Part 2. How do you build with CKKS?
As described in Section 1.1, there are many different fully homomorphic encryption schemes. At 0xPARC, we chose to build the CKKS fully homomorphic encryption scheme because it handles large amounts of data in parallel and operates natively over real numbers, making it well-suited to real-world workloads.
By the end of this section, founders, builders, and engineers should have a clear sense of what kinds of computations are realistic with CKKS and what kind of performance to expect from our system.
In this section, we'll cover:
- the interface: key generation, the data shape of a ciphertext, and the five fundamental operations on ciphertexts (2.1),
- the cost model that determines what computations are faster and what computations are slower (2.2),
- the tweakable parameters available during setup and how they affect runtime, precision, and security (2.3), and
- how the demos from Section 1.2 are built using these primitives (2.4).