We revisit the question of securely compressing multiparty correlations using only symmetric cryptography. A linear correlation C, defined by a linear subspace C ⊆ 𝔽ⁿ, samples a secret random 𝐜 ∈ C and assigns to each party a fixed subset of the entries of 𝐜. Gilboa and Ishai (Crypto 1999) and Cramer, Damgård and Ishai (TCC 2005) provide a general technique for securely compressing many independent samples from C by replicating independent keys of a pseudorandom function (PRF) among the parties. This implies a pseudorandom correlation function (PCF) for C from any PRF, where the PCF key size scales with the number of minimal-support codewords in C. We observe that the above generalizes to other types of useful target correlations C_T by using a secret replication pattern obtained via a random secret assignment of parties in C to parties in C_T. We present several corollaries of this general blueprint. These include a re-derivation of two-party PCF constructions for VOLE and subfield-VOLE over small domains (Roy, Crypto 2022) as well as new multiparty PCFs for small-domain VOLE-style correlations, including scalar-vector multiplication triples and their authenticated variants. Finally, we discuss applications to secure computation.
A Distributed Point Function (DPF) is a crypto-graphic primitive used for compressing additive secret shares of a secret unit vector across two parties. Many DPF applications require compressed shares of a sparse weight- t vector, namely a Distributed Multi-Point Function (DMPF). Despite the strong motivation and prior optimization efforts, in most use cases the best practical implementation of DMPF is still a simple brute-force combination of
tindependent DPFs. We present new constructions and optimized implementations of DMPFs in different parameter regimes, providing significant efficiency savings over existing approaches. We showcase our new constructions within applications of pseu-dorandom correlation generators (PCGs) and 2-server private set intersection (PSI). Incorporating our tools into the state-of-the-art PCG for “silent” generation of binary multiplication triples (FOLEAGE, Bombar et al, ePrint'24) yields a x2.68 improvement in throughput, with only x 1.4 blowup in the seed size. On a single core of our benchmark machine, our implementation silently generates up to 22.1 million triples per second, outperforming even the best “non-silent” protocol (Roy, CRYPTO'22), which generates 16 million triples per second.