lux-apps/papers
1
0
Fork 0
mirror of https://github.com/lux-apps/papers.git synced 2026-11-03 05:07:46 +00:00
Code Issues Packages Projects Releases Wiki Activity
Files
backup/local-pre-sync-2026-06-01
papers/legacy-md/FOLLOW-UPS-2026-05-12.md
T
zeekay 2f7b324257 docs(fhe): purge tfhe-rs / Zama refs — Lux FHE is self-contained, no external library dep 
Drops the named-competitor scaffold's TFHE-rs and Concrete (both Zama) entries from
lux-fhe-benchmarks and the matching bibliography. The CGGI academic citation for the
underlying TFHE scheme (Chillotti et al., Journal of Cryptology 2020) is kept — it's
the scheme we implement, not a library we depend on. Competitor coverage retained:
TFHE-CGGI reference, OpenFHE, VeloFHE/OpenFHE-GPU, Microsoft SEAL.

Other touch-ups in the same sweep:
  lux-fhe-runtime/08-related-work.tex   — rewrites "Concrete (Zama)" subsection
                                           as a generic "FHE compiler stacks" one.
  lux-fhe-smart-contracts main + 03    — positive framing ("on luxfi/lattice")
                                           instead of negative ("not tfhe-rs, not Zama").
  legacy-md/FOLLOW-UPS-2026-05-12.md   — drops Zama tfhe-rs from the bench harness
                                           plan; OpenFHE + Lux-FHE only.

Lux FHE is native: luxfi/fhe + luxfi/lattice, with optional C++ / GPU (Metal+CUDA) /
FPGA acceleration paths. No tfhe-rs, no Zama, no external library wrapping.
2026-06-10 22:35:36 -07:00

3.2 KiB
Raw Permalink Blame History

Open follow-ups as of 2026-05-12

Generated by the post-publish sweep. Items here need decisions or external action — they're not "more code to write."

Blocked on external action

  1. Hanzo blog KMS — kms.hanzo.ai/api/v3/secrets/raw?secretPath=/deploy&environment=prod returns empty. Populate CLOUDFLARE_API_TOKEN + CLOUDFLARE_ACCOUNT_ID there, OR grant HANZO_API_KEY GitHub secret read on /deploy. Brain/QZMQ/accel posts committed (commit 05d503b) but not deployed.
  2. jin GitHub archive flag — current gh user lacks admin on zenlm/jin. Run gh repo archive zenlm/jin --yes with an org admin token, or click Settings → Archive in web UI. Tag v0.1.0 already marks the archival point.
  3. @luxfi/[email protected] not on npm — pinned as optionalDependency in teleport/api. Downstream installs succeed; the actual PQAccount class needs the wallet team's publish step.
  4. @luxfi/[email protected] not on npm — pinned as file:../../p3q/contracts in teleport/contracts. Same pattern.

Engineering decisions needed

  1. CGGMP21 library choice — no maintained npm package found. @silencelaboratories/cggmp21 and @taurushq/cggmp21 don't exist on npm at this date. Options: (a) write our own TS CGGMP21, (b) FFI to a Rust impl, (c) skip CGGMP21 and stay on FROST + Pulsar/Corona. The teleport cggmp21 signer currently throws explicit unimplemented errors.
  2. pulsard + coronad daemons — the teleport TS mpc bus has wire-locked JSON-RPC clients for both. The Go side (luxfi/pulsar, luxfi/corona) exposes only KAT generators today. Build the long-running daemon binaries exposing the wire surface documented in teleport/mpc/src/signers/{pulsar,corona}.ts.
  3. P3Q audit lift — luxfi/p3q v0.0.1 ships audit-gated stubs. Once the formal audit completes and crates/p3q-verifier bodies land, bump p3q + @luxfi/p3q-evm and teleport picks up the upgrade by dep-bump.

Benchmark experiments to run (papers carry [TBD-measure])

  1. Consensus WAN testbed — 21- and 100-node AWS c6i.8xlarge tri-region, Mysticeti orchestrator harness, Quasar vs HotStuff-2 vs CometBFT, ≥5 runs of ≥600s, p50/p99 with confidence intervals. Numbers fill lux-pq-consensus-benchmarks and lux-quasar-benchmarks.
  2. EVM Mgas/s — mainnet RLP replay 17M-17.01M across geth/reth/erigon/evmone/Lux GPU-EVM. Numbers fill evmgpu-benchmark.
  3. FHE primitives — OpenFHE binfhe-benchmark + Lux-FHE on Apple M2 / EPYC / H100. Numbers fill lux-fhe-benchmarks. Lux FHE is from-scratch on luxfi/lattice — no external TFHE library bench dep.
  4. ZAP wire — kcchu/buffer-benchmarks harness + ZAP backend + two-node gRPC vs ZAP RTT under HDRHistogram. Numbers fill lux-zap-benchmarks.
  5. Hanzo stack — task-completion harness from pooya.blog/2026 across LangGraph/AutoGen/CrewAI/OpenAI Swarm/Hanzo, all hitting same OpenAI-compatible endpoint. Numbers fill hanzo-stack-comparison.

Top 5 gap papers (per /tmp/oss-paper-gap.md)

  1. zwing PQ-authenticated channel protocol (LP-9702).
  2. QZMQ standalone security paper.
  3. accel / LP-138 GPU dispatch fabric.
  4. P3Q standalone (currently only inside lp-stricte2epq §8).
  5. Hanzo Brain knowledge graph.
Reference in New Issue View Git Blame Copy Permalink