Lux Papers Build #122

v122

papers: remove zen-of-lux (renamed to ethos-of-lux in initial commit)

Lux Papers Build #117

papers: sanitize brand references in LP-305 and LP-306

- LP-305: generalize 'Simplici provides biometric KYC' to 'regulated custodians supply biometric KYC attestations'
- LP-306: remove Liquidity/Liquidity.io brand references from social DEX status table

Lux Papers Build #116

papers: sanitize brand references in LP-305 and LP-306

- LP-305: generalize 'Simplici provides biometric KYC' to 'regulated custodians supply biometric KYC attestations'
- LP-306: remove Liquidity/Liquidity.io brand references from social DEX status table

Lux Papers Build #108

papers: convert 2022-era whitepapers to LaTeX

- zen-of-lux.tex — founding manifesto, 8 principles (renamed from lp-001
  to avoid collision with lps/LP-001 digital-securities spec)
- lp-305 expanded — formal SVP/CVP/LWE/RLWE definitions, FHE correctness
  theorem, Lamport OTS appendix, 5 new academic citations
- lp-306 — LX social trading DEX evolution (2022 → 2026)
- lp-307 — Eisenstein series quantum consensus protocol
- lp-308 — QuantumSwap AMM with bonding curves + interest rate model
- lp-309 — per-chain PQ vulnerability comparison (BTC/ETH/SOL/ADA)

No Liquidity/Satschel branding. LUXG/LUXU for gold/uranium-pegged stables.
GLUX correctly noted as the brief governance token (unrelated to gold).
All compile clean via pdflatex.

Lux Papers Build #106

v106

paper(anchor): state address as-is, no compat narrative

Lux Papers Build #102

paper: LP-7200 — on-chain checkpoint anchoring for CRDT state

8-page LaTeX paper specifying the anchor precompile at 0x0700...00 and
its integration with off-chain CRDT replicas.

Problem: when an app stores data off-chain (CRDT op logs, SQLite
replicas, IPFS blobs), users need a way to prove the server hasn't
silently rewritten history. Anchoring the state's Merkle root to a
public chain at known heights gives tamper-evident durability without
putting every op on-chain.

Design:
- 72-byte payload (appID 32 + height 8 + root 32) → single tx.
- Height monotonicity enforced on-chain: retroactive rewrites revert.
- Q-Chain is the recommended anchor target (PQ finality, near-zero
  gas); C-Chain is the fallback for apps already paying gas there.
- Batched (1000 docs, Merkle root of anchors-per-batch): ~$0.000001
  per anchor amortised.

Verification: eth_call to get(appID, height), compare to local root.
Zero gas, ~50 ms round-trip.

Implementation: luxfi/precompile@5eb5ce4; SDK hanzo/base/crdt@HEAD.

Lux Papers Build #101

v101

paper: babyjubjub — align with reality (reduced form, real address/ga…

Lux Papers Build #100

audit: OOM and resource-exhaustion review of precompiles + dex + node

5-page audit covering lux/precompile (25 precompiles),
node/vms/dexvm (DEX VM), and the generic codec layer.

Findings (no critical reachable from untrusted network input):
- F-1 Medium: DEX referral payments slice unbounded (referral.go:336)
- F-2 Low:    3 dead fields in perpetuals engine (engine.go:94-96)
- F-3 Medium: ADL events slice unbounded (adl.go:193)
- F-4 Medium: keystore codec maxPackerSize = 1 GiB (too permissive)
- F-5 Low:    platformvm GenesisCodec math.MaxInt32 (read-once, ok)

Safe by design (confirmed):
- bls12381 MSM — gas-metered by numPairs
- blake3 XOF    — MaxOutputLength = 1024
- ring sig      — ringSize fits in byte
- stableswap    — n bounded by input length
- dexvm block   — txLen bounded by remaining input

Plus 12-item benchmark coverage plan identifying what should be
benchmarked that currently isn't: per-precompile gas/alloc ratios,
DEX matching memory, perpetuals growth, codec unmarshal cost,
mempool eviction, consensus fan-out, state-trie memory, GPU context,
bridge message bursts, MPC ceremony memory, plus existing
DAG-EVM / DAG-Z harnesses.

Lux Papers Build #99

audit: OOM and resource-exhaustion review of precompiles + dex + node

5-page audit covering lux/precompile (25 precompiles),
node/vms/dexvm (DEX VM), and the generic codec layer.

Findings (no critical reachable from untrusted network input):
- F-1 Medium: DEX referral payments slice unbounded (referral.go:336)
- F-2 Low:    3 dead fields in perpetuals engine (engine.go:94-96)
- F-3 Medium: ADL events slice unbounded (adl.go:193)
- F-4 Medium: keystore codec maxPackerSize = 1 GiB (too permissive)
- F-5 Low:    platformvm GenesisCodec math.MaxInt32 (read-once, ok)

Safe by design (confirmed):
- bls12381 MSM — gas-metered by numPairs
- blake3 XOF    — MaxOutputLength = 1024
- ring sig      — ringSize fits in byte
- stableswap    — n bounded by input length
- dexvm block   — txLen bounded by remaining input

Plus 12-item benchmark coverage plan identifying what should be
benchmarked that currently isn't: per-precompile gas/alloc ratios,
DEX matching memory, perpetuals growth, codec unmarshal cost,
mempool eviction, consensus fan-out, state-trie memory, GPU context,
bridge message bursts, MPC ceremony memory, plus existing
DAG-EVM / DAG-Z harnesses.

Lux Papers Build #98

papers: precompile specifications for all 17 missing precompiles

BLS12-381 (EIP-2537), Baby Jubjub, Curve25519, Pasta (Pallas/Vesta),
X25519 (RFC 7748), X-Wing hybrid KEM, AES-256-GCM, ChaCha20-Poly1305,
KZG (EIP-4844), Pedersen hash/commit, Poseidon2, ECIES, HPKE (RFC 9180),
secp256r1 (EIP-7212), Ed25519 (RFC 8032), SR25519 (Schnorrkel), BLAKE3.

Each paper covers: specification, security analysis, gas cost
justification, and implementation notes. 4-8 pages, standalone LaTeX.