1. 引言
在创建proof之前,Prover有a table of cell assignments that it claims satisfy the constraint system。该table具有
n
=
2
k
n=2^k
n=2k行,可分为 advice columns、instance columns 以及 fixed columns。 将第
i
i
i个fixed column中的第
j
j
j行的assignment定义为
F
i
,
j
F_{i,j}
Fi,j?,同理,定义advice 和 instance assignment为
A
i
,
j
A_{i,j}
Ai,j?。 【注意,此处将fixed column assignment 与advice/instance column assignment区分的主要原因是:fixed columns由Verifier提供,而advice columns和instance columns 由Prover提供。实际上,instance column 和 fixed column的commitment均由Prover和Verifier计算,仅advice commitment会存储在proof中。】
为了对这些assignment进行commit,需为每列构建degree
n
?
1
n-1
n?1 Lagrange polynomials,基于的evaluation domain size 为
n
n
n(其中
ω
\omega
ω为
n
n
n-th primitive root of unity):
-
a
i
(
X
)
a_i(X)
ai?(X) interpolates such that
a
i
(
ω
j
)
=
A
i
,
j
a_i(\omega^j) = A_{i,j}
ai?(ωj)=Ai,j?.
-
f
i
(
X
)
f_i(X)
fi?(X) interpolates such that
f
i
(
ω
j
)
=
F
i
,
j
f_i(\omega^j) = F_{i,j}
fi?(ωj)=Fi,j?.
然后为每列的polynomial创建blinding commitment:
A
=
[
Commit
(
a
0
(
X
)
)
,
…
,
Commit
(
a
i
(
X
)
)
]
\mathbf{A} = [\text{Commit}(a_0(X)), \dots, \text{Commit}(a_i(X))]
A=[Commit(a0?(X)),…,Commit(ai?(X))]
F
=
[
Commit
(
f
0
(
X
)
)
,
…
,
Commit
(
f
i
(
X
)
)
]
\mathbf{F} = [\text{Commit}(f_0(X)), \dots, \text{Commit}(f_i(X))]
F=[Commit(f0?(X)),…,Commit(fi?(X))]
F
\mathbf{F}
F会作为key generation的一部分生成,使用的blinding factor为
1
1
1。
A
\mathbf{A}
A由Prover构建并发送给Verifier。
2. Committing to the lookup permutations
1)首先,Verifier提供sampling challenge
θ
\theta
θ 用于keep individual columns within lookups independent。
2)然后,Prover commits to the permutations for each lookup:
-
已知a lookup 具有 input column polynomials
[
A
0
(
X
)
,
…
,
A
m
?
1
(
X
)
]
[A_0(X), \dots, A_{m-1}(X)]
[A0?(X),…,Am?1?(X)] 和 table column polynomials
[
S
0
(
X
)
,
…
,
S
m
?
1
(
X
)
]
[S_0(X), \dots, S_{m-1}(X)]
[S0?(X),…,Sm?1?(X)],Prover会构建2个压缩的多项式:
A
compressed
(
X
)
=
θ
m
?
1
A
0
(
X
)
+
θ
m
?
2
A
1
(
X
)
+
?
+
θ
A
m
?
2
(
X
)
+
A
m
?
1
(
X
)
A_\text{compressed}(X) = \theta^{m-1} A_0(X) + \theta^{m-2} A_1(X) + \dots + \theta A_{m-2}(X) + A_{m-1}(X)
Acompressed?(X)=θm?1A0?(X)+θm?2A1?(X)+?+θAm?2?(X)+Am?1?(X)
S
compressed
(
X
)
=
θ
m
?
1
S
0
(
X
)
+
θ
m
?
2
S
1
(
X
)
+
?
+
θ
S
m
?
2
(
X
)
+
S
m
?
1
(
X
)
S_\text{compressed}(X) = \theta^{m-1} S_0(X) + \theta^{m-2} S_1(X) + \dots + \theta S_{m-2}(X) + S_{m-1}(X)
Scompressed?(X)=θm?1S0?(X)+θm?2S1?(X)+?+θSm?2?(X)+Sm?1?(X) -
Prover会根据 lookup argument的rules 来 permutes
A
compressed
(
X
)
A_\text{compressed}(X)
Acompressed?(X) 和
S
compressed
(
X
)
S_\text{compressed}(X)
Scompressed?(X)。
3)Prover为所有的lookups创建blinding commitments,并将相应的blinding commitments发送给Verifier。
L
=
[
(
Commit
(
A
′
(
X
)
)
)
,
Commit
(
S
′
(
X
)
)
)
,
…
?
]
\mathbf{L} = \left[ (\text{Commit}(A'(X))), \text{Commit}(S'(X))), \dots \right]
L=[(Commit(A′(X))),Commit(S′(X))),…]
4)Verifier收到
A
\mathbf{A}
A,
F
\mathbf{F}
F 和
L
\mathbf{L}
L之后,发送将用于permutation argument和lookup argument中的random challenges
β
,
γ
\beta,\gamma
β,γ。(因为2个argument是独立的,因此可复用
β
,
γ
\beta,\gamma
β,γ。)
3. Committing to the equality constraint permutation
令
c
c
c为the number of columns that are enabled for equality constraints。
令
m
m
m为可容纳于column set 中的maximum number of columns,该值不会超过 PLONK配置的polynomial degree bound。
令
u
u
u为定义在 Permutation argument 中的 number of “usable” rows。
令
b
=
c
e
i
l
i
n
g
(
c
/
m
)
b = \mathsf{ceiling}(c/m)
b=ceiling(c/m)。
1)Prover构建长度为
b
u
bu
bu 的 vector
P
\mathbf{P}
P,对于每一个column set
0
≤
a
<
b
0\leq a <b
0≤a<b和每行
0
≤
j
<
u
0\leq j<u
0≤j<u,有:
P
a
u
+
j
=
∏
i
=
a
m
min
?
(
c
,
(
a
+
1
)
m
)
?
1
v
i
(
ω
j
)
+
β
?
δ
i
?
ω
j
+
γ
v
i
(
ω
j
)
+
β
?
s
i
(
ω
j
)
+
γ
.
\mathbf{P}_{au + j} = \prod\limits_{i=am}^{\min(c, (a+1)m)-1} \frac{v_i(\omega^j) + \beta \cdot \delta^i \cdot \omega^j + \gamma}{v_i(\omega^j) + \beta \cdot s_i(\omega^j) + \gamma}.
Pau+j?=i=am∏min(c,(a+1)m)?1?vi?(ωj)+β?si?(ωj)+γvi?(ωj)+β?δi?ωj+γ?. 2)Prover 从
1
1
1开始,计算a running product of
P
\mathbf{P}
P,同时计算a vector of polynomials
Z
P
,
0..
b
?
1
Z_{P,0..b-1}
ZP,0..b?1? that each have a Lagrange basis representation corresponding to a
u
u
u-sized slice of this running product,详细参看Permutation argument。 3)Prover为每个
Z
p
,
a
Z_{p,a}
Zp,a?多项式创建blinding commitments,并将这些blinding commitments值发送给Verifier:
Z
P
=
[
Commit
(
Z
P
,
0
(
X
)
)
,
…
,
Commit
(
Z
P
,
b
?
1
(
X
)
)
]
\mathbf{Z_P} = \left[\text{Commit}(Z_{P,0}(X)), \dots, \text{Commit}(Z_{P,b-1}(X))\right]
ZP?=[Commit(ZP,0?(X)),…,Commit(ZP,b?1?(X))]
4. Committing to the lookup permutation product columns
对于每个lookup,除了需要commit to the individual permuted lookups,Prover还需要commit to the permutation product column:
- Prover构建a vector
P
P
P:
P
j
=
(
A
compressed
(
ω
j
)
+
β
)
(
S
compressed
(
ω
j
)
+
γ
)
(
A
′
(
ω
j
)
+
β
)
(
S
′
(
ω
j
)
+
γ
)
P_j = \frac{(A_\text{compressed}(\omega^j) + \beta)(S_\text{compressed}(\omega^j) + \gamma)}{(A'(\omega^j) + \beta)(S'(\omega^j) + \gamma)}
Pj?=(A′(ωj)+β)(S′(ωj)+γ)(Acompressed?(ωj)+β)(Scompressed?(ωj)+γ)? - Prover构建a polynomial
Z
L
Z_L
ZL?,该polynomial
Z
L
Z_L
ZL?具有a Lagrange basis representation corresponding to a running product of
P
P
P,初始有
Z
L
(
1
)
=
1
Z_L(1)=1
ZL?(1)=1。
- Verifier收到
A
\mathbf{A}
A,
F
\mathbf{F}
F 和
L
\mathbf{L}
L(即已commit to all the cell values used in lookup columns和每个lookup的
A
′
(
X
)
,
S
′
(
X
)
A'(X),S'(X)
A′(X),S′(X))之后,发送将用于random challenges
β
,
γ
\beta,\gamma
β,γ。
- Prover使用
β
,
γ
\beta,\gamma
β,γ用于combine the permutation argument for
A
′
(
X
)
A'(X)
A′(X)和
S
′
(
X
)
S'(X)
S′(X)并保持两者独立。
- Prover为每个
Z
L
Z_L
ZL?多项式创建blinding commitments,并将这些blinding commitments 发送给Verifier:
Z
L
=
[
Commit
(
Z
L
(
X
)
)
,
…
?
]
\mathbf{Z_L} = \left[\text{Commit}(Z_L(X)), \dots \right]
ZL?=[Commit(ZL?(X)),…]
参考资料
[1] Halo2 Book 之 circuit commitment
|