Merge pull request #206 from zcash/book-sarkar

book: Move Sarkar explanation to implementation section

book/src/SUMMARY.md

@@ -22,6 +22,7 @@
     - [Inner product argument](design/proving-system/inner-product.md)
     - [Comparison to other work](design/proving-system/comparison.md)
   - [Implementation](design/implementation.md)
+    - [Fields](design/implementation/fields.md)
   - [Gadgets](design/gadgets.md)
     - [SHA-256](design/gadgets/sha256.md)
       - [16-bit table chip](design/gadgets/sha256/table16.md)

book/src/background/fields.md

@@ -256,95 +256,6 @@ quite large.)
 
 [ts-sqrt]: https://en.wikipedia.org/wiki/Tonelli%E2%80%93Shanks_algorithm
 
-### Sarkar square-root algorithm (table-based variant)
-We use a technique from [Sarkar2020](https://eprint.iacr.org/2020/1407.pdf) to compute
-square roots in `halo2`. The intuition behind the algorithm is that we can split the task
-into computing square roots in each multiplicative subgroup.
-
-Suppose we want to find the square root of $u$ modulo an odd prime $p$, where $u$ is a
-non-zero square in $\mathbb{Z}_p^\times$. We write $p - 1 \equiv 2^{n}m$ with $n \geq 1$
-and $m$ odd; $g = z^m$ where $z$ is a non-square in $\mathbb{Z}_p^\times$.
-
-Let $x_3 = uv^2, x_2 = x_3^{2^8}, x_1 = x_2^{2^8}, x_0 = x_1^{2^8}.$
-
-#### Precompute the following tables:
-$$
-gtab = \begin{bmatrix}
-g^0 & g^1 & ... & g^{255} \\
-(g^{2^8})^0 & (g^{2^8})^1 & ... & (g^{2^8})^{255} \\
-(g^{2^{16}})^0 & (g^{2^{16}})^1 & ... & (g^{2^{16}})^{255} \\
-(g^{2^{24}})^0 & (g^{2^{24}})^1 & ... & (g^{2^{24}})^{255}
-\end{bmatrix}
-$$
-
-$$
-invtab = \begin{bmatrix}
-(g^{2^{-24}})^0 & (g^{2^{-24}})^1 & ... & (g^{2^{-24}})^{255}
-\end{bmatrix}
-$$
-
-### i = 0, 1
-Using $invtab$, we lookup $t_0$ s.t. $x_0 = (g^{2^{-24}})^{t_0} \implies x_0 \cdot g^{t_0 \cdot 2^{24}} = 1.$
-
-Update global variable: $t = t_0.$
-
-Define $\alpha_1 = x_1 \cdot (g^{2^{16}})^{t}.$
-
-### i = 2
-Lookup $t_1$ s.t. 
-$$
-\begin{array}{l}
-\alpha_1 = (g^{2^{-24}})^{t_1} &\implies x_1 \cdot (g^{2^{16}})^{t_0} = (g^{2^{-24}})^{t_1} \\
-&\implies
-x_1 \cdot g^{(t_0 + 2^8 \cdot t_1) \cdot 2^{16}} = 1.
-\end{array}
-$$
-
-Update global variable:
-$t = t_0 + 2^8 \cdot t_1$
-
-Define $\alpha_2 = x_2 \cdot (g^{2^8})^{t}.$
-         
-### i = 3
-Lookup $t_2$ s.t. 
-
-$$
-\begin{array}{l}
-\alpha_2 = (g^{2^{-24}})^{t_2} &\implies x_2 \cdot (g^{2^8})^{t_0 + 2^8\cdot {t_1}} = (g^{2^{-24}})^{t_2} \\
-&\implies x_2 \cdot (g^{2^8})^{t_0 + 2^8 \cdot t_1 + 2^{16} \cdot t_2} = 1.
-\end{array}
-$$
-
-Update global variable:
-$t = t_0 + 2^8 \cdot t_1 + 2^{16} \cdot t_2$
-
-Define $\alpha_3 = x_3 \cdot g^{t}.$
-
-### Final result
-Lookup $t_3$ s.t.
-
-$$
-\begin{array}{l}
-\alpha_3 = (g^{2^{-24}})^{t_3} &\implies x_3 \cdot g^{t_0 + 2^8\cdot {t_1} + 2^{16} \cdot t_2} = (g^{2^{-24}})^{t_3} \\
-&\implies x_3 \cdot g^{t_0 + 2^8 \cdot t_1 + 2^{16} \cdot t_2 + 2^{24} \cdot t_3} = 1.
-\end{array}
-$$
-
-Update global variable:
-$t = t_0 + 2^8 \cdot t_1 + 2^{16} \cdot t_2 + 2^{24} \cdot t_3$
-
-We can now write
-$$
-\begin{array}{l}
-x_3 \cdot g^{t} = 1 &\implies x_3 \cdot g^{t + 1} = g \\
-&\implies uv^2 \cdot g^{t + 1} = g \\
-&\implies uv^2 = g^{-t} \\
-&\implies uv \cdot g^{t / 2} = v^{-1} g^{-t / 2}.
-\end{array}
-$$
-
-Squaring the RHS, we observe that $(v^{-1} g^{-t / 2})^2 = v^{-2}g^{-t} = u.$ Therefore, the square root of $u$ is $v^{-1} g^{-t / 2}.$
-
 ## Roots of unity
 
 In the previous sections we wrote $p - 1 = 2^k \cdot t$ with $t$ odd, and stated that an

book/src/design/implementation/fields.md

@@ -0,0 +1,97 @@
+# Fields
+
+The [Pasta curves](https://electriccoin.co/blog/the-pasta-curves-for-halo-2-and-beyond/)
+that we use in `halo2` are designed to be highly 2-adic, meaning that a large $2^S$
+[multiplicative subgroup](../../background/fields.md#multiplicative-subgroups) exists in
+each field. That is, we can write $p - 1 \equiv 2^S \cdot T$ with $T$ odd. For both Pallas
+and Vesta, $S = 32$; this helps to simplify the field implementations.
+
+## Sarkar square-root algorithm (table-based variant)
+
+We use a technique from [Sarkar2020](https://eprint.iacr.org/2020/1407.pdf) to compute
+[square roots](../../background/fields.md#square-roots) in `halo2`. The intuition behind
+the algorithm is that we can split the task into computing square roots in each
+multiplicative subgroup.
+
+Suppose we want to find the square root of $u$ modulo one of the Pasta primes $p$, where
+$u$ is a non-zero square in $\mathbb{Z}_p^\times$. We define a $2^S$
+[root of unity](../../background/fields.md#roots-of-unity) $g = z^T$ where $z$ is a
+non-square in $\mathbb{Z}_p^\times$, and precompute the following tables:
+
+$$
+gtab = \begin{bmatrix}
+g^0 & g^1 & ... & g^{2^8 - 1} \\
+(g^{2^8})^0 & (g^{2^8})^1 & ... & (g^{2^8})^{2^8 - 1} \\
+(g^{2^{16}})^0 & (g^{2^{16}})^1 & ... & (g^{2^{16}})^{2^8 - 1} \\
+(g^{2^{24}})^0 & (g^{2^{24}})^1 & ... & (g^{2^{24}})^{2^8 - 1}
+\end{bmatrix}
+$$
+
+$$
+invtab = \begin{bmatrix}
+(g^{2^{-24}})^0 & (g^{2^{-24}})^1 & ... & (g^{2^{-24}})^{2^8 - 1}
+\end{bmatrix}
+$$
+
+Let $v = u^{(T-1)/2}$. We can then define $x = uv \cdot v = u^T$ as an element of the
+$2^S$ multiplicative subgroup.
+
+Let $x_3 = x, x_2 = x_3^{2^8}, x_1 = x_2^{2^8}, x_0 = x_1^{2^8}.$
+
+### i = 0, 1
+Using $invtab$, we lookup $t_0$ such that
+$$
+x_0 = (g^{2^{-24}})^{t_0} \implies x_0 \cdot g^{t_0 \cdot 2^{24}} = 1.
+$$
+
+Define $\alpha_1 = x_1 \cdot (g^{2^{16}})^{t_0}.$
+
+### i = 2
+Lookup $t_1$ s.t. 
+$$
+\begin{array}{ll}
+\alpha_1 = (g^{2^{-24}})^{t_1} &\implies x_1 \cdot (g^{2^{16}})^{t_0} = (g^{2^{-24}})^{t_1} \\
+&\implies
+x_1 \cdot g^{(t_0 + 2^8 \cdot t_1) \cdot 2^{16}} = 1.
+\end{array}
+$$
+
+Define $\alpha_2 = x_2 \cdot (g^{2^8})^{t_0 + 2^8 \cdot t_1}.$
+         
+### i = 3
+Lookup $t_2$ s.t. 
+
+$$
+\begin{array}{ll}
+\alpha_2 = (g^{2^{-24}})^{t_2} &\implies x_2 \cdot (g^{2^8})^{t_0 + 2^8\cdot {t_1}} = (g^{2^{-24}})^{t_2} \\
+&\implies x_2 \cdot g^{(t_0 + 2^8 \cdot t_1 + 2^{16} \cdot t_2) \cdot 2^8} = 1.
+\end{array}
+$$
+
+Define $\alpha_3 = x_3 \cdot g^{t_0 + 2^8 \cdot t_1 + 2^{16} \cdot t_2}.$
+
+### Final result
+Lookup $t_3$ such that
+
+$$
+\begin{array}{ll}
+\alpha_3 = (g^{2^{-24}})^{t_3} &\implies x_3 \cdot g^{t_0 + 2^8\cdot {t_1} + 2^{16} \cdot t_2} = (g^{2^{-24}})^{t_3} \\
+&\implies x_3 \cdot g^{t_0 + 2^8 \cdot t_1 + 2^{16} \cdot t_2 + 2^{24} \cdot t_3} = 1.
+\end{array}
+$$
+
+Let $t = t_0 + 2^8 \cdot t_1 + 2^{16} \cdot t_2 + 2^{24} \cdot t_3$.
+
+We can now write
+$$
+\begin{array}{lclcl}
+x_3 \cdot g^{t} = 1 &\implies& x_3 &=& g^{-t} \\
+&\implies& uv^2 &=& g^{-t} \\
+&\implies& uv &=& v^{-1} \cdot g^{-t} \\
+&\implies& uv \cdot g^{t / 2} &=& v^{-1} \cdot g^{-t / 2}.
+\end{array}
+$$
+
+Squaring the RHS, we observe that $(v^{-1} g^{-t / 2})^2 = v^{-2}g^{-t} = u.$ Therefore,
+the square root of $u$ is $uv \cdot g^{t / 2}$; the first part we computed earlier, and
+the second part can be computed with three multiplications using lookups in $gtab$.