Proof and documentation
Visualizer: ML-DSA (Dilithium) Identity Cube
What the Identity Cube visualizer computes, the standardized parameters it uses, the evidence a reviewer can check, and the honest limits of what it does and does not do.
What it is
The Visualizer is a live, in-browser rendering of the lattice domain used by ML-DSA (CRYSTALS-Dilithium), the NIST post-quantum digital signature standard. It draws a cube of 4,096 points (256 polynomial coefficients across 16 layers, a 16 x 16 x 16 grid) using the real Dilithium field parameters, then computes and displays live statistics for that structure: minimum, maximum, mean, standard deviation, Shannon entropy, and a chi-square uniformity test per layer. It can also ingest a supplied Dilithium public key file and automatically report which parameter set it is (ML-DSA-44, ML-DSA-65, or ML-DSA-87) from the public-key byte length. The cube pattern itself is generated deterministically from an identity label using SHA-256, computed in your browser with the Web Crypto API. It is a visualization and statistical demonstration of the lattice domain, not a signing or verification operation.
What this proves to you (plain English)
Post-quantum signatures are not a black box you have to take on faith. This page shows the actual mathematical domain that ML-DSA works in, running entirely in your own browser with nothing pre-rendered and nothing fetched from a server. You can watch the coefficient distribution, confirm it is statistically uniform (the property the standard relies on), and see the tool correctly identify the strength tier of any Dilithium key you give it from its size alone. The point it proves is transparency: the math is open, standardized, and inspectable, not a vendor claim.
Standards and parameter set
- Signature scheme: ML-DSA (CRYSTALS-Dilithium), NIST FIPS 204. FACT.
- Primary view: ML-DSA-65, NIST security category 3, public key 1,952 bytes. FACT (FIPS 204).
- Field modulus q = 8,380,417 and polynomial degree n = 256. FACT (FIPS 204).
- Layer walk step delta = 524,288 (2 to the 19th power), which equals the ML-DSA-65 gamma1 bound in FIPS 204. FACT (FIPS 204).
- Parameter-set detection by public-key size: 1,312 bytes = ML-DSA-44 (category 2), 1,952 bytes = ML-DSA-65 (category 3), 2,592 bytes = ML-DSA-87 (category 5). FACT (FIPS 204).
- Context: NIST standardized ML-DSA in FIPS 204 (August 2024). Federal migration to post-quantum cryptography is directed under EO 14412 and OMB M-26-15. FACT (public record).
Correctness evidence (concrete, verifiable)
- Live computation, no server: the cube and all statistics are computed in the browser from SHA-256 using the Web Crypto API (crypto.subtle). Nothing is pre-rendered or fetched. FACT (demo source).
- Real standardized constants: the renderer uses q = 8,380,417, n = 256, and delta = 2 to the 19th power exactly as defined for ML-DSA-65 in FIPS 204. FACT (demo source cross-checked to FIPS 204).
- Uniformity test: each of the 16 layers is scored with a chi-square statistic against the published critical value 30.578 (alpha = 0.01, 15 degrees of freedom) and labeled PASS, MARGINAL, or REJECT. 30.578 is the standard chi-square table value for those parameters. FACT (statistical reference). INFERENCE: a PASS indicates the visualized coefficient distribution is statistically uniform; it is a property of the visualization, not a security rating of the signature scheme.
- Parameter-set recognition: given a Dilithium public key of 1,312 / 1,952 / 2,592 bytes, the tool reports ML-DSA-44 / 65 / 87 respectively. This is directly verifiable against the key sizes in FIPS 204. FACT (demo source cross-checked to FIPS 204).
Honesty caveats stated plainly:
- This demo does NOT sign, verify, or perform a key round-trip, so there is no byte-identical round-trip claim, no verify = TRUE / tamper = FALSE claim, and no NIST Known-Answer-Test (KAT) comparison here. Those belong to the Verifier and ZAP proofs, not this one.
- The cube pattern is seeded from the identity label, not derived from the private key or the bytes of a loaded key. It illustrates the lattice domain; it is not a cryptographic transform of key material. INFERENCE from demo source.
- The entropy and chi-square figures describe the visualization walk. They are not a measure of the security margin of ML-DSA itself. The hardness of ML-DSA rests on Module-LWE and Module-SIS, not on these display statistics. FACT (FIPS 204 security basis).
Measured performance
Omitted. This demonstration does not report measured timing or throughput numbers, so none are claimed here.
Client FAQ
1. Is this doing real cryptography or just an animation? It computes real SHA-256 values live in your browser and uses the real FIPS 204 field parameters for ML-DSA-65. It is an honest visualization of the lattice domain and its statistics. It is not a live signing engine, and the page says so.
2. Does my key or data leave my machine? No. The computation runs entirely in your browser with the Web Crypto API. Nothing is sent to a server, and nothing is pre-rendered.
3. What does the chi-square PASS actually tell me? That the visualized coefficient distribution is statistically uniform at the alpha = 0.01 level. Uniformity is the distribution property the standard depends on. It is a property of the display, not a certification of the signature scheme.
4. It identified my certificate as ML-DSA-65. How? From the public-key size. FIPS 204 fixes ML-DSA public keys at 1,312 bytes (ML-DSA-44), 1,952 bytes (ML-DSA-65), or 2,592 bytes (ML-DSA-87). The tool reads the size and reports the matching parameter set.
5. Which ML-DSA strength should we use? ML-DSA-65 is NIST category 3. For National Security Systems, CNSA 2.0 calls for the category 5 set, ML-DSA-87. Signatures grow with strength, which is the tradeoff to plan for.
6. Can we point this at our own certificate estate? Yes, as a working session. We will walk your certificates through identification and a migration path to ML-DSA and ML-KEM.
See it run