REDMARK FORGE SANDBOX
Measure · Model · Explore

Entropy & Information Studio

Shannon entropy, frequency distributions, redundancy, information limits, gzip-size estimates and first-order conditional entropy.

Browser-local: processed in this tab. Experimental learning tool; independently validate high-stakes conclusions.

01 · Input & analysis controls

Analyze text or a file; calculate Shannon entropy and compression ratio estimates.

No analysis yet.

02 · Information metrics

Entropy, theoretical coding limit, redundancy and compression estimate.

Samples—
Entropy—
Max entropy—
Redundancy—
Shannon limit—
Gzip size—
Run analysis for interpretation.

03 · Frequency table

Count, probability, self-information and cumulative probability.

No frequency table yet.

04 · Conditional entropy & entropy rate

For byte streams, estimate H(next byte | previous byte) using adjacent pairs.

Run main analysis first.
Finite samples and nonstationary sources bias estimates. Entropy is not a cryptographic security test.

05 · Min-entropy & randomness test battery

Rényi entropy chain (H₀ ≥ H₁ ≥ H₂ ≥ H∞), a conservative min-entropy bound, and NIST SP 800-22-style tests (frequency, block frequency, runs, longest run, cumulative sums) plus byte chi-square, serial correlation and Monte-Carlo π. Uses the text or file from panel 01.

Needs at least 1,000 bytes for meaningful p-values (use a file or generate sample data below).
Results appear here.

06 · Context entropy & sliding window

How much does knowing the previous k bytes help? Shows in-sample plug-in entropy and an honest adaptive-coding cost per order, plus an entropy profile along the input.

Run to see bits per byte for orders 0–5 (inputs up to 2 MB).

07 · Compression & Huffman lab

gzip / deflate / raw deflate sizes, the exact Huffman code for the symbol distribution (verified by encode → decode round trip) and how close it gets to the Shannon limit.

Results appear here.

08 · Distribution comparison & language guess

Compares the panel-01 input (A) with another text or file (B): cross-entropy, KL and Jensen–Shannon divergence, cosine similarity. Also estimates the language of A from letter frequencies.

Results appear here.

09 · Information-theory calculator

Enter probabilities or counts; or a joint table for mutual information; or a channel parameter for capacity.

Results appear here.