FHE schemes need chains of NTT-compatible primes with specific arithmetic form. Ethoryx generates full coefficient modulus chains in milliseconds, with every prime guaranteed to satisfy q ≡ 1 (mod 2n) for any polynomial degree. Drop-in compatible with CoeffModulus::Create and equivalent interfaces across all major FHE libraries.
FHE schemes require not one prime but a chain of NTT-compatible primes that define the RNS (Residue Number System) decomposition. Ethoryx generates the full chain in a single API call, with every prime satisfying both the NTT constraint and the scheme's bit-size specification.
The generated chains plug directly into Microsoft SEAL's CoeffModulus::Create() and OpenFHE's equivalent constructors. Zama's Concrete/TFHE-rs toolchain is also compatible through standard NTT parameter interfaces.
We offer 50-62 bit primes for FHE-friendly NTT configurations; the exact range required for CKKS, BGV, and BFV schemes. Chain depths from 3 to 20 primes, suitable for inference-only through deep-circuit FHE machine learning.
Fetch an FHE coefficient modulus chain from the Ethoryx API, pass it straight to SEAL. No custom parameter-generation code needed.
# 1. Fetch a 5-prime chain for polynomial degree n=4096, 55 bits each curl "https://api.ethoryx.io/v1/generate/fhe?n=4096&count=5&bits_each=55" \ -H "X-API-Key: YOUR_KEY" # 2. Response (abbreviated) { "poly_degree_n": 4096, "primes": [ "36028797018963791", "36028797018963977", "36028797018964181", "36028797018964397", "36028797018964571" ], "ntt_compatible": true, "seal_compatible": true, "usage_example": "seal::CoeffModulus::Create(4096, {55, 55, 55, 55, 55})" } // 3. Pass directly to SEAL using namespace seal; EncryptionParameters parms(scheme_type::ckks); parms.set_poly_modulus_degree(4096); parms.set_coeff_modulus(CoeffModulus::Create(4096, {55,55,55,55,55})); SEALContext context(parms); // Or build from the exact primes returned above for bit-exact reproducibility
Serve ML predictions over encrypted inputs without seeing the data. Ethoryx FHE chains are suitable for CKKS-based inference on encrypted patient records, encrypted financial data, and encrypted user queries. Used at polynomial degrees n=8192 and n=16384.
Two parties compute the intersection of their data sets without either revealing their full set. Applicable to contact discovery (WhatsApp-style), fraud detection across banks, and genomic overlap analysis. Uses BFV-style schemes with 3-7 prime chains.
Query a database without revealing the query. CKKS and BGV both support range queries and dot-product queries over encrypted data. Used by Zama, Duality Technologies, and Inpher in production systems.
FHE primitives underlie many SMPC protocols. Generating the shared public parameters (coefficient moduli) correctly and reproducibly is essential and deterministic-start Ethoryx output makes cross-party verification straightforward.
| Polynomial degree | Standard schemes | Typical chain depth | Use case |
|---|---|---|---|
| n = 1024 | BFV / BGV | 2–3 primes | Simple boolean circuits |
| n = 2048 | BFV / BGV / CKKS | 3–5 primes | Small arithmetic circuits |
| n = 4096 | CKKS / BFV / BGV | 5–8 primes | ML inference · PSI |
| n = 8192 | CKKS | 8–14 primes | Deep ML · large-dim PSI |
| n = 16384 | CKKS | 14–20 primes | Training · deep networks |