Hawkynt

Context Tree Weighting (Willems)

Genuine Context Tree Weighting (Willems/Shtarkov/Tjalkens): a depth-16 binary context tree with a Krichevsky-Trofimov estimator per node, recursively weighted between each node’s own estimate and the product of its children, driving a binary arithmetic coder. Distinct from β€œContext Predictor (order-2/1/0)” in ctw.js, which despite its historic filename is an unrelated most-frequent-symbol predictor.

Properties

Property Value
Category Compression Algorithms
Sub-category Context Mixing
Security status πŸŽ“ Educational Only
Complexity Advanced
Inventor Frans Willems, Yuri Shtarkov, Tjalling Tjalkens
Year 1995
Origin πŸ‡³πŸ‡± Netherlands
Source algorithms/compression/ctw-willems.js

Security

Status: πŸŽ“ Educational Only

No vulnerabilities are recorded for this implementation.

Documentation

References

Test vectors

7 vectors ship with this algorithm and run in the test suite. Byte values are hexadecimal.

Vector 1 β€” Empty input

Field Value
input (empty)
expected 00000000

Vector 2 β€” Single byte literal

Field Value
input 41
expected 010000006280

Vector 3 β€” Two bytes

Field Value
input 4142
expected 0200000062cf00

Vector 4 β€” Text with no repetition

Field Value
input 48656c6c6f20576f726c64
expected 0b00000066312f2a8b787dd38a2b3780

Vector 5 β€” Repeated text sample (4x)

Field Value
input 74686520717569636b2062726f776e20 666f78206a756d7073206f7665722074 6865206c617a7920646f672e20746865 20717569636b2062726f776e20666f78 206a756d7073206f7665722074686520 6c617a7920646f672e20746865207175 69636b2062726f776e20666f78206a75 6d7073206f76657220746865206c617a 7920646f672e2074686520717569636b 2062726f776e20666f78206a756d7073 206f76657220746865206c617a792064 6f672e20
expected b400000077578191c1ef6951402bbff4 0fe06e7fad93efe76eb8f98e1d9de8bb 6a368862d7cdf641fdba96d3defdd470 4484af640788b71c09c0defd04d4ebfe 1e479d281c4c77864b08fd5b07143359 2a699635b9714ef58ca147a1bdb6360a 763ad1d9a0d36fd82886b218

Vector 6 β€” 256 repeated bytes

Field Value
input 61616161616161616161616161616161 61616161616161616161616161616161 61616161616161616161616161616161 61616161616161616161616161616161 61616161616161616161616161616161 61616161616161616161616161616161 61616161616161616161616161616161 61616161616161616161616161616161 61616161616161616161616161616161 61616161616161616161616161616161 61616161616161616161616161616161 61616161616161616161616161616161 61616161616161616161616161616161 61616161616161616161616161616161 61616161616161616161616161616161 61616161616161616161616161616161
expected 0001000070e865c932a56d

Vector 7 β€” All 256 byte values

Field Value
input 000102030405060708090a0b0c0d0e0f 101112131415161718191a1b1c1d1e1f 202122232425262728292a2b2c2d2e2f 303132333435363738393a3b3c3d3e3f 404142434445464748494a4b4c4d4e4f 505152535455565758595a5b5c5d5e5f 606162636465666768696a6b6c6d6e6f 707172737475767778797a7b7c7d7e7f 808182838485868788898a8b8c8d8e8f 909192939495969798999a9b9c9d9e9f a0a1a2a3a4a5a6a7a8a9aaabacadaeaf b0b1b2b3b4b5b6b7b8b9babbbcbdbebf c0c1c2c3c4c5c6c7c8c9cacbcccdcecf d0d1d2d3d4d5d6d7d8d9dadbdcdddedf e0e1e2e3e4e5e6e7e8e9eaebecedeeef f0f1f2f3f4f5f6f7f8f9fafbfcfdfeff
expected 00010000244b9e5f41569b1e4e74c0ae 47a08ce9844e4ff26153655c8e96877c 32b6d6f2e2308caf9d1470a06095d867 ab488e38112ff7ab7038045b6bce6523 … (261 bytes; the full value is in the source)

← All algorithms