CKKS (Approximate Homomorphic Encryption)

CKKS is widely used for floating-point–like computation on encrypted data: it supports approximate arithmetic over real/complex values with controllable precision. It is not an IEEE-754 bit-exact floating-point circuit, but it is often the most practical choice for encrypted analytics and ML workloads.

Key ideas

• Packed encoding: multiple values can be encoded in one ciphertext (SIMD-style).
• Approximation: results have small numerical errors (controlled by parameters).
• Depth budget: multiplications consume “levels”; additions are much cheaper.

Parameters (high level)

• multiplicativeDepth: max multiplication layers supported.
• scalingModSize: controls precision / noise growth (bits per level).
• batchSize: number of packed slots used for SIMD.

Next: see the implemented examples and run them from the browser.