zheng proposals

design proposals for zheng proof system evolution. zheng-2 is the integrated architecture — all other proposals feed into it.

architecture

proposal	status	target
zheng-2	draft	dual-algebra proof architecture: 2 PCS, 1 IOP, 1 hash, 14 languages

PCS backends

proposal	status	target
binius-pcs	draft	F₂ tower native PCS — 32-64× cheaper bitwise ops
brakedown-pcs	draft	Merkle-free PCS via expander-graph codes — 1-5 KiB proofs

proof size and verification

proposal	status	target
algebraic-extraction	accepted	batch algebraic opening — 157 KiB → 5-12 KiB
gravity-commitment	accepted	mass-weighted encoding — verification ∝ importance

composition and recursion

proposal	status	target
folding-first	accepted	HyperNova composition — 70K constraints → 30 field ops
proof-carrying	accepted	zero proving latency — proof emerges from computation
universal-accumulator	draft	fold ALL proof types into one ~200 byte object

prover optimization

proposal	status	target
tensor-compression	accepted	O(√N) memory, O(N) streaming prover — mobile devices
gpu-prover	draft	full pipeline in VRAM — 45-100× throughput

domain-specific

proposal	status	target
ring-aware-fhe	draft	native TFHE bootstrapping — ring-structured CCS + jets

combined targets (all proposals)

                        zheng-1 (Whirlaway)     zheng-2             improvement
proof size (128-bit):   157 KiB                 1-5 KiB             30-150×
verification:           1.0 ms                  10-50 μs            20-100×
recursive step:         70K constraints          30 field ops        2,300×
prover memory:          O(N)                    O(√N)               √N ×
prover time:            O(N log N)              O(N) streaming      log N ×
proving latency:        separate step           zero (proof-carrying) ∞ ×
binary workloads:       32-64× overhead         native              32-64×
FHE bootstrapping:      catastrophic            native (ring-aware)  orders of magnitude
GPU throughput:         CPU only                45-100× on commodity  45-100×

recommended sequence

phase 1:  algebraic extraction      near-term proof size win on existing WHIR
phase 2:  folding-first             20× block proving, 1000× epoch composition
phase 3:  binius-pcs + kuro         binary workloads go native
phase 4:  proof-carrying            zero proving latency
phase 5:  ring-aware-fhe            native FHE bootstrapping
phase 6:  brakedown-pcs             Merkle-free replaces WHIR
phase 7:  gravity-commitment        power-law verification cost
phase 8:  tensor-compression        O(√N) memory, mobile provers
phase 9:  universal-accumulator     one proof for everything
phase 10: gpu-prover               full pipeline on GPU

cross-repo proposals

proposal	repo	relation to zheng
algebraic-nmt	bbg	polynomial commitment replaces NMT trees — uses zheng PCS
goldilocks-fhe	mudra	q = Goldilocks for FHE — enables ring-aware-fhe

lifecycle

status	meaning
draft	idea captured, open for discussion
accepted	approved — ready to implement
rejected	decided against, kept for rationale
implemented	done — migrated to relevant spec file

Dimensions

trident/reference/props

Design Proposals Proposals for language and VM design changes. Not spec — these are desires documented for future consideration. Each proposal is a standalone markdown file. Status is tracked in the frontmatter. Status values | Status | Meaning | |--------|---------| | `draft` | Idea captured, open…

nox/reference/props

props

hemera/reference/props

hemera proposals design proposals for hemera hash function evolution. hemera-2 (pre-genesis upgrade) | proposal | status | target | |----------|--------|--------| | inversion-sbox | draft | x⁻¹ partial S-box + 16 rounds: 36% fewer constraints, 5.4× MPC depth, 2^1046 algebraic degree | |…

bbg/reference/props

bbg proposals design proposals for bbg authenticated state layer evolution. proposals compose with zheng-2 proof architecture and hemera-2 hash. polynomial state | proposal | status | target | |----------|--------|--------| | algebraic-nmt | draft | 9 NMT trees → 1 polynomial, 30× fewer constraints…

zheng/reference/props.md