next up previous
Next: Quadratic Newton Iteration Up: Symbolic Newton Iteration Previous: Symbolic Newton Iteration

Linear Newton Iteration

Theorem 12   Let R be an Euclidean domain with a regular Euclidean size $ \delta$. Let p $ \in$ R be a prime element. Let $ \phi$ $ \in$ R[y] and let a, a0 $ \in$ R such that
  1. $ \phi$(a) = 0,
  2. a $ \equiv$ a0modp,
  3. $ \phi$$\scriptstyle \prime$(a0) $ \neq$ 0 modp.
If a is unknown, but $ \phi$ and a0 are known then we can compute a p-adic expansion of a. Therefore we compute a exactly.

Proof. Let (a0, a1,..., ad) be a p-adic expansion of a w.r.t. p. Let k < d + 1 be a positive integer. By Proposition 13, the element

a(k) = a0 + a1p + ... ak-1pk-1 (92)

is a p-adic approximation of a at order k. By Proposition 11 there exists a polynomial $ \psi$ $ \in$ R[y] such that

$\displaystyle \phi$(a(k) + akpk) = $\displaystyle \phi$(a(k)) + $\displaystyle \phi$$\scriptstyle \prime$(a(k))akpk + $\displaystyle \psi$(a(k) + akpk)(akpk)2. (93)

Since we have

a $\displaystyle \equiv$ a(k) + akpkmodpk+1    

we deduce from Proposition 14

$\displaystyle \phi$(a) $\displaystyle \equiv$ $\displaystyle \phi$(a(k) + akpk)modpk+1 (94)

Since $ \phi$(a) = 0 this shows that $ \phi$(a(k) + akpk) is in the ideal generated by pk+1. Similarly, $ \phi$(a(k)) is in the ideal generated by pk. Therefore we can divide $ \phi$(a(k) + akpk) and $ \phi$(a(k)) by pk, leading to

$\displaystyle {\frac{{{\phi}(a^{(k)} + a_k p^k)}}{{p^k}}}$ = $\displaystyle {\frac{{{\phi}(a^{(k)})}}{{p^k}}}$ + $\displaystyle \phi$$\scriptstyle \prime$(a(k))ak + $\displaystyle \psi$(a(k) + akpk)(ak)2pk. (95)

Now observe that

$\displaystyle {\frac{{{\phi}(a^{(k)} + a_k p^k)}}{{p^k}}}$ $\displaystyle \equiv$ 0 $\displaystyle \equiv$ $\displaystyle \psi$(a(k) + akpk)(ak)2pkmodp. (96)

Let us denote by $ \Phi_{{p}}^{{}}$ the canonical homomorphism from R to R/$ \langle$p$ \rangle$. Then we obtain

$\displaystyle \Phi_{{p}}^{{}}$($\displaystyle {\frac{{{\phi}(a^{(k)})}}{{p^k}}}$) = - $\displaystyle \Phi_{{p}}^{{}}$($\displaystyle \phi$$\scriptstyle \prime$(a(k)))$\displaystyle \Phi_{{p}}^{{}}$(ak) (97)

Now, since a $ \equiv$ a0modp holds we have

$\displaystyle \Phi_{{p}}^{{}}$($\displaystyle \phi$$\scriptstyle \prime$(a(k))) $\displaystyle \equiv$ $\displaystyle \Phi_{{p}}^{{}}$($\displaystyle \phi$$\scriptstyle \prime$(a0))modp (98)

Finally, since $ \phi$$\scriptstyle \prime$(a0) $ \neq$ 0 modp holds we can solve Equatiion 97 for $ \Phi_{{p}}^{{}}$(ak). $ \qedsymbol$

Theorem 13   Let R be a commutative ring with identity element and let $ \cal {I}$ be a finitely generated ideal of R. Let r be a positive integer. Let f1,..., fn $ \in$ R[x1,..., xr] be n multivariate polynomials in the r variables x1,..., xr. Let a1,..., ar $ \in$ R be elements. Let U be the Jacobian matrix of f1,..., fn evaluated at (a1,..., ar). That is, U is the n×r matrix defined by

U = (uijwhere  uij = $\displaystyle {\frac{{{\partial} f_i}}{{{\partial} x_j}}}$(a1,..., ar) (99)

We assume that the following properties hold
  1. for every i = 1 ... n we have fi(a1,..., ar) $ \equiv$ 0 mod$ \cal {I}$.
  2. the Jacobian matrix U is left-invertible.
Then, for every positive integer $ \ell$ we can compute a($\scriptstyle \ell$)1,..., a($\scriptstyle \ell$)r $ \in$ R such that
  1. for every i = 1,..., n we have fi(a($\scriptstyle \ell$)1,..., a($\scriptstyle \ell$)r) $ \equiv$ 0 mod$ \cal {I}$$\scriptstyle \ell$,
  2. for every j = 1,..., r we have a($\scriptstyle \ell$)j $ \equiv$ ajmod$ \cal {I}$.

Proof. We proceed by induction on $ \ell$ $ \geq$ 1. For $ \ell$ = 1 the claim follows from the hypothesis of the theorem. So let $ \ell$ $ \geq$ 1 be such that the claim is true. Hence there exist a($\scriptstyle \ell$)1,..., a($\scriptstyle \ell$)r $ \in$ R such that

fi(a($\scriptstyle \ell$)1,..., a($\scriptstyle \ell$)r) $\displaystyle \equiv$ 0 mod$\displaystyle \cal {I}$$\scriptstyle \ell$i = 1,..., n (100)

and

a($\scriptstyle \ell$)j $\displaystyle \equiv$ ajmod$\displaystyle \cal {I}$i = 1,..., r (101)

Since $ \cal {I}$ is finitely generated, then so is $ \cal {I}$$\scriptstyle \ell$ and let g1,..., gs $ \in$ R such that

$\displaystyle \cal {I}$$\scriptstyle \ell$ = $\displaystyle \langle$g1,..., gs$\displaystyle \rangle$ (102)

Therefore, for every i = 1,..., n, there exist qi1,...qis $ \in$ R such that

fi(a($\scriptstyle \ell$)1,..., a($\scriptstyle \ell$)r)  =  $\displaystyle \Sigma_{{{k=1}}}^{{{k=s}}}$qikgk (103)

For each j = 1,..., r we want to compute Bj $ \in$ R such that

a($\scriptstyle \ell$+1)j = a($\scriptstyle \ell$)j + Bj (104)

is the desired next approximation. We impose Bj $ \in$ $ \cal {I}$$\scriptstyle \ell$ so let bj1,..., bjs $ \in$ R be such that

Bj  =  $\displaystyle \Sigma_{{{k=1}}}^{{{k=s}}}$bjkgk (105)

Using Proposition 12 we obtain

fi(a($\scriptstyle \ell$+1)1,..., a($\scriptstyle \ell$+1)r) =

fi(a($\scriptstyle \ell$)1 + B1,..., a($\scriptstyle \ell$)r + Br)

  $\displaystyle \equiv$ fi(a($\scriptstyle \ell$)1,..., a($\scriptstyle \ell$)r) + $\displaystyle \Sigma_{{{j=1}}}^{{{j=r}}}$u($\scriptstyle \ell$)ijBjmod$\displaystyle \cal {I}$$\scriptstyle \ell$+1
  $\displaystyle \equiv$ $\displaystyle \Sigma_{{{k=1}}}^{{{k=s}}}$qikgk + $\displaystyle \Sigma_{{{j=1}}}^{{{j=r}}}$u($\scriptstyle \ell$)ijBjmod$\displaystyle \cal {I}$$\scriptstyle \ell$+1
  $\displaystyle \equiv$ $\displaystyle \Sigma_{{{k=1}}}^{{{k=s}}}$qikgk + $\displaystyle \Sigma_{{{j=1}}}^{{{j=r}}}$u($\scriptstyle \ell$)ij$\displaystyle \left(\vphantom{{\Sigma}_{k=1}^{k=s} b_{jk} g_k }\right.$$\displaystyle \Sigma_{{{k=1}}}^{{{k=s}}}$bjkgk$\displaystyle \left.\vphantom{{\Sigma}_{k=1}^{k=s} b_{jk} g_k }\right)$mod$\displaystyle \cal {I}$$\scriptstyle \ell$+1
  $\displaystyle \equiv$ $\displaystyle \Sigma_{{{k=1}}}^{{{k=s}}}$$\displaystyle \left(\vphantom{q_{ik} + {\Sigma}_{j=1}^{j=r} u^{({\ell})}_{ij} b_{jk} }\right.$qik + $\displaystyle \Sigma_{{{j=1}}}^{{{j=r}}}$u($\scriptstyle \ell$)ijbjk$\displaystyle \left.\vphantom{q_{ik} + {\Sigma}_{j=1}^{j=r} u^{({\ell})}_{ij} b_{jk} }\right)$gkmod$\displaystyle \cal {I}$$\scriptstyle \ell$+1
(106)

where $ \left(\vphantom{ u^{({\ell})}_{ij} }\right.$u($\scriptstyle \ell$)ij$ \left.\vphantom{ u^{({\ell})}_{ij} }\right)$ is the Jacobian matrix of (f1,..., fn) at (a($\scriptstyle \ell$)1,..., a($\scriptstyle \ell$)r).

Hence, solving for a($\scriptstyle \ell$+1)1,..., a($\scriptstyle \ell$+1)r $ \in$ R such that

fi(a($\scriptstyle \ell$+1)1,..., a($\scriptstyle \ell$+1)r) $\displaystyle \equiv$ 0 mod$\displaystyle \cal {I}$$\scriptstyle \ell$+1    

leads to solving the system of linear equations:

qik + $\displaystyle \Sigma_{{{j=1}}}^{{{j=r}}}$u($\scriptstyle \ell$)ijbjk $\displaystyle \equiv$ 0 mod$\displaystyle \cal {I}$ (107)

for k = 1,..., s and i = 1,..., n.

Now using a($\scriptstyle \ell$)j $ \equiv$ ajmod$ \cal {I}$ for j = 1,..., r we obtain

u($\scriptstyle \ell$)ij $\displaystyle \equiv$ uijmod$\displaystyle \cal {I}$ (108)

Therefore the system linear equations given by Relation (107) has solutions. $ \qedsymbol$


next up previous
Next: Quadratic Newton Iteration Up: Symbolic Newton Iteration Previous: Symbolic Newton Iteration
Marc Moreno Maza
2004-04-27