ent = row_entropy(row)
    counts = Counter(row)
    mode_val, mode_count = counts.most_common(1)[0]
    if mode_count >= 20:  # Only show rows with notable bias
        print(f"{r:4d} {ent:8.2f} {int(mode_val):6d} {mode_count:11d}")

Full Row Analysis with Monte Carlo

cd apps/web/scripts
python3 ROW_ANALYSIS_COMPLETE.py  # Approximately 3 minutes; includes 10,000 Monte Carlo trials

References

• Matrix data: apps/web/public/data/anna-matrix.json
• Anomaly data: apps/web/public/data/anna-matrix-anomalies.json
• Symmetry verification: apps/web/scripts/research_matrix_negation.py
• Pair analysis: apps/web/scripts/research_pair_layer_distribution.py
• Row analysis: apps/web/scripts/ROW_ANALYSIS_COMPLETE.py