XLOG

A GPU-native logic programming language for unified symbolic reasoning.

Documentation: xlog.md · Whitepaper (PDF) · Language reference · Examples

When symbolic reasoning runs on the CPU and neural computation runs on the GPU, training can pay for host–device transfers between the two. XLOG addresses that boundary with a shared compiler and provider-owned CUDA runtime services supporting four reasoning paradigms — deterministic Datalog, exact and approximate probabilistic inference, epistemic reasoning, and differentiable neural-symbolic training — with SAT/MaxSAT verification available as a shared solver service. Transfer-free claims are route-specific: the certified resident recursive core and resident Monte Carlo sampled core record zero tracked host–device transfers within their measured regions. The recursive graph then reads one pinned terminal receipt; Monte Carlo reads a three-word postflight record, with query results read separately when requested.

Implemented in Rust with…