void
_nmod_poly_div_basecase_2(mp_ptr Q, mp_ptr W,
mp_srcptr A, long A_len, mp_srcptr B, long B_len,
nmod_t mod)
{
long coeff, i, len;
mp_limb_t lead_inv = n_invmod(B[B_len - 1], mod.n);
mp_ptr B2, R2;
mp_srcptr Btop;
B2 = W;
for (i = 0; i < B_len - 1; i++)
{
B2[2 * i] = B[i];
B2[2 * i + 1] = 0;
}
Btop = B2 + 2*(B_len - 1);
R2 = W + 2*(B_len - 1);
for (i = 0; i < A_len - B_len + 1; i++)
{
R2[2 * i] = A[B_len + i - 1];
R2[2 * i + 1] = 0;
}
coeff = A_len - B_len;
while (coeff >= 0)
{
mp_limb_t r_coeff;
r_coeff =
n_ll_mod_preinv(R2[2 * coeff + 1], R2[2 * coeff], mod.n, mod.ninv);
while (coeff >= 0 && r_coeff == 0L)
{
Q[coeff--] = 0L;
if (coeff >= 0)
r_coeff =
n_ll_mod_preinv(R2[2 * coeff + 1], R2[2 * coeff], mod.n,
mod.ninv);
}
if (coeff >= 0)
{
mp_limb_t c, * R_sub;
Q[coeff] =
n_mulmod2_preinv(r_coeff, lead_inv, mod.n, mod.ninv);
c = n_negmod(Q[coeff], mod.n);
len = FLINT_MIN(B_len - 1, coeff);
R_sub = R2 + 2 * (coeff - len);
if (len > 0)
mpn_addmul_1(R_sub, Btop - 2*len, 2 * len, c);
coeff--;
}
}
}
void
_nmod_poly_divrem_basecase_1(mp_ptr Q, mp_ptr R, mp_ptr W,
mp_srcptr A, slong lenA, mp_srcptr B, slong lenB,
nmod_t mod)
{
const mp_limb_t invL = n_invmod(B[lenB - 1], mod.n);
slong iR;
mp_ptr ptrQ = Q - lenB + 1;
mp_ptr R1 = W;
flint_mpn_copyi(R1, A, lenA);
for (iR = lenA - 1; iR >= lenB - 1; iR--)
{
if (R1[iR] == 0)
{
ptrQ[iR] = WORD(0);
}
else
{
ptrQ[iR] = n_mulmod2_preinv(R1[iR], invL, mod.n, mod.ninv);
if (lenB > 1)
{
const mp_limb_t c = n_negmod(ptrQ[iR], mod.n);
mpn_addmul_1(R1 + iR - lenB + 1, B, lenB - 1, c);
}
}
}
if (lenB > 1)
_nmod_vec_reduce(R, R1, lenB - 1, mod);
}
void _nmod_poly_rem_basecase_1(mp_ptr R, mp_ptr W,
mp_srcptr A, long lenA, mp_srcptr B, long lenB,
nmod_t mod)
{
if (lenB > 1)
{
const mp_limb_t invL = n_invmod(B[lenB - 1], mod.n);
long iR;
mp_ptr R1 = W;
mpn_copyi(R1, A, lenA);
for (iR = lenA - 1; iR >= lenB - 1; iR--)
{
if (R1[iR] != 0)
{
const mp_limb_t q = n_mulmod2_preinv(R1[iR], invL, mod.n, mod.ninv);
const mp_limb_t c = n_negmod(q, mod.n);
mpn_addmul_1(R1 + iR - lenB + 1, B, lenB - 1, c);
}
}
_nmod_vec_reduce(R, R1, lenB - 1, mod);
}
}
static __inline__ mp_limb_t
nmod_set_si(long v, nmod_t mod)
{
mp_limb_t u = n_mod2_preinv(FLINT_ABS(v), mod.n, mod.ninv);
if (v < 0)
u = n_negmod(u, mod.n);
return u;
}
void
_nmod_poly_divrem_basecase_3(mp_ptr Q, mp_ptr R, mp_ptr W,
mp_srcptr A, slong lenA, mp_srcptr B, slong lenB,
nmod_t mod)
{
const mp_limb_t invL = n_invmod(B[lenB - 1], mod.n);
slong iR, i;
mp_ptr B3 = W, R3 = W + 3*(lenB - 1), ptrQ = Q - lenB + 1;
for (i = 0; i < lenB - 1; i++)
{
B3[3 * i] = B[i];
B3[3 * i + 1] = 0;
B3[3 * i + 2] = 0;
}
for (i = 0; i < lenA; i++)
{
R3[3 * i] = A[i];
R3[3 * i + 1] = 0;
R3[3 * i + 2] = 0;
}
for (iR = lenA - 1; iR >= lenB - 1; )
{
mp_limb_t r =
n_lll_mod_preinv(R3[3 * iR + 2], R3[3 * iR + 1],
R3[3 * iR], mod.n, mod.ninv);
while ((iR + 1 >= lenB) && (r == WORD(0)))
{
ptrQ[iR--] = WORD(0);
if (iR + 1 >= lenB)
r = n_lll_mod_preinv(R3[3 * iR + 2], R3[3 * iR + 1],
R3[3 * iR], mod.n, mod.ninv);
}
if (iR + 1 >= lenB)
{
ptrQ[iR] = n_mulmod2_preinv(r, invL, mod.n, mod.ninv);
if (lenB > 1)
{
const mp_limb_t c = n_negmod(ptrQ[iR], mod.n);
mpn_addmul_1(R3 + 3 * (iR - lenB + 1), B3, 3 * lenB - 3, c);
}
iR--;
}
}
for (iR = 0; iR < lenB - 1; iR++)
R[iR] = n_lll_mod_preinv(R3[3 * iR + 2], R3[3 * iR + 1],
R3[3 * iR], mod.n, mod.ninv);
}
void _fq_nmod_trace(fmpz_t rop2, const mp_limb_t *op, slong len,
const fq_nmod_ctx_t ctx)
{
const slong d = fq_nmod_ctx_degree(ctx);
ulong i, l;
mp_limb_t *t, rop;
t = _nmod_vec_init(d);
_nmod_vec_zero(t, d);
t[0] = n_mod2_preinv(d, ctx->mod.n, ctx->mod.ninv);
for (i = 1; i < d; i++)
{
for (l = ctx->len - 2; l >= 0 && ctx->j[l] >= d - (i - 1); l--)
{
t[i] = n_addmod(t[i],
n_mulmod2_preinv(t[ctx->j[l] + i - d], ctx->a[l], ctx->mod.n, ctx->mod.ninv),
ctx->mod.n);
}
if (l >= 0 && ctx->j[l] == d - i)
{
t[i] = n_addmod(t[i],
n_mulmod2_preinv(ctx->a[l], i, ctx->mod.n, ctx->mod.ninv),
ctx->mod.n);
}
t[i] = n_negmod(t[i], ctx->mod.n);
}
rop = WORD(0);
for (i = 0; i < d; i++)
{
rop = n_addmod(rop,
n_mulmod2_preinv(op[i], t[i], ctx->mod.n, ctx->mod.ninv),
ctx->mod.n);
}
_nmod_vec_clear(t);
fmpz_set_ui(rop2, rop);
}
void
_nmod_poly_div_basecase_1(mp_ptr Q, mp_ptr W,
mp_srcptr A, long A_len, mp_srcptr B, long B_len,
nmod_t mod)
{
mp_limb_t lead_inv = n_invmod(B[B_len - 1], mod.n);
long len, coeff = A_len - B_len;
mp_ptr R1 = W;
mp_srcptr Btop = B + B_len - 1;
mpn_copyi(R1, A + B_len - 1, A_len - B_len + 1);
while (coeff >= 0)
{
R1[coeff] = n_mod2_preinv(R1[coeff], mod.n, mod.ninv);
while (coeff >= 0 && R1[coeff] == 0L)
{
Q[coeff--] = 0L;
if (coeff >= 0)
R1[coeff] = n_mod2_preinv(R1[coeff], mod.n, mod.ninv);
}
if (coeff >= 0)
{
mp_limb_t c, * R_sub;
Q[coeff] =
n_mulmod2_preinv(R1[coeff], lead_inv, mod.n, mod.ninv);
c = n_negmod(Q[coeff], mod.n);
len = FLINT_MIN(B_len - 1, coeff);
R_sub = R1 + coeff - len;
if (len > 0)
mpn_addmul_1(R_sub, Btop - len, len, c);
coeff--;
}
}
}
void _nmod_poly_rem_basecase_3(mp_ptr R, mp_ptr W,
mp_srcptr A, long lenA, mp_srcptr B, long lenB,
nmod_t mod)
{
if (lenB > 1)
{
const mp_limb_t invL = n_invmod(B[lenB - 1], mod.n);
long iR, i;
mp_ptr B3 = W, R3 = W + 3*(lenB - 1);
for (i = 0; i < lenB - 1; i++)
{
B3[3 * i] = B[i];
B3[3 * i + 1] = 0;
B3[3 * i + 2] = 0;
}
for (i = 0; i < lenA; i++)
{
R3[3 * i] = A[i];
R3[3 * i + 1] = 0;
R3[3 * i + 2] = 0;
}
for (iR = lenA - 1; iR >= lenB - 1; iR--)
{
const mp_limb_t r = n_lll_mod_preinv(R3[3*iR + 2], R3[3*iR + 1],
R3[3*iR], mod.n, mod.ninv);
if (r != 0)
{
const mp_limb_t q = n_mulmod2_preinv(r, invL, mod.n, mod.ninv);
const mp_limb_t c = n_negmod(q, mod.n);
mpn_addmul_1(R3 + 3 * (iR - lenB + 1), B3, 3 * lenB - 3, c);
}
}
for (iR = 0; iR < lenB - 1; iR++)
R[iR] = n_lll_mod_preinv(R3[3 * iR + 2], R3[3 * iR + 1],
R3[3 * iR], mod.n, mod.ninv);
}
}
void fmpz_powm_ui(fmpz_t f, const fmpz_t g, ulong e, const fmpz_t m)
{
if (fmpz_sgn(m) <= 0)
{
printf("Exception (fmpz_powm_ui). Modulus is less than 1.\n");
abort();
}
if (fmpz_is_one(m))
{
fmpz_zero(f);
}
else if (e == 0)
{
fmpz_one(f);
}
else /* e != 0, m > 0 */
{
fmpz g2 = *g;
fmpz m2 = *m;
if (!COEFF_IS_MPZ(m2)) /* m is small */
{
if (!COEFF_IS_MPZ(g2)) /* g is small */
{
mp_limb_t minv = n_preinvert_limb(m2);
_fmpz_demote(f);
if (g2 >= 0)
{
g2 = n_mod2_preinv(g2, m2, minv);
*f = n_powmod2_preinv(g2, e, m2, minv);
}
else
{
g2 = n_mod2_preinv(-g2, m2, minv);
*f = n_powmod2_preinv(g2, e, m2, minv);
if ((e & 1UL))
*f = n_negmod(*f, m2);
}
}
else /* g is large */
{
__mpz_struct *ptr = _fmpz_promote(f);
mpz_t m3;
mpz_init_set_ui(m3, m2);
mpz_powm_ui(ptr, COEFF_TO_PTR(g2), e, m3);
mpz_clear(m3);
_fmpz_demote_val(f);
}
}
else /* m is large */
{
if (!COEFF_IS_MPZ(g2)) /* g is small */
{
__mpz_struct *ptr = _fmpz_promote(f);
mpz_t g3;
mpz_init_set_si(g3, g2);
mpz_powm_ui(ptr, g3, e, COEFF_TO_PTR(m2));
mpz_clear(g3);
_fmpz_demote_val(f);
}
else /* g is large */
{
__mpz_struct *ptr = _fmpz_promote(f);
mpz_powm_ui(ptr, COEFF_TO_PTR(g2), e, COEFF_TO_PTR(m2));
_fmpz_demote_val(f);
}
}
}
}
