Structured Numeric Compression  ·  100-Run Verified Benchmark  ·  Sep 2026
AIBH Gravitational Infall  ·  Shannon Entropy Zones  ·  Packed Float32 Binary Output
1208×
Mean compression ratio
vs 1.94× zlib  ·  15 datasets  ·  Python 3.12 Apple M-series  ·  verified Sep 2026
12,500×Peak ratio
125×Minimum ratio
0.035msp50 latency
64 bytesOutput size
FREEAPI access
↓ explore the data
What UUON CFE-A02 actually does
The model
Gravitational Field Zones
Input data is partitioned into gravitational zones via AIBH infall dynamics. Each zone collapses to a centroid — the mathematical center of mass of that region of the field.
The output
Packed Float32 Binary
8 zones produces 16 floats — 64 bytes fixed, regardless of input size. 800,000 bytes encodes to 64 bytes. Shannon entropy H returned per zone on every call.
The tradeoff
Lossy Structural
Not byte-for-byte lossless. Approximate reconstruction from centroids. Use zlib or zstd for exact recovery. Use UUON CFE for field structure transmission, telemetry, ML features, and IoT streams.
15 datasets · 100 runs each · all results verified · Sep 2026

Select a dataset

Compression ratio — UUON CFE-A02 vs zlib
125×Binary ratio
64.8×vs zlib adv.
0.035msp50 latency
0.039msp99 latency
DatasetNRatiovs zlibp50
Full results — all datasets · all metrics
DatasetNOrig bytesUUON bytesRatiovs zlib adv.p50 msp99 msmin ms
Sine Wave 1K1,0008,00064125.0x64.8x0.03510.03940.0312
Sine Wave 10K10,00080,000641,250.0x637.8x0.29480.31300.2710
Sine Wave 100K100,000800,0006412,500.0x6,345.2x3.03233.30712.9100
Cosine Mix 1K1,0008,00064125.0x64.4x0.03400.04200.0311
Random Uniform 1K1,0008,00064125.0x64.8x0.03280.03840.0299
Random Uniform 10K10,00080,000641,250.0x637.8x0.30200.32000.2780
Gaussian 1K1,0008,00064125.0x64.4x0.03260.04230.0298
Gaussian 10K10,00080,000641,250.0x634.5x0.29630.32880.2720
Spiky Chaos 1K1,0008,00064125.0x64.4x0.03260.04350.0299
Constant 1K1,0008,00064125.0x1.3x0.03430.03890.0310
Linear Ramp 1K1,0008,00064125.0x34.5x0.03350.04220.0305
Step Function 1K1,0008,00064125.0x1.7x0.03370.04440.0308
Sensor Telemetry5,00040,00064625.0x297.6x0.15290.17970.1410
Financial Series1,0008,00064125.0x64.1x0.03340.03510.0305
Extreme Values1,0008,00064125.0x1.5x0.03350.03940.0308
Methodology: 100 iterations per dataset, Python 3.12, Apple M-series, local execution.
UUON output: 8 zones x (centroid + max_deviation) x float32 = 64 bytes fixed, regardless of input size.
zlib baseline: level-1 compression of string-encoded float64 array.
Compression class: Lossy structural. Not lossless. Approximate reconstruction from zone centroids only.
Advantage column: UUON binary ratio divided by zlib ratio. Different compression classes — not a direct equivalence.
Interactive · calls compress.uuon.world live · demo key included

Test the engine

Input — paste numbers or load a sample
Stress test — concurrent calls to live Fargate
// Click to fire N concurrent calls against the live endpoint
Output — AIBH result from live API
// Select a sample dataset above // or enter comma-separated numbers // then click Compress // // Returns: // compression_ratio_b64 actual byte ratio // H_total Shannon entropy // H_per_zone entropy per zone // compressed_b64 packed float32 binary // compressed_bytes_b64 output size in bytes
Copy · paste · run · no SDK required

One endpoint. Any language.

# Step 1 - register for your free key
curl -X POST http://compress.uuon.world/register \
  -H "Content-Type: application/json" \
  -d '{"email": "you@example.com"}'

# Step 2 - compress (binary endpoint - up to 12500x ratio)
curl -X POST http://compress.uuon.world/algorithms/compression-field/compress-binary \
  -H "Content-Type: application/json" \
  -H "x-api-key: YOUR_UUON_KEY" \
  -d '{
    "data": [1.2, 3.4, 5.6, 7.8, 9.0, 2.1, 4.3, 6.5, 8.7, 0.9],
    "zone_count": 8
  }'

# Response:
#   compressed_b64          base64 packed float32 binary (64 bytes)
#   compression_ratio_b64   actual byte compression ratio
#   H_total                 Shannon field entropy
#   H_per_zone              entropy breakdown per zone
#   original_bytes          input size in bytes
#   compressed_bytes_b64    output size in bytes
import requests, base64, struct, math

API_KEY = "your_uuon_key"
BASE    = "http://compress.uuon.world"

data = [math.sin(i * 0.1) * 50 + 50 for i in range(1000)]

r = requests.post(
    f"{BASE}/algorithms/compression-field/compress-binary",
    headers={"x-api-key": API_KEY},
    json={"data": data, "zone_count": 8}
).json()

o = r["output"]
print(f"Compression ratio : {o['compression_ratio_b64']}x")
print(f"Shannon entropy H : {o['H_total']}")
print(f"Original size     : {o['original_bytes']} bytes")
print(f"Compressed size   : {o['compressed_bytes_b64']} bytes")

# Decode the packed binary
raw    = base64.b64decode(o["compressed_b64"])
floats = struct.unpack(f"{len(raw)//4}f", raw)
zones  = [(floats[i], floats[i+1]) for i in range(0, len(floats), 2)]
for i, (centroid, deviation) in enumerate(zones):
    print(f"Zone {i}: centroid={centroid:.4f}  max_dev={deviation:.4f}")
const KEY  = 'your_uuon_key';
const BASE = 'http://compress.uuon.world';

const data = Array.from({length: 1000}, (_, i) => Math.sin(i * 0.1) * 50 + 50);

const res = await fetch(`${BASE}/algorithms/compression-field/compress-binary`, {
  method: 'POST',
  headers: { 'Content-Type': 'application/json', 'x-api-key': KEY },
  body: JSON.stringify({ data, zone_count: 8 })
});

const { output: o } = await res.json();
console.log(`Ratio  : ${o.compression_ratio_b64}x`);
console.log(`Entropy: ${o.H_total}`);
console.log(`Size   : ${o.original_bytes} to ${o.compressed_bytes_b64} bytes`);

// Decode binary in browser
const raw  = atob(o.compressed_b64);
const buf  = new ArrayBuffer(raw.length);
const u8   = new Uint8Array(buf);
for (let i = 0; i < raw.length; i++) u8[i] = raw.charCodeAt(i);
const f32  = new Float32Array(buf);
// f32 = [centroid_z0, maxdev_z0, centroid_z1, maxdev_z1, ...]
package main

import (
    "bytes"
    "encoding/base64"
    "encoding/binary"
    "encoding/json"
    "fmt"
    "math"
    "net/http"
)

const apiKey = "your_uuon_key"
const base   = "http://compress.uuon.world"

func main() {
    data := make([]float64, 1000)
    for i := range data {
        data[i] = math.Sin(float64(i)*0.1)*50 + 50
    }
    body, _ := json.Marshal(map[string]interface{}{
        "data": data, "zone_count": 8,
    })
    req, _ := http.NewRequest("POST",
        base+"/algorithms/compression-field/compress-binary",
        bytes.NewBuffer(body))
    req.Header.Set("Content-Type", "application/json")
    req.Header.Set("x-api-key", apiKey)

    resp, _ := http.DefaultClient.Do(req)
    defer resp.Body.Close()

    var result map[string]interface{}
    json.NewDecoder(resp.Body).Decode(&result)
    o := result["output"].(map[string]interface{})
    fmt.Printf("Ratio  : %vx\n", o["compression_ratio_b64"])
    fmt.Printf("Entropy: %v\n", o["H_total"])

    raw, _ := base64.StdEncoding.DecodeString(o["compressed_b64"].(string))
    for i := 0; i < len(raw)/4; i += 2 {
        c := math.Float32frombits(binary.LittleEndian.Uint32(raw[i*4:]))
        d := math.Float32frombits(binary.LittleEndian.Uint32(raw[(i+1)*4:]))
        fmt.Printf("Zone %d: centroid=%.4f  max_dev=%.4f\n", i/2, c, d)
    }
}
use base64::{engine::general_purpose, Engine};
use reqwest::Client;
use serde_json::json;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let client = Client::new();
    let data: Vec<f64> = (0..1000)
        .map(|i| (i as f64 * 0.1).sin() * 50.0 + 50.0)
        .collect();

    let res: serde_json::Value = client
        .post("http://compress.uuon.world/algorithms/compression-field/compress-binary")
        .header("x-api-key", "your_uuon_key")
        .json(&json!({ "data": data, "zone_count": 8 }))
        .send().await?.json().await?;

    let o = &res["output"];
    println!("Ratio  : {}x", o["compression_ratio_b64"]);
    println!("Entropy: {}",  o["H_total"]);

    let raw = general_purpose::STANDARD.decode(o["compressed_b64"].as_str().unwrap())?;
    let floats: Vec<f32> = raw.chunks(4)
        .map(|c| f32::from_le_bytes(c.try_into().unwrap()))
        .collect();
    for (i, chunk) in floats.chunks(2).enumerate() {
        println!("Zone {}: centroid={:.4}  max_dev={:.4}", i, chunk[0], chunk[1]);
    }
    Ok(())
}
import fetch from 'node-fetch'; // npm install node-fetch

const KEY  = 'your_uuon_key';
const BASE = 'http://compress.uuon.world';

const data = Array.from({length: 1000}, (_, i) => Math.sin(i * 0.1) * 50 + 50);

const res = await fetch(`${BASE}/algorithms/compression-field/compress-binary`, {
  method: 'POST',
  headers: { 'Content-Type': 'application/json', 'x-api-key': KEY },
  body: JSON.stringify({ data, zone_count: 8 })
});

const { output: o } = await res.json();
console.log(`Ratio  : ${o.compression_ratio_b64}x`);
console.log(`Entropy: ${o.H_total}`);

const buf    = Buffer.from(o.compressed_b64, 'base64');
const floats = [];
for (let i = 0; i < buf.length; i += 4) floats.push(buf.readFloatLE(i));
floats.forEach((v, i) => {
  if (i % 2 === 0) process.stdout.write(`Zone ${i/2}: centroid=${v.toFixed(4)}`);
  else console.log(`  max_dev=${v.toFixed(4)}`);
});
library(httr)
library(jsonlite)
library(base64enc)

API_KEY <- "your_uuon_key"
data    <- sin(seq(0, 99.9, by=0.1)) * 50 + 50

res <- POST(
  "http://compress.uuon.world/algorithms/compression-field/compress-binary",
  add_headers("x-api-key" = API_KEY, "Content-Type" = "application/json"),
  body = toJSON(list(data = data, zone_count = 8L), auto_unbox = TRUE)
)

o <- content(res)$output
cat("Ratio  :", o$compression_ratio_b64, "x\n")
cat("Entropy:", o$H_total, "\n")

raw    <- base64decode(o$compressed_b64)
con    <- rawConnection(raw)
floats <- readBin(con, "numeric", n = length(raw)/4, size = 4, endian = "little")
close(con)
matrix(floats, ncol=2, byrow=TRUE, dimnames=list(NULL, c("centroid","max_dev")))
Free · no card · no expiry · instant

Get your
API key

One email. One key. Immediate access to the compression endpoint. Usage tracked per key from the first call.

  • 125x to 12,500x compression on numeric data
  • Shannon entropy H on every response
  • Zone-level gravitational field analysis
  • Packed float32 binary decodable in any language
  • REST JSON in / binary out / no SDK needed
  • Usage logged and visible via admin endpoint
  • Valid permanently
Your API Key — click to copy
Click to copy · store safely · never share