passagemath-ntl 10.6.36__cp314-cp314t-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl

sage/rings/padics/padic_ZZ_pX_CR_element.pyx

@@ -0,0 +1,3277 @@
+# sage_setup: distribution = sagemath-ntl
+# distutils: libraries = NTL_LIBRARIES gmp M_LIBRARIES
+# distutils: extra_compile_args = NTL_CFLAGS
+# distutils: include_dirs = NTL_INCDIR
+# distutils: library_dirs = NTL_LIBDIR
+# distutils: extra_link_args = NTL_LIBEXTRA
+# distutils: language = c++
+# sage.doctest: needs sage.rings.padics
+r"""
+`p`-adic ``ZZ_pX`` CR Element
+
+This file implements elements of Eisenstein and unramified extensions
+of `\ZZ_p` and `\QQ_p` with capped relative precision.
+
+For the parent class see :mod:`sage.rings.padics.padic_extension_leaves`.
+
+The underlying implementation is through NTL's ``ZZ_pX`` class.  Each
+element contains the following data:
+
+- ``ordp`` -- ``long``; a power of the uniformizer to scale the unit
+  by.  For unramified extensions this uniformizer is `p`, for Eisenstein
+  extensions it is not.  A value equal to the maximum value of a ``long``
+  indicates that the element is an exact zero.
+
+- ``relprec`` -- ``long``; a signed integer giving the precision to
+  which this element is defined.  For nonzero ``relprec``, the
+  absolute value gives the power of the uniformizer modulo which the
+  unit is defined.  A positive value indicates that the element is
+  normalized (ie ``unit`` is actually a unit: in the case of
+  Eisenstein extensions the constant term is not divisible by `p`, in
+  the case of unramified extensions that there is at least one
+  coefficient that is not divisible by `p`).  A negative value
+  indicates that the element may or may not be normalized.  A zero
+  value indicates that the element is zero to some precision.  If so,
+  ``ordp`` gives the absolute precision of the element.  If ``ordp``
+  is greater than ``maxordp``, then the element is an exact zero.
+
+- ``unit`` -- ``ZZ_pX_c``; an ntl ``ZZ_pX`` storing the unit part
+  The variable `x` is the uniformizer in the case of Eisenstein
+  extensions. If the element is not normalized, the ``unit`` may or
+  may not actually be a unit.  This ``ZZ_pX`` is created with global
+  ntl modulus determined by the absolute value of ``relprec``.  If
+  ``relprec`` is 0, ``unit`` **is not initialized**, or destructed if
+  normalized and found to be zero.  Otherwise, let `r` be ``relprec`` and
+  `e` be the ramification index over `\QQ_p` or `\ZZ_p`.
+  Then the modulus of unit is given by `p^{ceil(r/e)}`.  Note that all
+  kinds of problems arise if you try to mix moduli.
+  ``ZZ_pX_conv_modulus`` gives a semi-safe way to convert between
+  different moduli without having to pass through ``ZZX``.
+
+- ``prime_pow`` (some subclass of :class:`PowComputer_ZZ_pX`) -- a class,
+  identical among all elements with the same parent, holding common
+  data.
+
+  + ``prime_pow.deg`` -- the degree of the extension
+
+  + ``prime_pow.e`` -- the ramification index
+
+  + ``prime_pow.f`` -- the inertia degree
+
+  + ``prime_pow.prec_cap`` -- the unramified precision cap.  For
+    Eisenstein extensions this is the smallest power of `p` that is
+    zero.
+
+  + ``prime_pow.ram_prec_cap`` -- the ramified precision cap.  For
+    Eisenstein extensions this will be the smallest power of `x` that
+    is indistinguishable from zero.
+
+  + ``prime_pow.pow_ZZ_tmp``, prime_pow.pow_mpz_t_tmp``,
+    ``prime_pow.pow_Integer`` -- functions for accessing powers of
+    `p`.  The first two return pointers.  See
+    :mod:`sage.rings.padics.pow_computer_ext` for examples and important
+    warnings.
+
+  + ``prime_pow.get_context``, ``prime_pow.get_context_capdiv``,
+    ``prime_pow.get_top_context`` -- obtain an
+    ``ntl_ZZ_pContext_class`` corresponding to `p^n`.  The capdiv
+    version divides by ``prime_pow.e`` as appropriate.
+    ``top_context`` corresponds to `p^{\texttt{prec\_cap}}`.
+
+  + ``prime_pow.restore_context``,
+    ``prime_pow.restore_context_capdiv``,
+    ``prime_pow.restore_top_context`` -- restores the given context.
+
+  + ``prime_pow.get_modulus``, ``get_modulus_capdiv``,
+    ``get_top_modulus`` -- returns a ``ZZ_pX_Modulus_c*`` pointing to
+    a polynomial modulus defined modulo `p^n` (appropriately divided
+    by ``prime_pow.e`` in the capdiv case).
+
+EXAMPLES:
+
+An Eisenstein extension::
+
+    sage: R = Zp(5,5)
+    sage: S.<x> = R[]
+    sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+    sage: W.<w> = R.ext(f); W
+    5-adic Eisenstein Extension Ring in w defined by x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+    sage: z = (1+w)^5; z
+    1 + w^5 + w^6 + 2*w^7 + 4*w^8 + 3*w^10 + w^12 + 4*w^13 + 4*w^14 + 4*w^15 + 4*w^16 + 4*w^17 + 4*w^20 + w^21 + 4*w^24 + O(w^25)
+    sage: y = z >> 1; y
+    w^4 + w^5 + 2*w^6 + 4*w^7 + 3*w^9 + w^11 + 4*w^12 + 4*w^13 + 4*w^14 + 4*w^15 + 4*w^16 + 4*w^19 + w^20 + 4*w^23 + O(w^24)
+    sage: y.valuation()
+    4
+    sage: y.precision_relative()
+    20
+    sage: y.precision_absolute()
+    24
+    sage: z - (y << 1)
+    1 + O(w^25)
+    sage: (1/w)^12+w
+    w^-12 + w + O(w^13)
+    sage: (1/w).parent()
+    5-adic Eisenstein Extension Field in w defined by x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+
+Unramified extensions::
+
+    sage: g = x^3 + 3*x + 3
+    sage: A.<a> = R.ext(g)
+    sage: z = (1+a)^5; z
+    (2*a^2 + 4*a) + (3*a^2 + 3*a + 1)*5 + (4*a^2 + 3*a + 4)*5^2 + (4*a^2 + 4*a + 4)*5^3 + (4*a^2 + 4*a + 4)*5^4 + O(5^5)
+    sage: z - 1 - 5*a - 10*a^2 - 10*a^3 - 5*a^4 - a^5
+    O(5^5)
+    sage: y = z >> 1; y
+    (3*a^2 + 3*a + 1) + (4*a^2 + 3*a + 4)*5 + (4*a^2 + 4*a + 4)*5^2 + (4*a^2 + 4*a + 4)*5^3 + O(5^4)
+    sage: 1/a
+    (3*a^2 + 4) + (a^2 + 4)*5 + (3*a^2 + 4)*5^2 + (a^2 + 4)*5^3 + (3*a^2 + 4)*5^4 + O(5^5)
+    sage: FFp = R.residue_field()
+    sage: R(FFp(3))
+    3 + O(5)
+    sage: QQq.<zz> = Qq(25,4)
+    sage: QQq(FFp(3))
+    3 + O(5)
+    sage: FFq = QQq.residue_field(); QQq(FFq(3))
+    3 + O(5)
+    sage: zz0 = FFq.gen(); QQq(zz0^2)
+    (zz + 3) + O(5)
+
+Different printing modes::
+
+    sage: R = Zp(5, print_mode='digits'); S.<x> = R[]; f = x^5 + 75*x^3 - 15*x^2 + 125*x -5; W.<w> = R.ext(f)
+    sage: z = (1+w)^5; repr(z)
+    '...4110403113210310442221311242000111011201102002023303214332011214403232013144001400444441030421100001'
+    sage: R = Zp(5, print_mode='bars'); S.<x> = R[]; g = x^3 + 3*x + 3; A.<a> = R.ext(g)
+    sage: z = (1+a)^5; repr(z)
+    '...[4, 4, 4]|[4, 4, 4]|[4, 4, 4]|[4, 4, 4]|[4, 4, 4]|[4, 4, 4]|[4, 4, 4]|[4, 4, 4]|[4, 4, 4]|[4, 4, 4]|[4, 4, 4]|[4, 4, 4]|[4, 4, 4]|[4, 4, 4]|[4, 4, 4]|[4, 4, 4]|[4, 4, 4]|[4, 3, 4]|[1, 3, 3]|[0, 4, 2]'
+    sage: R = Zp(5, print_mode='terse'); S.<x> = R[]; f = x^5 + 75*x^3 - 15*x^2 + 125*x -5; W.<w> = R.ext(f)
+    sage: z = (1+w)^5; z
+    6 + 95367431640505*w + 25*w^2 + 95367431640560*w^3 + 5*w^4 + O(w^100)
+    sage: R = Zp(5, print_mode='val-unit'); S.<x> = R[]; f = x^5 + 75*x^3 - 15*x^2 + 125*x -5; W.<w> = R.ext(f)
+    sage: y = (1+w)^5 - 1; y
+    w^5 * (2090041 + 19073486126901*w + 1258902*w^2 + 674*w^3 + 16785*w^4) + O(w^100)
+
+You can get at the underlying ntl unit::
+
+    sage: z._ntl_rep()
+    [6 95367431640505 25 95367431640560 5]
+    sage: y._ntl_rep()
+    [2090041 19073486126901 1258902 674 16785]
+    sage: y._ntl_rep_abs()
+    ([5 95367431640505 25 95367431640560 5], 0)
+
+.. NOTE::
+
+    If you get an error ``internal error: can't grow this _ntl_gbigint,``
+    it indicates that moduli are being mixed inappropriately somewhere.
+
+    For example, when calling a function with a ``ZZ_pX_c`` as an
+    argument, it copies.  If the modulus is not
+    set to the modulus of the ``ZZ_pX_c``, you can get errors.
+
+AUTHORS:
+
+- David Roe (2008-01-01): initial version
+
+- Robert Harron (2011-09): fixes/enhancements
+
+- Julian Rueth (2014-05-09): enable caching through ``_cache_key``
+"""
+# ****************************************************************************
+#       Copyright (C) 2008 David Roe <roed.math@gmail.com>
+#                          William Stein <wstein@gmail.com>
+#                     2014 Julian Rueth <julian.rueth@fsfe.org>
+#
+#  Distributed under the terms of the GNU General Public License (GPL)
+#  as published by the Free Software Foundation; either version 2 of
+#  the License, or (at your option) any later version.
+#
+#                  https://www.gnu.org/licenses/
+# ****************************************************************************
+
+from cysignals.signals cimport sig_on, sig_off
+from sage.ext.stdsage cimport PY_NEW
+include "sage/libs/ntl/decl.pxi"
+
+from sage.rings.integer cimport Integer
+from sage.rings.rational cimport Rational
+from sage.libs.gmp.mpz cimport *
+from sage.libs.gmp.mpq cimport *
+from sage.libs.ntl.ntl_ZZX cimport ntl_ZZX
+from sage.libs.ntl.ntl_ZZ cimport ntl_ZZ
+from sage.libs.ntl.ntl_ZZ_p cimport ntl_ZZ_p
+from sage.libs.ntl.ntl_ZZ_pContext cimport ntl_ZZ_pContext_class
+from sage.rings.padics.padic_generic_element cimport pAdicGenericElement
+from cypari2.gen import Gen as pari_gen
+from sage.interfaces.abc import GpElement
+from sage.rings.finite_rings.integer_mod import IntegerMod_abstract
+from sage.rings.padics.padic_ext_element cimport pAdicExtElement
+from sage.rings.padics.precision_error import PrecisionError
+
+from sage.rings.padics.pow_computer_ext cimport PowComputer_ZZ_pX_small_Eis
+from sage.rings.padics.pow_computer_ext cimport PowComputer_ZZ_pX_big_Eis
+from sage.rings.finite_rings.integer_mod_ring import IntegerModRing
+
+cdef object infinity
+from sage.rings.infinity import infinity
+
+cdef long maxordp = (1L << (sizeof(long) * 8 - 2)) -1
+cdef long minusmaxordp = -maxordp
+
+cdef inline int check_ordp(long a) except -1:
+    if a > maxordp or a < minusmaxordp:
+        raise ValueError("valuation overflow")
+
+
+cdef class pAdicZZpXCRElement(pAdicZZpXElement):
+    def __init__(self, parent, x, absprec=infinity, relprec=infinity, empty=False):
+        r"""
+        Create an element of a capped relative precision, unramified
+        or Eisenstein extension of `\ZZ_p` or `\QQ_p`.
+
+        INPUT:
+
+        - ``parent`` -- either an ``EisensteinRingCappedRelative`` or
+          ``UnramifiedRingCappedRelative``
+
+        - ``x`` -- integer; rational, `p`-adic element, polynomial,
+          list, integer_mod, pari int/frac/poly_t/pol_mod, an
+          ``ntl_ZZ_pX``, an ``ntl_ZZ``, an ``ntl_ZZ_p``, an
+          ``ntl_ZZX``, or something convertible into parent.residue_field()
+
+        - ``absprec`` -- an upper bound on the absolute precision of the
+          element created
+
+        - ``relprec`` -- an upper bound on the relative precision of
+          the element created
+
+        - ``empty`` -- whether to return after initializing to zero
+          (without setting the valuation)
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = (1+w)^5; z  # indirect doctest
+            1 + w^5 + w^6 + 2*w^7 + 4*w^8 + 3*w^10 + w^12 + 4*w^13 + 4*w^14 + 4*w^15 + 4*w^16 + 4*w^17 + 4*w^20 + w^21 + 4*w^24 + O(w^25)
+            sage: W(pari('3 + O(5^3)'))
+            3 + O(w^15)
+            sage: W(R(3,3))
+            3 + O(w^15)
+            sage: W.<w> = R.ext(x^625 + 915*x^17 - 95)
+            sage: W(3)
+            3 + O(w^3125)
+            sage: W(w, 14)
+            w + O(w^14)
+
+        TESTS:
+
+        Check that :issue:`3865` is fixed::
+
+            sage: W(gp('3 + O(5^10)'))
+            3 + O(w^3125)
+
+
+        Check that :issue:`13612` has been fixed::
+
+            sage: R = Zp(3)
+            sage: S.<a> = R[]
+            sage: W.<a> = R.extension(a^2 + 1)
+            sage: W(W.residue_field().zero())
+            O(3)
+
+            sage: K = Qp(3)
+            sage: S.<a> = K[]
+            sage: L.<a> = K.extension(a^2 + 1)
+            sage: L(L.residue_field().zero())
+            O(3)
+        """
+        pAdicZZpXElement.__init__(self, parent)
+        self.relprec = 0
+        if empty:
+            return
+        cdef long aprec, rprec, ctx_prec, ltmp
+        if relprec is not infinity and not isinstance(relprec, Integer):
+            relprec = Integer(relprec)
+        if (relprec is infinity) or (relprec > parent.precision_cap()):
+            rprec = self.prime_pow.ram_prec_cap
+        else:
+            rprec = mpz_get_si((<Integer>relprec).value)
+        if rprec < 0:
+            rprec = 0
+        if absprec is not infinity:
+            if not isinstance(absprec, Integer):
+                absprec = Integer(absprec)
+            if mpz_fits_slong_p((<Integer>absprec).value) == 0:
+                absprec = infinity
+            else:
+                aprec = mpz_get_si((<Integer>absprec).value)
+        cdef mpz_t tmp
+        cdef ZZ_c tmp_z
+        cdef Py_ssize_t i
+        cdef Integer tmp_Int
+        cdef Rational xlift
+        if isinstance(x, pAdicGenericElement):
+            if self.prime_pow.in_field == 0 and x.valuation() < 0:
+                raise ValueError("element has negative valuation")
+            if x._is_base_elt(self.prime_pow.prime):
+                xlift = Rational(x.lift())
+                if mpq_sgn(xlift.value) == 0:
+                    if (<pAdicGenericElement>x)._is_exact_zero():
+                        if absprec is infinity:
+                            self._set_exact_zero()
+                        else:
+                            self._set_inexact_zero(aprec)
+                        return
+                ltmp = mpz_get_si((<Integer>x.precision_absolute()).value) * self.prime_pow.e
+                if absprec is infinity or ltmp < aprec:
+                    aprec = ltmp
+                self._set_from_mpq_both(xlift.value, aprec, rprec)
+                return
+        if isinstance(x, GpElement):
+            x = x.__pari__()
+        if isinstance(x, pari_gen):
+            if x.type() == "t_PADIC":
+                if x.variable() != self.prime_pow.prime:
+                    raise TypeError("cannot coerce a pari p-adic with the wrong prime")
+                ltmp = x.padicprec(self.prime_pow.prime) * self.prime_pow.e
+                if absprec is infinity or ltmp < aprec:
+                    aprec = ltmp
+                    absprec = 0 # absprec just has to be non-infinite: everything else uses aprec
+                x = x.lift()
+            if x.type() == 't_INT':
+                x = Integer(x)
+            elif x.type() == 't_FRAC':
+                x = Rational(x)
+            elif x.type() == 't_POLMOD' or x.type == 't_POL':
+                # This code doesn't check to see if the primes are the same.
+                L = []
+                x = x.lift().lift()
+                for i from 0 <= i <= x.poldegree():
+                    L.append(Integer(x.polcoef(i)))
+                x = L
+            else:
+                raise TypeError("unsupported coercion from pari: only p-adics, integers, rationals, polynomials and pol_mods allowed")
+        elif isinstance(x, IntegerMod_abstract):
+            mpz_init(tmp)
+            ctx_prec = mpz_remove(tmp, (<Integer>x.modulus()).value, self.prime_pow.prime.value)
+            if mpz_cmp_ui(tmp, 1) == 0:
+                mpz_clear(tmp)
+                x = x.lift()
+                if absprec is infinity or ctx_prec < aprec:
+                    aprec = ctx_prec
+                    absprec = 0 # absprec just has to be non-infinite: everything else uses aprec
+            else:
+                mpz_clear(tmp)
+                raise TypeError("cannot coerce from the given integer mod ring (not a power of the same prime)")
+        elif isinstance(x, ntl_ZZ_p):
+            ctx_prec = ZZ_remove(tmp_z, (<ntl_ZZ>x.modulus()).x, self.prime_pow.pow_ZZ_tmp(1)[0])
+            if ZZ_IsOne(tmp_z):
+                x = x.lift()
+                tmp_Int = PY_NEW(Integer)
+                ZZ_to_mpz(tmp_Int.value, &(<ntl_ZZ>x).x)
+                x = tmp_Int
+                if absprec is infinity or ctx_prec < aprec:
+                    aprec = ctx_prec
+                    absprec = 0 # absprec just has to be non-infinite: everything else uses aprec
+            else:
+                raise TypeError("cannot coerce the given ntl_ZZ_p (modulus not a power of the same prime)")
+        elif isinstance(x, ntl_ZZ):
+            tmp_Int = PY_NEW(Integer)
+            ZZ_to_mpz(tmp_Int.value, &(<ntl_ZZ>x).x)
+            x = tmp_Int
+        elif isinstance(x, int):
+            x = Integer(x)
+        elif x in parent.residue_field() and x.parent().is_finite():
+            # Should only reach here if x is not in F_p
+            z = parent.gen()
+            poly = x.polynomial().list()
+            x = sum([poly[i].lift() * (z ** i) for i in range(len(poly))], parent.zero())
+            if absprec is infinity or 1 < aprec:
+                aprec = 1
+                absprec = 0 # absprec just has to be non-infinite: everything else uses aprec
+        cdef pAdicZZpXCRElement _x
+        if isinstance(x, Integer):
+            if absprec is infinity:
+                self._set_from_mpz_rel((<Integer>x).value, rprec)
+            else:
+                self._set_from_mpz_both((<Integer>x).value, aprec, rprec)
+        elif isinstance(x, Rational):
+            if absprec is infinity:
+                self._set_from_mpq_rel((<Rational>x).value, rprec)
+            else:
+                self._set_from_mpq_both((<Rational>x).value, aprec, rprec)
+        elif isinstance(x, ntl_ZZ_pX):
+            if absprec is infinity:
+                self._set_from_ZZ_pX_rel(&(<ntl_ZZ_pX>x).x, (<ntl_ZZ_pX>x).c, rprec)
+            else:
+                self._set_from_ZZ_pX_both(&(<ntl_ZZ_pX>x).x, (<ntl_ZZ_pX>x).c, aprec, rprec)
+        elif isinstance(x, ntl_ZZX):
+            if absprec is infinity:
+                self._set_from_ZZX_rel((<ntl_ZZX>x).x, rprec)
+            else:
+                self._set_from_ZZX_both((<ntl_ZZX>x).x, aprec, rprec)
+        elif isinstance(x, pAdicExtElement):
+            if x.parent() is parent:
+                _x = <pAdicZZpXCRElement>x
+                if _x.relprec == 0:
+                    if absprec is infinity or aprec > _x.ordp:
+                        self._set_inexact_zero(_x.ordp) # this works for exact zeros too.
+                    else:
+                        self._set_inexact_zero(aprec)
+                elif _x.relprec < 0:
+                    if -_x.relprec < rprec:
+                        rprec = _x.relprec
+                    else:
+                        rprec = -rprec
+                    if absprec is infinity or aprec > _x.ordp - rprec:
+                        self._set(&_x.unit, _x.ordp, rprec)
+                    elif aprec > _x.ordp:
+                        self._set(&_x.unit, _x.ordp, _x.ordp - aprec) #negating relprec to indicate non-normalized.
+                    else:
+                        self._set_inexact_zero(aprec)
+                else:
+                    if _x.relprec < rprec:
+                        rprec = _x.relprec
+                    if absprec is infinity or aprec > _x.ordp + rprec:
+                        self._set(&_x.unit, _x.ordp, rprec)
+                    elif aprec > _x.ordp:
+                        self._set(&_x.unit, _x.ordp, aprec - _x.ordp)
+                    else:
+                        self._set_inexact_zero(aprec)
+            elif x.parent().fraction_field() is parent:
+                if isinstance(x, pAdicZZpXCRElement):
+                    _x = <pAdicZZpXCRElement>x
+                    if _x.relprec < 0:
+                        _x._normalize()
+                    if _x._is_exact_zero():
+                        self._set_exact_zero()
+                    elif _x._is_inexact_zero():
+                        self._set_inexact_zero(_x.ordp)
+                    else:
+                        if _x.relprec < rprec:
+                            rprec = _x.relprec
+                        self._set(&_x.unit, _x.ordp, rprec)
+                else:
+                    # x is a pAdicZZpXCAElement
+                    xordp = x.valuation()
+                    xprec = x.precision_absolute()
+                    if xordp == xprec:
+                        self._set_inexact_zero(mpz_get_si((<Integer>xordp).value))
+                    else:
+                        poly = x._ntl_rep_abs()[0]
+                        if absprec is infinity:
+                            self._set_from_ZZ_pX_rel(&(<ntl_ZZ_pX>poly).x,(<ntl_ZZ_pX>poly).c, rprec)
+                        else:
+                            self._set_from_ZZ_pX_both(&(<ntl_ZZ_pX>poly).x,(<ntl_ZZ_pX>poly).c, aprec, rprec)
+            elif x.parent() is parent.fraction_field():
+                _x = <pAdicZZpXCRElement>x
+                if _x.relprec < 0:
+                    _x._normalize()
+                if _x._is_exact_zero():
+                    self._set_exact_zero()
+                elif _x._is_inexact_zero():
+                    self._set_inexact_zero(_x.ordp)
+                else:
+                    if _x.relprec < rprec:
+                        rprec = _x.relprec
+                    self._set(&_x.unit, _x.ordp, rprec)
+            else:
+                raise NotImplementedError("conversion from different p-adic extensions not yet supported")
+        else:
+            try:
+                x = list(x)
+            except TypeError:
+                try:
+                    x = x.list()
+                except AttributeError:
+                    raise TypeError("cannot convert x to a p-adic element")
+            if absprec is infinity:
+                self._set_from_list_rel(x, rprec)
+            else:
+                self._set_from_list_both(x, aprec, rprec)
+
+    def _cache_key(self):
+        r"""
+        Return a hashable key which identifies this element.
+
+        This makes it possible to use this element in caches such as
+        functions or methods decorated with ``@cached_function`` or
+        ``@cached_method`` respectively.
+
+        EXAMPLES:
+
+        In the following example, ``a`` and ``b`` compare equal. They cannot
+        have a meaningful hash value since then their hash value would have to
+        be the same::
+
+            sage: K.<a> = Qq(9)
+            sage: b = a + O(3)
+            sage: a == b
+            True
+            sage: hash(a)
+            Traceback (most recent call last):
+            ...
+            TypeError: ...unhashable type: 'sage.rings.padics.qadic_flint_CR.qAdicCappedRelativeElement'...
+
+        However, we want to cache computations which depend on them. Therefore
+        they define a ``_cache_key`` which is hashable and uniquely identifies
+        them::
+
+            sage: a._cache_key()
+            (..., ((0, 1),), 0, 20)
+            sage: b._cache_key()
+            (..., ((0, 1),), 0, 1)
+
+        TESTS:
+
+        Check that zero values are handled correctly::
+
+            sage: K.zero()._cache_key()
+            (..., 0)
+            sage: K(0,1)._cache_key()
+            (..., 1, 0)
+        """
+        if self._is_exact_zero():
+            return (self.parent(), 0)
+        elif self._is_inexact_zero():
+            return (self.parent(), 0, self.valuation())
+        else:
+            return (self.parent(),
+                    tuple(tuple(c) if isinstance(c, list) else c
+                          for c in self.unit_part().expansion()),
+                    self.valuation(), self.precision_relative())
+
+    cdef int _set_inexact_zero(self, long absprec) except -1:
+        """
+        Set ``self`` to be zero with valuation absprec.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = W(0,6); z  # indirect doctest
+            O(w^6)
+            sage: z.valuation()
+            6
+            sage: z.precision_absolute()
+            6
+            sage: z.precision_relative()
+            0
+
+        TESTS::
+
+            sage: R = Zp(17, 3)
+            sage: S.<x> = R[]
+            sage: W.<w> = R.ext(x^34 - 289*x^5 + 17)
+            sage: z = W(0, 6); z
+            O(w^6)
+            sage: z.valuation()
+            6
+            sage: z.precision_absolute()
+            6
+            sage: z.precision_relative()
+            0
+        """
+        self.ordp = absprec
+        self.relprec = 0
+
+    cdef int _set_exact_zero(self) except -1:
+        """
+        Set ``self`` to be an exact zero.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = R(0); z  # indirect doctest
+            0
+            sage: z.valuation()
+            +Infinity
+            sage: z.precision_absolute()
+            +Infinity
+            sage: z.precision_relative()
+            0
+
+        TESTS::
+
+            sage: R = Zp(89, 3)
+            sage: S.<x> = R[]
+            sage: W.<w> = R.ext(x^34 - 2*89*x^5 + 89)
+            sage: z = R(0); z  # indirect doctest
+            0
+            sage: z.valuation()
+            +Infinity
+            sage: z.precision_absolute()
+            +Infinity
+            sage: z.precision_relative()
+            0
+        """
+        self.ordp = maxordp
+        self.relprec = 0
+
+    cpdef bint _is_exact_zero(self) except -1:
+        """
+        Test if ``self`` is an exact zero.
+
+        EXAMPLES::
+
+            sage: R = Qp(3,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = W(0)
+            sage: z._is_exact_zero()
+            True
+            sage: z = W(0,6)
+            sage: z._is_exact_zero()
+            False
+
+        TESTS::
+
+            sage: R = Qp(53, 3)
+            sage: S.<x> = R[]
+            sage: W.<w> = R.ext(x^34 - 2*53^5*x^9 + 53)
+            sage: z = W(0)
+            sage: z._is_exact_zero()
+            True
+            sage: z = W(0,6)
+            sage: z._is_exact_zero()
+            False
+        """
+        if self.ordp == maxordp:
+            return 1
+        else:
+            return 0
+
+    cpdef bint _is_inexact_zero(self) except -1:
+        """
+        Test if ``self`` is an inexact zero.
+
+        EXAMPLES::
+
+            sage: R = Zp(7,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = W(0)
+            sage: z._is_inexact_zero()
+            False
+            sage: z = W(0,6)
+            sage: z._is_inexact_zero()
+            True
+
+        TESTS::
+
+            sage: R = Qp(29, 3)
+            sage: S.<x> = R[]
+            sage: W.<w> = R.ext(x^29 - 2*29^5*x - 29)
+            sage: z = W(0)
+            sage: z._is_inexact_zero()
+            False
+            sage: z = W(0,6)
+            sage: z._is_inexact_zero()
+            True
+        """
+        self._normalize()
+        if self.relprec == 0:
+            return not self._is_exact_zero()
+        else:
+            return False
+
+    cdef int _set(self, ZZ_pX_c* unit, long ordp, long relprec) except -1:
+        """
+        Set ``unit``, ``ordp`` and ``relprec`` directly.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: F = W.fraction_field()
+            sage: z = F(1 + w); z  # indirect doctest
+            1 + w + O(w^25)
+
+        TESTS::
+
+            sage: R = Zp(17,30)
+            sage: S.<x> = R[]
+            sage: f = x^51 - 34
+            sage: W.<w> = R.ext(f)
+            sage: F = W.fraction_field()
+            sage: z = F(1 + w); z  # indirect doctest
+            1 + w + O(w^1530)
+            sage: z = F(w + w^2, relprec=0); z
+            O(w)
+        """
+        self.ordp = ordp
+        self._set_prec_rel(relprec)
+        if self.relprec != 0:
+            ZZ_pX_conv_modulus(self.unit, unit[0], self.prime_pow.get_context_capdiv(relprec).x)
+
+    cdef int _set_from_mpz_rel(self, mpz_t x, long relprec) except -1:
+        """
+        Set ``self`` from an ``mpz_t`` with relative precision bounded by ``relprec``.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: W(70, relprec=8)  # indirect doctest
+            4*w^5 + 3*w^7 + w^9 + 2*w^10 + 2*w^11 + O(w^13)
+            sage: W(70, relprec=0)
+            O(w^5)
+
+        TESTS::
+
+            sage: R = Qp(13,50)
+            sage: S.<x> = R[]
+            sage: f = x^169 - 13
+            sage: W.<w> = R.ext(f)
+            sage: a = W(65, relprec=8); a.valuation()  # indirect doctest
+            169
+            sage: W(65, relprec=0)
+            O(w^169)
+        """
+        if mpz_sgn(x) == 0:
+            self._set_exact_zero()
+            return 0
+        cdef mpz_t tmp_m
+        cdef ZZ_c tmp_z
+        cdef long shift
+        mpz_init(tmp_m)
+        sig_on()
+        shift = mpz_remove(tmp_m, x, self.prime_pow.prime.value)
+        sig_off()
+        self._set_prec_rel(relprec)
+        mpz_to_ZZ(&tmp_z, tmp_m)
+        mpz_clear(tmp_m)
+        if self.relprec != 0:
+            ZZ_pX_SetCoeff(self.unit, 0, ZZ_to_ZZ_p(tmp_z))
+            self.ordp = 0
+            self._pshift_self(shift)
+        else:
+            self.ordp = shift * self.prime_pow.e
+
+    cdef int _set_from_mpz_both(self, mpz_t x, long absprec, long relprec) except -1:
+        """
+        Set ``self`` from an ``mpz_t`` with relative precision bounded by ``relprec``
+        and absolute precision bounded by ``absprec``.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: W(70, 8)  # indirect doctest
+            4*w^5 + 3*w^7 + O(w^8)
+            sage: W(70, absprec=4)
+            O(w^4)
+
+        TESTS::
+
+            sage: R = Zp(7,3)
+            sage: S.<x> = R[]
+            sage: f = x^49 + 7*x^21 - 14
+            sage: W.<w> = R.ext(f)
+            sage: W(70, 100)  # indirect doctest
+            5*w^49 + 6*w^70 + 3*w^91 + O(w^100)
+            sage: W(70, absprec=4)
+            O(w^4)
+        """
+        if mpz_sgn(x) == 0:
+            self._set_inexact_zero(absprec)
+            return 0
+        cdef mpz_t tmp_m
+        cdef ZZ_c tmp_z
+        cdef long shift
+        mpz_init(tmp_m)
+        sig_on()
+        shift = mpz_remove(tmp_m, x, self.prime_pow.prime.value)
+        sig_off()
+        self.ordp = shift * self.prime_pow.e
+        if self._set_prec_both(absprec, relprec) == 1:
+            # This indicates that self._set_inexact_zero was called
+            mpz_clear(tmp_m)
+            return 0
+        mpz_to_ZZ(&tmp_z, tmp_m)
+        mpz_clear(tmp_m)
+        if self.relprec != 0:
+            ZZ_pX_SetCoeff(self.unit, 0, ZZ_to_ZZ_p(tmp_z))
+            self.ordp = 0
+            self._pshift_self(shift)
+
+    cdef int _set_from_mpq_rel(self, mpq_t x, long relprec) except -1:
+        """
+        Set ``self`` from an ``mpq_t`` with relative precision
+        bounded by ``relprec``.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = W(70/3, relprec=9); z  # indirect doctest
+            3*w^5 + w^7 + 2*w^9 + 2*w^10 + 4*w^11 + w^12 + 2*w^13 + O(w^14)
+            sage: z * 3
+            4*w^5 + 3*w^7 + w^9 + 2*w^10 + 2*w^11 + w^13 + O(w^14)
+            sage: W(70)
+            4*w^5 + 3*w^7 + w^9 + 2*w^10 + 2*w^11 + w^13 + 3*w^16 + w^17 + w^18 + 4*w^20 + 4*w^21 + w^22 + 2*w^23 + 3*w^25 + w^27 + O(w^30)
+            sage: F = W.fraction_field()
+            sage: y = F(3/700); y
+            w^-10 + w^-8 + 4*w^-6 + w^-3 + 4*w^-2 + 3*w^-1 + 3 + 4*w + w^3 + 4*w^4 + w^5 + 4*w^6 + 2*w^7 + 3*w^8 + 4*w^9 + 3*w^10 + 4*w^11 + w^12 + O(w^15)
+            sage: y * 700
+            3 + O(w^25)
+            sage: W(70/3, relprec=0)
+            O(w^5)
+            sage: c = F(5^-1 + O(5^2)); c
+            w^-5 + 3*w^-3 + 2*w^3 + 4*w^5 + 4*w^6 + 3*w^7 + w^9 + O(w^10)
+            sage: c * 5
+            1 + O(w^15)
+
+        TESTS::
+
+            sage: R = Zp(11, 8, print_mode='digits')
+            sage: S.<x> = R[]
+            sage: f = x^3 + 1331 * x^2 - 11 * x + 11
+            sage: W.<w> = R.ext(f)
+            sage: z = W(77/3, relprec=11); repr(z)[3:]
+            '304107A2555000'
+            sage: repr(z*3)[3:]
+            '56698765444000'
+            sage: repr(W(77))[3:]
+            '5800A6604678856698765444000'
+            sage: F = W.fraction_field()
+            sage: y = F(3/847); repr(y)[3:]
+            '5563A4105291255628.148272'
+            sage: repr(y*847)[3:]
+            '000000000000000000000003'
+            sage: repr(W(77/3, relprec=0))[3:]
+            '000'
+            sage: c = F(11^-1 + O(11^2)); repr(c)[3:]
+            '011111.01A'
+            sage: repr(c * 11)[3:]
+            '000000001'
+        """
+        if mpq_sgn(x) == 0:
+            self._set_exact_zero()
+            return 0
+        cdef mpz_t num_unit, den_unit
+        self._set_from_mpq_part1(num_unit, den_unit, x)
+        self._set_prec_rel(relprec)
+        self._set_from_mpq_part2(num_unit, den_unit)
+
+    cdef int _set_from_mpq_both(self, mpq_t x, long absprec, long relprec) except -1:
+        """
+        Set ``self`` from an ``mpq_t`` with relative precision
+        bounded by ``relprec`` and absolute precision bounded by
+        ``absprec``.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = W(70/3, 14); z  # indirect doctest
+            3*w^5 + w^7 + 2*w^9 + 2*w^10 + 4*w^11 + w^12 + 2*w^13 + O(w^14)
+            sage: z * 3
+            4*w^5 + 3*w^7 + w^9 + 2*w^10 + 2*w^11 + w^13 + O(w^14)
+            sage: W(70)
+            4*w^5 + 3*w^7 + w^9 + 2*w^10 + 2*w^11 + w^13 + 3*w^16 + w^17 + w^18 + 4*w^20 + 4*w^21 + w^22 + 2*w^23 + 3*w^25 + w^27 + O(w^30)
+            sage: F = W.fraction_field()
+            sage: y = F(3/700,-2); y
+            w^-10 + w^-8 + 4*w^-6 + w^-3 + O(w^-2)
+            sage: y * 700
+            3 + O(w^8)
+            sage: W(70/3, absprec=4)
+            O(w^4)
+        """
+        if mpq_sgn(x) == 0:
+            self._set_inexact_zero(absprec)
+            return 0
+        cdef mpz_t num_unit, den_unit
+        self._set_from_mpq_part1(num_unit, den_unit, x)
+        if self._set_prec_both(absprec, relprec) == 1:
+            # indicates an inexact zero
+            mpz_clear(num_unit)
+            mpz_clear(den_unit)
+            return 0
+        self._set_from_mpq_part2(num_unit, den_unit)
+
+    cdef int _set_from_mpq_part1(self, mpz_t num_unit, mpz_t den_unit, mpq_t x) except -1:
+        """
+        Set ``num_unit`` to be the unit of the numerator, ``den_unit`` to be
+        the unit of the denominator and sets ``self.ordp`` correctly.
+
+        TESTS::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = W(7000/3, 23); z  # indirect doctest
+            2*w^15 + 2*w^17 + 3*w^19 + w^22 + O(w^23)
+        """
+        cdef long num_ordp, den_ordp
+        sig_on()
+        mpz_init(num_unit)
+        mpz_init(den_unit)
+        num_ordp = mpz_remove(num_unit, mpq_numref(x), self.prime_pow.prime.value)
+        den_ordp = mpz_remove(den_unit, mpq_denref(x), self.prime_pow.prime.value)
+        sig_off()
+        self.ordp = (num_ordp - den_ordp) * self.prime_pow.e
+        if self.ordp < 0 and self.prime_pow.in_field == 0:
+            mpz_clear(num_unit)
+            mpz_clear(den_unit)
+            raise ValueError("p divides the denominator")
+
+    cdef int _set_from_mpq_part2(self, mpz_t num_unit, mpz_t den_unit) except -1:
+        """
+        Given that ``self.ordp`` and ``self.relprec`` have been set, takes
+        ``num_unit`` and ``den_unit`` and sets ``self.unit``.
+
+        TESTS::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: W(QQ(0), 23)  # indirect doctest
+            O(w^23)
+            sage: W(QQ(0))
+            0
+        """
+        cdef ZZ_c num_zz, den_zz
+        cdef ZZ_p_c tmp_zp
+        cdef long val = self.ordp / self.prime_pow.e
+        cdef mpz_t tmp_m
+        if self.relprec != 0:
+            mpz_init(tmp_m)
+            mpz_set(tmp_m, num_unit)
+            mpz_to_ZZ(&num_zz, tmp_m)
+            mpz_set(tmp_m, den_unit)
+            mpz_to_ZZ(&den_zz, tmp_m)
+            mpz_clear(tmp_m)
+            #The context has been restored in setting self.relprec
+            ZZ_p_div(tmp_zp, ZZ_to_ZZ_p(num_zz), ZZ_to_ZZ_p(den_zz))
+            ZZ_pX_SetCoeff(self.unit, 0, tmp_zp)
+            self.ordp = 0
+            self._pshift_self(val)
+
+    cdef int _set_from_ZZX_rel(self, ZZX_c poly, long relprec) except -1:
+        """
+        Set ``self`` from a ``ZZX`` with relative precision bounded by
+        ``relprec``.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = W(ntl.ZZX([4,1,16]), relprec=14); z  # indirect doctest
+            4 + w + w^2 + 3*w^7 + w^9 + 2*w^11 + 4*w^13 + O(w^14)
+            sage: z._ntl_rep()
+            [4 1 16]
+            sage: z = W(ntl.ZZX([5^40,5^42,3*5^41]), relprec=14); z
+            w^200 + 4*w^207 + 4*w^209 + w^210 + 2*w^211 + 2*w^213 + O(w^214)
+            sage: W(5)^40 + w*W(5)^42 + w^2 * W(3) * W(5)^41
+            w^200 + 4*w^207 + 4*w^209 + w^210 + 2*w^211 + 2*w^213 + 2*w^215 + w^217 + 2*w^218 + w^220 + w^221 + w^222 + 3*w^224 + O(w^225)
+            sage: z = W(ntl.ZZX([5^40,5^42,3*5^41]), relprec=0); z
+            O(w^200)
+        """
+        if ZZX_IsZero(poly):
+            self._set_exact_zero()
+            return 0
+        if ZZX_deg(poly) >= self.prime_pow.deg:
+            raise NotImplementedError
+        # the -1 in the next line signals that there is no absprec specified
+        self._set_from_ZZX_part1(poly, -1, relprec)
+        # context was restored in _set_from_ZZX_part1
+        if relprec == 0:
+            self._set_prec_rel(relprec)
+            return 0
+        if self.relprec + self.ordp != 0:
+            self.prime_pow.restore_context_capdiv(self.relprec + self.ordp)
+            ZZX_to_ZZ_pX(self.unit, poly)
+            self._internal_lshift(-self.ordp)
+
+    cdef int _set_from_ZZX_both(self, ZZX_c poly, long absprec, long relprec) except -1:
+        """
+        Set ``self`` from a ``ZZX`` with relative precision bounded by
+        ``relprec`` and absolute precision bounded by ``absprec``.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = W(ntl.ZZX([4,1,16]), 12); z  # indirect doctest
+            4 + w + w^2 + 3*w^7 + w^9 + 2*w^11 + O(w^12)
+            sage: z._ntl_rep()
+            [4 1 16]
+            sage: z = W(ntl.ZZX([5^40,5^42,3*5^41]), 212); z
+            w^200 + 4*w^207 + 4*w^209 + w^210 + 2*w^211 + O(w^212)
+            sage: z = W(ntl.ZZX([5^40,5^42,3*5^41]), 197); z
+            O(w^197)
+        """
+        if ZZX_IsZero(poly) or absprec <= 0:
+            self._set_inexact_zero(absprec)
+            return 0
+        if ZZX_deg(poly) >= self.prime_pow.deg:
+            raise NotImplementedError
+        if self._set_from_ZZX_part1(poly, absprec, relprec) == -2:
+            # indicates _set_inexact_zero was called
+            return 0
+        # context was restored in _set_from_ZZX_part1
+        if self.relprec + self.ordp != 0:
+            self.prime_pow.restore_context_capdiv(self.relprec + self.ordp)
+            ZZX_to_ZZ_pX(self.unit, poly)
+            self._internal_lshift(-self.ordp)
+
+    cdef int _set_from_ZZX_part1(self, ZZX_c poly, long absprec, long relprec) except -1:
+        """
+        Set ``self.ordp`` from ``poly`` and restores the context. ``poly`` must
+        have degree less than ``self.prime_pow.deg``.
+
+        TESTS::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = W(ntl.ZZX([4,1,16]), 12); z  # indirect doctest
+            4 + w + w^2 + 3*w^7 + w^9 + 2*w^11 + O(w^12)
+        """
+        cdef long i = 0
+        cdef long deg = ZZX_deg(poly)
+        cdef long mini = -1
+        cdef long minval
+        cdef long curval
+        cdef ZZ_c tmp_z
+        while mini == -1:
+            if not ZZ_IsZero(ZZX_coeff(poly,i)):
+                minval = ZZ_remove(tmp_z, ZZX_coeff(poly, i), self.prime_pow.pow_ZZ_tmp(1)[0])
+                mini = i
+            i += 1
+        while i <= deg:
+            if not ZZ_IsZero(ZZX_coeff(poly,i)):
+                curval = ZZ_remove(tmp_z, ZZX_coeff(poly, i), self.prime_pow.pow_ZZ_tmp(1)[0])
+                if curval < minval:
+                    minval = curval
+                    mini = i
+            i += 1
+        if self.prime_pow.e == 1:
+            self.ordp = minval
+        else:
+            self.ordp = minval * self.prime_pow.e + mini
+        if absprec == -1: # indicates that _set_from_ZZX_rel is calling
+            self._set_prec_rel(relprec)
+        elif self._set_prec_both(absprec, relprec):
+            # indicates self._set_inexact_zero was called
+            return -2
+        # _set_prec_rel or both has restored the context so that part2 works.
+
+    cdef int _set_from_ZZ_pX_rel(self, ZZ_pX_c* poly, ntl_ZZ_pContext_class ctx, long relprec) except -1:
+        """
+        Set ``self`` from a ``ZZ_pX`` with relative precision bounded by
+        ``relprec``.
+
+        If ``ctx`` is ``None`` and ``poly`` is 0 this function will raise an
+        error (a ``ZZ_pX`` cannot represent something with infinite absolute
+        precision).
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = W(ntl.ZZ_pX([4,1,16],5^2)); z  # indirect doctest
+            4 + w + w^2 + 3*w^7 + w^9 + O(w^10)
+            sage: z._ntl_rep()
+            [4 1 16]
+            sage: z = W(ntl.ZZ_pX([5^40,5^42,3*5^41], 5^44)); z
+            w^200 + 4*w^207 + 4*w^209 + w^210 + 2*w^211 + 2*w^213 + 2*w^215 + w^217 + 2*w^218 + O(w^220)
+            sage: z = W(ntl.ZZ_pX([5^40,5^42,3*5^41], 5^44), relprec=0); z
+            O(w^200)
+        """
+        cdef long ctx_prec = -1
+        if ctx is not None:
+            ctx_prec = self._check_ZZ_pContext(ctx) * self.prime_pow.e
+        if ZZ_pX_IsZero(poly[0]):
+            if ctx_prec == -1:
+                raise ValueError("must specify either a context or an absolute precision bound")
+            else:
+                self._set_inexact_zero(ctx_prec)
+            return 0
+        self._set_from_ZZ_pX_part1(poly)
+        if relprec == 0:
+            self._set_prec_rel(relprec)
+            return 0
+        if ctx_prec == -1:
+            self._set_prec_rel(self.ordp + relprec)
+        else:
+            self._set_prec_rel(min(ctx_prec, self.ordp + relprec))
+        self._set_from_ZZ_pX_part2(poly)
+
+    cdef int _set_from_ZZ_pX_both(self, ZZ_pX_c* poly, ntl_ZZ_pContext_class ctx, long absprec, long relprec) except -1:
+        """
+        Set ``self`` from a ``ZZ_pX`` with relative precision bounded by
+        ``relprec`` and absolute precision bounded by ``absprec``.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = W(ntl.ZZ_pX([4,1,16],5^2), absprec=8, relprec=12); z  # indirect doctest
+            4 + w + w^2 + 3*w^7 + O(w^8)
+            sage: z._ntl_rep()
+            [4 1 16]
+            sage: z = W(ntl.ZZ_pX([5^40,5^42,3*5^41], 5^50), 220); z
+            w^200 + 4*w^207 + 4*w^209 + w^210 + 2*w^211 + 2*w^213 + 2*w^215 + w^217 + 2*w^218 + O(w^220)
+            sage: z = W(ntl.ZZ_pX([5^40,5^42,3*5^41], 5^44), absprec=77); z
+            O(w^77)
+        """
+        cdef long ctx_prec
+        if ctx is not None:
+            ctx_prec = self._check_ZZ_pContext(ctx)
+            if ctx_prec * self.prime_pow.e < absprec:
+                absprec = ctx_prec * self.prime_pow.e
+        if ZZ_pX_IsZero(poly[0]):
+            self._set_inexact_zero(absprec)
+            return 0
+        self._set_from_ZZ_pX_part1(poly)
+        if absprec <= self.ordp:
+            self._set_inexact_zero(absprec)
+        else:
+            self._set_prec_rel(min(absprec, self.ordp + relprec))
+            self._set_from_ZZ_pX_part2(poly)
+
+    cdef int _set_from_ZZ_pX_part1(self, ZZ_pX_c* poly) except -1:
+        """
+        Set ``self.ordp`` based on ``poly``.  ``poly`` must not be 0.
+
+        TESTS::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = W(ntl.ZZ_pX([4,1,16],5^2), absprec=8, relprec=12); z  # indirect doctest
+            4 + w + w^2 + 3*w^7 + O(w^8)
+        """
+        cdef long val = 0, index = 0
+        ZZ_pX_min_val_coeff(val, index, poly[0], self.prime_pow.pow_ZZ_tmp(1)[0])
+        if self.prime_pow.e == 1:
+            self.ordp = val
+        else:
+            self.ordp = val * self.prime_pow.e + index
+
+    cdef int _set_from_ZZ_pX_part2(self, ZZ_pX_c* poly) except -1:
+        """
+        Assuming that ``self.ordp`` and ``self.relprec`` have been set, sets
+        ``self.unit`` to ``poly`` and then normalizes.
+
+        TESTS::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = W(ntl.ZZ_pX([4,1,16],5^2), absprec=8, relprec=12); z  # indirect doctest
+            4 + w + w^2 + 3*w^7 + O(w^8)
+        """
+        # We've set self.relprec to what is actually the absolute precision.
+        if self.relprec != 0:
+            ZZ_pX_conv_modulus(self.unit, poly[0], self.prime_pow.get_context_capdiv(self.relprec).x)
+            self.relprec -= self.ordp
+            self._internal_lshift(-self.ordp)
+
+    cdef bint _set_prec_rel(self, long relprec) except -1:
+        """
+        Safely set the relative precision of ``self`` to be the absolute
+        value of ``relprec``.
+
+        Returns ``True`` iff ``self.relprec`` was reset.
+
+        Note that this will wipe out anything in ``self.unit``.  Be
+        careful resetting ``self.unit`` directly: if you set it to a
+        different modulus, NTL may have problems.  The safest way to
+        reset ``self.unit`` to a different modulus is::
+
+            self.prime_pow.restore_context_capdiv(self.relprec)
+            cdef ZZ_pX_c tmp = self.unit
+            self._set_prec_rel(new_rel_prec)
+            ZZ_pX_conv_modulus(self.unit, tmp, self.prime_pow.get_context_capdiv(self.relprec).x)
+
+        You may be able to just set ``self.relprec`` and
+        ``ZZ_pX_conv_modulus`` if you're decreasing precision.  I'm
+        not sure.
+
+        TESTS::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: W(70, relprec=8)  # indirect doctest
+            4*w^5 + 3*w^7 + w^9 + 2*w^10 + 2*w^11 + O(w^13)
+        """
+        if self.relprec == relprec:
+            return False
+        if relprec != 0:
+            self.prime_pow.restore_context_capdiv(relprec)
+            self.unit = ZZ_pX_c()
+        self.relprec = relprec
+        return True
+
+    cdef bint _set_prec_both(self, long absprec, long relprec) except -1:
+        """
+        Assuming ``self.ordp`` is set, sets the relative precision of ``self``
+        to the minimum of ``abs(relprec)`` and ``absprec-self.ordp``.
+
+        If ``relprec`` is negative, will set ``self.relprec`` to be negative
+        (indicating unnormalized unit)
+
+        Returns ``True`` iff ``self.relprec = 0``, ie ``self`` was set to an
+        inexact zero.
+
+        Note that this will wipe out anything in ``self.unit``.  Be
+        careful resetting ``self.unit`` directly: if you set it to a
+        different modulus, NTL may have problems.  The safest way to
+        reset ``self.unit`` to a different modulus is:
+
+            self.prime_pow.restore_context_capdiv(self.relprec)
+            cdef ZZ_pX_c tmp = self.unit
+            self._set_prec_rel(new_rel_prec)
+            ZZ_pX_conv_modulus(self.unit, tmp, self.prime_pow.get_context_capdiv(self.relprec).x)
+
+        You may be able to just set ``self.relprec`` and
+        ``ZZ_pX_conv_modulus`` if you're decreasing precision.  I'm
+        not sure.
+
+        TESTS::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: W(70, 8)  # indirect doctest
+            4*w^5 + 3*w^7 + O(w^8)
+        """
+        self.relprec = absprec - self.ordp
+        cdef long arelprec
+        if relprec < 0:
+            arelprec = -relprec
+        else:
+            arelprec = relprec
+        if self.relprec <= 0:
+            self._set_inexact_zero(absprec)
+        else:
+            if arelprec < self.relprec:
+                self.relprec = arelprec
+            if self.relprec != 0:
+                self.prime_pow.restore_context_capdiv(self.relprec)
+                self.unit = ZZ_pX_c()
+                if relprec < 0:
+                    self.relprec = -self.relprec
+        return self.relprec == 0
+
+    cdef int _normalize(self) except -1:
+        """
+        Normalize ``self``, adjusting ``self.ordp``, ``self.relprec``, and
+        ``self.unit`` so that ``self.unit`` actually represents a unit.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = (1+w)^5
+            sage: y = z - 1
+            sage: y._ntl_rep_unnormalized()
+            [5 3005 25 3060 5]
+            sage: y  # indirect doctest
+            w^5 + w^6 + 2*w^7 + 4*w^8 + 3*w^10 + w^12 + 4*w^13 + 4*w^14 + 4*w^15 + 4*w^16 + 4*w^17 + 4*w^20 + w^21 + 4*w^24 + O(w^25)
+            sage: y._ntl_rep_unnormalized()
+            [41 26 152 49 535]
+        """
+        cdef long minval = 0, mini = 0, shift
+        if self.relprec < 0:
+            if ZZ_pX_IsZero(self.unit):
+                self.ordp -= self.relprec # note that self.relprec < 0
+                self.relprec = 0
+            else:
+                ZZ_pX_min_val_coeff(minval, mini, self.unit, self.prime_pow.pow_ZZ_tmp(1)[0])
+                if self.prime_pow.e == 1:
+                    shift = minval
+                else:
+                    shift = minval * self.prime_pow.e + mini
+                if shift >= -self.relprec:
+                    self.ordp -= self.relprec # note that self.relprec < 0
+                    self.relprec = 0
+                elif shift > 0:
+                    self.relprec = -self.relprec - shift
+                    self.ordp += shift
+                    self._internal_lshift(-shift)
+                else:
+                    self.relprec = -self.relprec
+
+    def _is_normalized(self):
+        """
+        Whether this element is currently normalized.
+
+        EXAMPLES::
+
+            sage: R.<a> = ZqCR(125,implementation="NTL"); b = 5*a + 4; c = 10*a^2 + 6; d = b + c
+            sage: d._is_normalized()
+            False
+            sage: d.valuation()
+            1
+            sage: d._is_normalized()
+            True
+        """
+        return self.relprec >= 0
+
+    cdef int _internal_lshift(self, long shift) except -1:
+        """
+        Multiply ``self.unit`` by ``x^shift``.
+
+        Note that ``self.relprec`` must be set before calling this
+        function and should not be 0, and self.unit must be defined to
+        precision ``self.relprec - shift``
+
+        This function does not alter ``self.ordp`` even though it WILL
+        change the valuation of ``self.unit``
+
+        Also note that if you call this function you should usually
+        manually set ``self.relprec = -self.relprec`` since this function
+        will usually unnormalize ``self``.
+
+        TESTS::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = (1+w)^5
+            sage: y = z - 1
+            sage: y  # indirect doctest
+            w^5 + w^6 + 2*w^7 + 4*w^8 + 3*w^10 + w^12 + 4*w^13 + 4*w^14 + 4*w^15 + 4*w^16 + 4*w^17 + 4*w^20 + w^21 + 4*w^24 + O(w^25)
+        """
+        if self.relprec == 0:
+            raise ValueError("p-adic Internal l-shift called with relative precision 0")
+        cdef ZZ_pX_c tmpP
+        cdef ZZ_pX_Modulus_c* mod
+        if self.prime_pow.e == 1:
+            if shift > 0:
+                ZZ_pX_left_pshift(self.unit, self.unit, self.prime_pow.pow_ZZ_tmp(shift)[0], self.prime_pow.get_context(self.relprec).x)
+            else:
+                ZZ_pX_right_pshift(self.unit, self.unit, self.prime_pow.pow_ZZ_tmp(-shift)[0], self.prime_pow.get_context(self.relprec).x)
+        else:
+            if shift > 0:
+                self.prime_pow.restore_context_capdiv(self.relprec)
+                mod = self.prime_pow.get_modulus_capdiv(self.relprec)
+                ZZ_pX_PowerXMod_long_pre(tmpP, shift, mod[0])
+                ZZ_pX_MulMod_pre(self.unit, self.unit, tmpP, mod[0])
+            elif shift < 0:
+                self.prime_pow.eis_shift_capdiv(&self.unit, &self.unit, -shift, self.relprec)
+
+    cdef int _pshift_self(self, long shift) except -1:
+        """
+        Multiply ``self`` by ``p^shift``.
+
+        This function assumes that ``self.relprec``, ``self.ordp`` and
+        ``self.unit`` are already set (in the case ``self.prime_pow.e
+        != 1``), and is more reasonable to call externally than
+        ``_internal_lshift``
+
+        EXAMPLES::
+
+            sage: R = Qp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = W(3/25, relprec=6); z
+            3*w^-10 + 3*w^-8 + 2*w^-6 + O(w^-4)
+            sage: z * 25
+            3 + O(w^6)
+        """
+        cdef ZZ_pX_c high_shifter
+        cdef ZZ_pX_Modulus_c *modulus
+        cdef ntl_ZZ_pContext_class c
+        cdef ZZ_pX_c* high_array
+        cdef long i, high_length
+        if self.prime_pow.e == 1:
+            self.ordp += shift
+        else:
+            self.ordp += shift * self.prime_pow.e
+            if shift < 0:
+                shift = -shift
+                c = self.prime_pow.get_context_capdiv(self.relprec)
+                c.restore_c()
+                modulus = self.prime_pow.get_modulus_capdiv(self.relprec)
+                if isinstance(self.prime_pow, PowComputer_ZZ_pX_big_Eis):
+                    high_array = (<PowComputer_ZZ_pX_big_Eis>self.prime_pow).high_shifter
+                elif isinstance(self.prime_pow, PowComputer_ZZ_pX_small_Eis):
+                    high_array = (<PowComputer_ZZ_pX_small_Eis>self.prime_pow).high_shifter
+                else:
+                    raise TypeError("unrecognized PowComputer type")
+                ZZ_pX_conv_modulus(high_shifter, high_array[0], c.x)
+                ZZ_pX_InvMod_newton_ram(high_shifter, high_shifter, modulus[0], c.x)
+                ZZ_pX_PowerMod_long_pre(high_shifter, high_shifter, shift, modulus[0])
+                ZZ_pX_MulMod_pre(self.unit, self.unit, high_shifter, modulus[0])
+            elif shift > 0:
+                i = 0
+                c = self.prime_pow.get_context_capdiv(self.relprec)
+                c.restore_c()
+                modulus = self.prime_pow.get_modulus_capdiv(self.relprec)
+                if isinstance(self.prime_pow, PowComputer_ZZ_pX_big_Eis):
+                    high_array = (<PowComputer_ZZ_pX_big_Eis>self.prime_pow).high_shifter
+                    high_length = (<PowComputer_ZZ_pX_big_Eis>self.prime_pow).high_length
+                elif isinstance(self.prime_pow, PowComputer_ZZ_pX_small_Eis):
+                    high_array = (<PowComputer_ZZ_pX_small_Eis>self.prime_pow).high_shifter
+                    high_length = (<PowComputer_ZZ_pX_small_Eis>self.prime_pow).high_length
+                else:
+                    raise TypeError("unrecognized PowComputer type")
+                if shift >= self.prime_pow.prec_cap:
+                    # high_shifter = p^(2^(high_length - 1))/x^(e*2^(high_length - 1))
+                    ZZ_pX_conv_modulus(high_shifter, high_array[high_length-1], c.x)
+                    # if shift = r + s * 2^(high_length - 1)
+                    # then high_shifter = p^(s*2^(high_length - 1))/x^(e*s*2^(high_length - 1))
+                    ZZ_pX_PowerMod_long_pre(high_shifter, high_shifter, (shift / (1L << (high_length - 1))), modulus[0])
+                    ZZ_pX_MulMod_pre(self.unit, self.unit, high_shifter, modulus[0])
+                    # Now we only need to multiply self.unit by p^r/x^(e*r) where r < 2^(high_length - 1), which is tractable.
+                    shift = shift % (1L << (high_length - 1))
+                while shift > 0:
+                    if shift & 1:
+                        ZZ_pX_conv_modulus(high_shifter, high_array[i], c.x)
+                        ZZ_pX_MulMod_pre(self.unit, self.unit, high_shifter, modulus[0])
+                    shift = shift >> 1
+                    i += 1
+
+    cdef pAdicZZpXCRElement _new_c(self, long relprec):
+        """
+        Return a new element with the same parent as ``self`` and
+        relative precision ``relprec``
+
+        Note that if ``relprec`` is nonpositive, the convention is that
+        ``relprec = 0`` indicates an exact or inexact zero, ``relprec < 0``
+        indicates an unnormalized element.
+
+        EXAMPLES::
+
+            sage: R = Qp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: w^5 + 1  # indirect doctest
+            1 + w^5 + O(w^25)
+        """
+        cdef pAdicZZpXCRElement ans = pAdicZZpXCRElement.__new__(pAdicZZpXCRElement)
+        ans._parent = self._parent
+        ans.prime_pow = self.prime_pow
+        if relprec > 0:
+            self.prime_pow.restore_context_capdiv(relprec)
+            ans.relprec = relprec
+        elif relprec == 0:
+            ans._set_exact_zero()
+        else:
+            self.prime_pow.restore_context_capdiv(-relprec)
+            ans.relprec = relprec
+        return ans
+
+    def __reduce__(self):
+        """
+        Pickles ``self``.
+
+        EXAMPLES::
+
+            sage: R = Qp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = (1 + w)^5 - 1
+            sage: loads(dumps(z)) == z
+            True
+        """
+        cdef Integer relprec, ordp
+        relprec = PY_NEW(Integer)
+        ordp = PY_NEW(Integer)
+        mpz_set_si(relprec.value, self.relprec)
+        mpz_set_si(ordp.value, self.ordp)
+        if self.relprec == 0:
+            return make_ZZpXCRElement, (self.parent(), None, ordp, relprec, 0)
+        self.prime_pow.restore_context_capdiv(self.relprec)
+        cdef ntl_ZZ_pX holder = ntl_ZZ_pX.__new__(ntl_ZZ_pX)
+        holder.c = self.prime_pow.get_context_capdiv(self.relprec)
+        holder.x = self.unit
+        return make_ZZpXCRElement, (self.parent(), holder, ordp, relprec, 0)
+
+    cdef int _cmp_units(left, pAdicGenericElement right) except -2:
+        """
+        For units ``left`` and ``right``, return 0 if they are equal up to
+        the lesser of the two precisions, or 1 if they are not.
+
+        EXAMPLES::
+
+            sage: R = Qp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: w == 1  # indirect doctest
+            False
+            sage: y = 1 + w + O(w^7)
+            sage: z = 1 + w + w^10 + O(w^13)
+            sage: y == z
+            True
+        """
+        # This function needs improvement.  In particular, there are a lot of
+        # speed improvements to be had, and it should be changed so that it
+        # returns 1 only half the time (and -1 the other half) when left and
+        # right are not equal.
+        cdef pAdicZZpXCRElement diff = <pAdicZZpXCRElement> (left - right)
+        diff._normalize()
+        if diff.relprec == 0:
+            return 0
+        # for now, just return 1
+        return 1
+
+    def __invert__(self):
+        """
+        Return the inverse of this element.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = (1 + w)^5
+            sage: y = ~z; y  # indirect doctest
+            1 + 4*w^5 + 4*w^6 + 3*w^7 + w^8 + 2*w^10 + w^11 + w^12 + 2*w^14 + 3*w^16 + 3*w^17 + 4*w^18 + 4*w^19 + 2*w^20 + 2*w^21 + 4*w^22 + 3*w^23 + 3*w^24 + O(w^25)
+            sage: y.parent()
+            5-adic Eisenstein Extension Field in w defined by x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: z = z - 1
+            sage: ~z
+            w^-5 + 4*w^-4 + 4*w^-3 + 4*w^-2 + 2*w^-1 + 1 + w + 4*w^2 + 4*w^3 + 4*w^4 + w^5 + w^6 + w^7 + 4*w^8 + 4*w^9 + 2*w^10 + w^11 + 2*w^12 + 4*w^13 + 4*w^14 + O(w^15)
+            sage: ~z * z
+            1 + O(w^20)
+        """
+        if self._is_exact_zero():
+            raise ZeroDivisionError("cannot divide by zero")
+        if self._is_inexact_zero(): # this calls _normalize
+            raise PrecisionError("cannot divide by something indistinguishable from zero")
+        cdef pAdicZZpXCRElement ans = self._new_c(self.relprec)
+        if not ans.prime_pow.in_field:
+            ans._parent = self._parent.fraction_field()
+            ans.prime_pow = ans._parent.prime_pow
+        ans.ordp = -self.ordp
+        sig_on()
+        if self.prime_pow.e == 1:
+            ZZ_pX_InvMod_newton_unram(ans.unit, self.unit, self.prime_pow.get_modulus(ans.relprec)[0], self.prime_pow.get_context(ans.relprec).x, self.prime_pow.get_context(1).x)
+        else:
+            ZZ_pX_InvMod_newton_ram(ans.unit, self.unit, self.prime_pow.get_modulus_capdiv(ans.relprec)[0], self.prime_pow.get_context_capdiv(ans.relprec).x)
+        sig_off()
+        return ans
+
+    cdef pAdicZZpXCRElement _lshift_c(self, long n):
+        """
+        Multiply ``self`` by the uniformizer raised to the power ``n``.  If
+        ``n`` is negative, right shifts by ``-n``.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = (1 + w)^5
+            sage: z
+            1 + w^5 + w^6 + 2*w^7 + 4*w^8 + 3*w^10 + w^12 + 4*w^13 + 4*w^14 + 4*w^15 + 4*w^16 + 4*w^17 + 4*w^20 + w^21 + 4*w^24 + O(w^25)
+            sage: z << 17  # indirect doctest
+            w^17 + w^22 + w^23 + 2*w^24 + 4*w^25 + 3*w^27 + w^29 + 4*w^30 + 4*w^31 + 4*w^32 + 4*w^33 + 4*w^34 + 4*w^37 + w^38 + 4*w^41 + O(w^42)
+            sage: z << (-1)
+            w^4 + w^5 + 2*w^6 + 4*w^7 + 3*w^9 + w^11 + 4*w^12 + 4*w^13 + 4*w^14 + 4*w^15 + 4*w^16 + 4*w^19 + w^20 + 4*w^23 + O(w^24)
+        """
+        if not self.prime_pow.in_field and n < -self.ordp:
+            return self._rshift_c(-n)
+        check_ordp(n)
+        cdef pAdicZZpXCRElement ans
+        if self._is_exact_zero() or n == 0:
+            return self
+        elif self._is_inexact_zero():
+            ans = self._new_c(0)
+        else:
+            ans = self._new_c(self.relprec)
+            ans.unit = self.unit
+        ans.ordp = self.ordp + n
+        check_ordp(ans.ordp)
+        return ans
+
+    def __lshift__(pAdicZZpXCRElement self, shift):
+        """
+        Multiply ``self`` by the uniformizer raised to the power ``n``.  If
+        ``n`` is negative, right shifts by ``-n``.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = (1 + w)^5
+            sage: z
+            1 + w^5 + w^6 + 2*w^7 + 4*w^8 + 3*w^10 + w^12 + 4*w^13 + 4*w^14 + 4*w^15 + 4*w^16 + 4*w^17 + 4*w^20 + w^21 + 4*w^24 + O(w^25)
+            sage: z << 17  # indirect doctest
+            w^17 + w^22 + w^23 + 2*w^24 + 4*w^25 + 3*w^27 + w^29 + 4*w^30 + 4*w^31 + 4*w^32 + 4*w^33 + 4*w^34 + 4*w^37 + w^38 + 4*w^41 + O(w^42)
+            sage: z << (-1)
+            w^4 + w^5 + 2*w^6 + 4*w^7 + 3*w^9 + w^11 + 4*w^12 + 4*w^13 + 4*w^14 + 4*w^15 + 4*w^16 + 4*w^19 + w^20 + 4*w^23 + O(w^24)
+        """
+        cdef pAdicZZpXCRElement ans
+        if not isinstance(shift, Integer):
+            shift = Integer(shift)
+        if mpz_fits_slong_p((<Integer>shift).value) == 0:
+            if self._is_exact_zero():
+                return self
+            if self.prime_pow.in_field or mpz_sgn((<Integer>shift).value) > 0:
+                raise ValueError("shift does not fit in long")
+            else:
+                ans = self._new_c(0)
+                ans.ordp = 0
+                return ans
+        return self._lshift_c(mpz_get_si((<Integer>shift).value))
+
+    cdef pAdicZZpXCRElement _rshift_c(self, long n):
+        """
+        Divide ``self`` by the uniformizer raised to the power ``n``.  If
+        parent is not a field, throws away the nonpositive part of
+        the series expansion.  If ``n`` is negative, left shifts by ``-n``.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5,print_mode='digits')
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = (1 + w)^5
+            sage: for m in range(26): repr(z >> m)  # indirect doctest
+            '...4001400444441030421100001'
+            '...400140044444103042110000'
+            '...40014004444410304211000'
+            '...4001400444441030421100'
+            '...400140044444103042110'
+            '...40014004444410304211'
+            '...4001400444441030421'
+            '...400140044444103042'
+            '...40014004444410304'
+            '...4001400444441030'
+            '...400140044444103'
+            '...40014004444410'
+            '...4001400444441'
+            '...400140044444'
+            '...40014004444'
+            '...4001400444'
+            '...400140044'
+            '...40014004'
+            '...4001400'
+            '...400140'
+            '...40014'
+            '...4001'
+            '...400'
+            '...40'
+            '...4'
+            '...'
+            sage: repr(z >> (-4))
+            '...40014004444410304211000010000'
+        """
+        if self.prime_pow.in_field or n <= self.ordp:
+            return self._lshift_c(-n)
+        if self._is_exact_zero() or n == 0:
+            return self
+        cdef long arelprec
+        if self.relprec < 0:
+            arelprec = -self.relprec
+        else:
+            arelprec = self.relprec
+        cdef pAdicZZpXCRElement ans
+        if arelprec > n - self.ordp:
+            ans = self._new_c(arelprec - (n - self.ordp))
+            if self.prime_pow.e == 1:
+                ZZ_pX_right_pshift(ans.unit, self.unit, self.prime_pow.pow_ZZ_tmp(n - self.ordp)[0], self.prime_pow.get_context(ans.relprec).x)
+            else:
+                self.prime_pow.eis_shift_capdiv(&ans.unit, &self.unit, n - self.ordp, ans.relprec)
+        else:
+            ans = self._new_c(0)
+        ans.ordp = 0
+        ans.relprec = -ans.relprec
+        return ans
+
+    def __rshift__(pAdicZZpXCRElement self, shift):
+        """
+        Divide ``self`` by the uniformizer raised to the power ``n``.  If
+        parent is not a field, throws away the nonpositive part of
+        the series expansion.  If ``n`` is negative, left shifts by ``-n``.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = (1 + w)^5
+            sage: z
+            1 + w^5 + w^6 + 2*w^7 + 4*w^8 + 3*w^10 + w^12 + 4*w^13 + 4*w^14 + 4*w^15 + 4*w^16 + 4*w^17 + 4*w^20 + w^21 + 4*w^24 + O(w^25)
+            sage: z >> (6)  # indirect doctest
+            1 + 2*w + 4*w^2 + 3*w^4 + w^6 + 4*w^7 + 4*w^8 + 4*w^9 + 4*w^10 + 4*w^11 + 4*w^14 + w^15 + 4*w^18 + O(w^19)
+            sage: z >> (-4)
+            w^4 + w^9 + w^10 + 2*w^11 + 4*w^12 + 3*w^14 + w^16 + 4*w^17 + 4*w^18 + 4*w^19 + 4*w^20 + 4*w^21 + 4*w^24 + w^25 + 4*w^28 + O(w^29)
+            sage: F = W.fraction_field()
+            sage: z = F(z)
+            sage: z >> 7
+            w^-7 + w^-2 + w^-1 + 2 + 4*w + 3*w^3 + w^5 + 4*w^6 + 4*w^7 + 4*w^8 + 4*w^9 + 4*w^10 + 4*w^13 + w^14 + 4*w^17 + O(w^18)
+        """
+        cdef pAdicZZpXCRElement ans
+        if not isinstance(shift, Integer):
+            shift = Integer(shift)
+        if mpz_fits_slong_p((<Integer>shift).value) == 0:
+            if self._is_exact_zero():
+                return self
+            if self.prime_pow.in_field or mpz_sgn((<Integer>shift).value) < 0:
+                raise ValueError("valuation overflow")
+            else:
+                ans = self._new_c(0)
+                ans.ordp = 0
+                return ans
+        return self._rshift_c(mpz_get_si((<Integer>shift).value))
+
+    cpdef _neg_(self):
+        """
+        Negation.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: z = (1 + w)^5; z
+            1 + w^5 + w^6 + 2*w^7 + 4*w^8 + 3*w^10 + w^12 + 4*w^13 + 4*w^14 + 4*w^15 + 4*w^16 + 4*w^17 + 4*w^20 + w^21 + 4*w^24 + O(w^25)
+            sage: -z  # indirect doctest
+            4 + 3*w^5 + 4*w^6 + w^7 + w^8 + w^9 + w^10 + w^11 + 2*w^12 + 4*w^13 + 4*w^15 + 3*w^16 + w^17 + 2*w^18 + 3*w^19 + 2*w^21 + 4*w^23 + 4*w^24 + O(w^25)
+            sage: y = z + (-z); y
+            O(w^25)
+            sage: -y
+            O(w^25)
+            sage: -W(0)
+            0
+        """
+        cdef pAdicZZpXCRElement ans = self._new_c(self.relprec)
+        ans.ordp = self.ordp
+        if self.relprec != 0:
+            self.prime_pow.restore_context_capdiv(self.relprec)
+            ZZ_pX_negate(ans.unit, self.unit)
+        return ans
+
+#                                             / 1 + \alpha^p \pi_K^{p \lambda}                      mod \mathfrak{p}_K^{p \lambda + 1}   if 1 \le \lambda < \frac{e_K}{p-1}
+#        (1 + \alpha \pi^{\lambda})^p \equiv {  1 + (\alpha^p - \epsilon \alpha) \pi_K^{p \lambda}  mod \mathfrak{p}_K^{p \lambda + 1}   if \lambda = \frac{e_K}{p-1}
+#                                             \ 1 - \epsilon \alpha \pi_K^{\lambda + e}             mod \mathfrak{p}_K^{\lambda + e + 1} if \lambda > \frac{e_K}{p-1}
+
+    def __pow__(pAdicZZpXCRElement self, _right, m): # m ignored
+        r"""
+        Compute ``self^right``.
+
+        Note: when ``right`` is divisible by `p` then one can get more
+        precision than expected.
+
+        Lemma 2.1 (Constructing Class Fields over Local Fields, Sebastian Pauli):
+
+        Let `\alpha` be in `\mathcal{O}_K`.  Let
+
+        .. MATH::
+
+            p = -\pi_K^{e_K} \epsilon
+
+        be the factorization of `p` where `\epsilon` is a unit.  Then
+        the `p`-th power of `1 + \alpha \pi_K^{\lambda}` satisfies
+
+        .. MATH::
+
+            (1 + \alpha \pi^{\lambda})^p \equiv \left{ \begin{array}{lll}
+            1 + \alpha^p \pi_K^{p \lambda} & \mod \mathfrak{p}_K^{p \lambda + 1} & \mbox{if $1 \le \lambda < \frac{e_K}{p-1}$} \\
+            1 + (\alpha^p - \epsilon \alpha) \pi_K^{p \lambda} & \mod \mathfrak{p}_K^{p \lambda + 1} & \mbox{if $\lambda = \frac{e_K}{p-1}$} \\
+            1 - \epsilon \alpha \pi_K^{\lambda + e} & \mod \mathfrak{p}_K^{\lambda + e + 1} & \mbox{if $\lambda > \frac{e_K}{p-1}$}
+            \end{array} \right.
+
+
+        So if ``right`` is divisible by `p^k` we can multiply the
+        relative precision by `p` until we exceed `e/(p-1)`, then add
+        `e` until we have done a total of `k` things: the precision of
+        the result can therefore be greater than the precision of
+        ``self``.
+
+        There is also the issue of `p`-adic exponents, and determining
+        how the precision of the exponent affects the precision of the
+        result.
+
+        In computing `(a + O(\pi^k))^{b + O(p^m)}`, one needs that the
+        reduction of `a` mod `\pi` is in the prime field
+        `\GF{p}` (so that the `p^m` power of the Teichmuller
+        part is constant as `m` increases).  Given this restriction,
+        we can factor out the Teichmuller part and use the above lemma
+        to find the first spot where
+
+        .. MATH::
+
+            (1 + \alpha \pi^{\lambda})^{p^m}
+
+        differs from 1.  We compare this with the precision bound
+        given by computing `(a + O(\pi^k))^b` and take the lesser of
+        the two.
+
+        In order to do this we need to compute the valuation of ``(self
+        / self.parent().teichmuller(self)) - 1``.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: (1 + w)^5  # indirect doctest
+            1 + w^5 + w^6 + 2*w^7 + 4*w^8 + 3*w^10 + w^12 + 4*w^13 + 4*w^14 + 4*w^15 + 4*w^16 + 4*w^17 + 4*w^20 + w^21 + 4*w^24 + O(w^25)
+            sage: (1 + w)^-5
+            1 + 4*w^5 + 4*w^6 + 3*w^7 + w^8 + 2*w^10 + w^11 + w^12 + 2*w^14 + 3*w^16 + 3*w^17 + 4*w^18 + 4*w^19 + 2*w^20 + 2*w^21 + 4*w^22 + 3*w^23 + 3*w^24 + O(w^25)
+            sage: (1 + w + O(w^19))^5
+            1 + w^5 + w^6 + 2*w^7 + 4*w^8 + 3*w^10 + w^12 + 4*w^13 + 4*w^14 + 4*w^15 + 4*w^16 + 4*w^17 + 4*w^20 + w^21 + O(w^24)
+            sage: (1 + O(w))^5
+            1 + O(w^5)
+            sage: (1 + w + O(w^3))^25
+            1 + w^10 + w^11 + 4*w^12 + O(w^13)
+            sage: (3 + 2*w + w^2 + O(w^6))^(15 + O(125))
+            2 + 4*w^6 + w^7 + 3*w^8 + 3*w^9 + 4*w^10 + O(w^11)
+            sage: (3 + 2*w + w^2 + O(w^6))^(15 + O(25))
+            2 + 4*w^6 + w^7 + 3*w^8 + 3*w^9 + O(w^10)
+            sage: (3 + w^2 + O(w^6))^(15+O(25))
+            2 + w^5 + 4*w^7 + w^9 + 3*w^10 + O(w^11)
+            sage: R = Zp(2, 10)
+            sage: S.<x> = R[]
+            sage: f = x^34 + 18*x^5 - 72*x^3 + 2
+            sage: W.<w> = R.ext(f)
+            sage: (1+w+O(w^2))^8
+            1 + w^8 + O(w^16)
+            sage: (1+w+O(w^2))^16
+            1 + w^16 + O(w^32)
+            sage: (1+w+O(w^2))^32
+            1 + w^32 + w^50 + w^55 + w^60 + O(w^64)
+            sage: (1+w+O(w^2))^64
+            1 + w^64 + w^66 + w^71 + w^76 + w^81 + w^84 + w^86 + w^91 + w^94 + w^96 + O(w^98)
+
+        TESTS:
+
+            We define ``0^0`` to be unity, :issue:`13786`::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: type(W(0))
+            <class 'sage.rings.padics.padic_ZZ_pX_CR_element.pAdicZZpXCRElement'>
+            sage: W(0)^0
+            1 + O(w^25)
+            sage: W(0)^0 == W(1)
+            True
+
+        The value returned from ``0^0`` should belong to our ring::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: type(W(0)^0) == type(W(0))
+            True
+        """
+        self._normalize()
+        cdef Integer right
+        cdef bint padic_exp
+        cdef long exp_prec
+        cdef long exp_val
+        cdef long relprec
+        cdef long threshold # e / (p-1)
+        cdef mpz_t tmp, tmp2
+        if mpz_fits_slong_p(self.prime_pow.prime.value) == 0:
+            threshold = 0
+        else:
+            threshold = self.prime_pow.e / (mpz_get_si(self.prime_pow.prime.value) - 1)
+        cdef Integer base_level
+        cdef pAdicZZpXCRElement ans
+        cdef long i
+        if self._is_exact_zero():
+            # Return 0 except for 0^0 error or type error on the exponent.
+            if isinstance(_right, (Integer, Rational, int)) or (isinstance(_right, pAdicGenericElement) and _right._is_base_elt(self.prime_pow.prime)):
+                if _right == 0:
+                    return self.parent(1)
+                return self
+            else:
+                raise TypeError("exponent must be an integer, rational or base p-adic with the same prime")
+        elif self._is_inexact_zero():
+            # If an integer exponent, return an inexact zero of valuation right * self.ordp.  Otherwise raise an error.
+            if isinstance(_right, int):
+                _right = Integer(_right)
+            if isinstance(_right, Integer):
+                ans = self._new_c(0)
+                mpz_init_set_si(tmp, self.ordp)
+                mpz_mul(tmp, tmp, (<Integer>_right).value)
+                if mpz_cmp_si(tmp, maxordp) >= 0 or mpz_cmp_si(tmp, -maxordp) <= 0:
+                    raise ValueError("valuation overflow")
+                ans.ordp = mpz_get_si(tmp)
+                mpz_clear(tmp)
+                return ans
+            elif isinstance(_right, Rational) or (isinstance(_right, pAdicGenericElement) and _right._is_base_elt(self.prime_pow.prime)):
+                raise ValueError("Need more precision")
+            else:
+                raise TypeError("exponent must be an integer, rational or base p-adic with the same prime")
+        if isinstance(_right, int):
+            _right = Integer(_right)
+        if isinstance(_right, Integer):
+            right = <Integer> _right
+            if right == 0:
+                # return 1 to maximum precision
+                ans = self._new_c(self.prime_pow.ram_prec_cap)
+                ans.ordp = 0
+                ZZ_pX_SetCoeff_long(ans.unit, 0, 1)
+                return ans
+            padic_exp = False
+            exp_val = _right.valuation(self.prime_pow.prime) ##
+        elif isinstance(_right, pAdicGenericElement) and _right._is_base_elt(self.prime_pow.prime):
+            if self.ordp != 0:
+                raise ValueError("in order to raise to a p-adic exponent, base must be a unit")
+            right = Integer(_right)
+            padic_exp = True
+            exp_prec = _right.precision_absolute() ##
+            exp_val = _right.valuation() ##
+            if exp_val < 0:
+                raise NotImplementedError("negative valuation exponents not yet supported")
+            # checks to see if the residue of self.unit is in the prime field.
+            if self.prime_pow.e == 1:
+                for i from 1 <= i <= ZZ_pX_deg(self.unit):
+                    if not ZZ_divide_test(ZZ_p_rep(ZZ_pX_coeff(self.unit, i)), self.prime_pow.pow_ZZ_tmp(1)[0]):
+                        raise ValueError("in order to raise to a p-adic exponent, base must reduce to an element of F_p mod the uniformizer")
+            # compute the "level"
+            teich_part = self.parent().teichmuller(self)
+            base_level = (self / teich_part - 1).valuation() ##
+        elif isinstance(_right, Rational):
+            raise NotImplementedError
+        else:
+            raise TypeError("exponent must be an integer, rational or base p-adic with the same prime")
+        # Now we compute the increased relprec due to the exponent having positive p-adic valuation
+        if exp_val > 0:
+            mpz_init_set_si(tmp, self.relprec)
+            while mpz_cmp_si(tmp, threshold) <= 0 and exp_val > 0:
+                mpz_mul(tmp, tmp, self.prime_pow.prime.value)
+                exp_val -= 1
+            if exp_val > 0:
+                mpz_init_set_si(tmp2, self.prime_pow.e)
+                mpz_addmul_ui(tmp, tmp2, exp_val)
+                mpz_clear(tmp2)
+            if mpz_cmp_si(tmp, self.prime_pow.ram_prec_cap) > 0:
+                relprec = self.prime_pow.ram_prec_cap
+            else:
+                relprec = mpz_get_si(tmp)
+            mpz_clear(tmp)
+        else:
+            relprec = self.relprec
+        # Now we compute the limit on relprec due to a non-infinite precision on the exponent.
+        if padic_exp:
+            if exp_prec > 0:
+                # I can freely change base_level, so I use it in place of tmp above.
+                while mpz_cmp_si(base_level.value, threshold) <= 0 and exp_prec > 0:
+                    mpz_mul(base_level.value, base_level.value, self.prime_pow.prime.value)
+                    exp_prec -= 1
+                if exp_prec > 0:
+                    mpz_init_set_si(tmp2, self.prime_pow.e)
+                    mpz_addmul_ui(base_level.value, tmp2, exp_prec)
+                    mpz_clear(tmp2)
+                if mpz_cmp_si(base_level.value, relprec) < 0:
+                    relprec = mpz_get_si(base_level.value)
+            else:
+                ans = self._new_c(0)
+                ans.ordp = 0
+                return ans
+        ans = self._new_c(relprec)
+        if self.ordp == 0:
+            ans.ordp = 0
+        else:
+            mpz_init_set(tmp, right.value)
+            mpz_mul_si(tmp, tmp, self.ordp)
+            if mpz_cmp_si(tmp, maxordp) >= 0 or mpz_cmp_si(tmp, -maxordp) <= 0:
+                raise ValueError("valuation overflow")
+            ans.ordp = mpz_get_si(tmp)
+            mpz_clear(tmp)
+        cdef ntl_ZZ rZZ = ntl_ZZ.__new__(ntl_ZZ)
+        mpz_to_ZZ(&rZZ.x, right.value)
+        sig_on()
+        if mpz_sgn(right.value) < 0:
+            if self.prime_pow.e == 1:
+                ZZ_pX_InvMod_newton_unram(ans.unit, self.unit, self.prime_pow.get_modulus(ans.relprec)[0], self.prime_pow.get_context(ans.relprec).x, self.prime_pow.get_context(1).x)
+            else:
+                ZZ_pX_InvMod_newton_ram(ans.unit, self.unit, self.prime_pow.get_modulus_capdiv(ans.relprec)[0], self.prime_pow.get_context_capdiv(ans.relprec).x)
+            ZZ_negate(rZZ.x, rZZ.x)
+            ZZ_pX_PowerMod_pre(ans.unit, ans.unit, rZZ.x, self.prime_pow.get_modulus_capdiv(ans.relprec)[0])
+        else:
+            ZZ_pX_PowerMod_pre(ans.unit, self.unit, rZZ.x, self.prime_pow.get_modulus_capdiv(ans.relprec)[0])
+        sig_off()
+        return ans
+
+    cpdef _add_(self, _right):
+        """
+        Compute the sum of ``self`` and ``right``.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: (4*w^5 + 3*w^7 + w^9 + 2*w^10 + 2*w^11 + O(w^13)) - 69  # indirect doctest
+            1 + O(w^13)
+            sage: -69 + (4*w^5 + 3*w^7 + w^9 + 2*w^10 + 2*w^11 + O(w^13))
+            1 + O(w^13)
+            sage: y = (4*w^5 + 3*w^7 + w^9 + 2*w^10 + 2*w^11 + O(w^13))
+            sage: y - 70
+            O(w^13)
+            sage: y + 0
+            4*w^5 + 3*w^7 + w^9 + 2*w^10 + 2*w^11 + O(w^13)
+        """
+        cdef pAdicZZpXCRElement right = <pAdicZZpXCRElement>_right
+        cdef pAdicZZpXCRElement ans
+        cdef long tmpL
+        cdef ZZ_pX_c tmpP
+        if self.relprec == 0:
+            if self.ordp >= right.ordp + right.relprec: # or self._is_exact_zero()
+                return right
+            elif self.ordp <= right.ordp:
+                ans = self._new_c(0)
+                ans.ordp = self.ordp
+            else:
+                ans = self._new_c(self.ordp - right.ordp)
+                ZZ_pX_conv_modulus(ans.unit, right.unit, self.prime_pow.get_context_capdiv(ans.relprec).x)
+                if right.relprec < 0:
+                    ans.relprec = -ans.relprec
+                ans.ordp = right.ordp
+            return ans
+        if right.relprec == 0:
+            if right.ordp >= self.ordp + self.relprec: # or right._is_exact_zero()
+                return self
+            elif right.ordp <= self.ordp:
+                ans = self._new_c(0)
+                ans.ordp = right.ordp
+            else:
+                ans = self._new_c(right.ordp - self.ordp)
+                ZZ_pX_conv_modulus(ans.unit, self.unit, self.prime_pow.get_context_capdiv(ans.relprec).x)
+                if self.relprec < 0:
+                    ans.relprec = -ans.relprec
+                ans.ordp = self.ordp
+            return ans
+        cdef long srprec = self.relprec
+        if srprec < 0:
+            srprec = -srprec
+        cdef long rrprec = right.relprec
+        if rrprec < 0:
+            rrprec = -rrprec
+        if self.ordp == right.ordp:
+            # The relative precision of the sum is the minimum of the relative precisions in this case, possibly decreasing if we got cancellation
+            # Since the valuations are the same, we could just add the units, if they had the same modulus.
+            # But they don't necessarily, so we may have to conv_modulus
+            if srprec == rrprec:
+                ans = self._new_c(-srprec) # -srprec indicates that ans is not normalized
+                self.prime_pow.restore_context_capdiv(srprec)
+                ZZ_pX_add(ans.unit, self.unit, right.unit)
+            elif srprec < rrprec:
+                ans = self._new_c(-srprec)
+                ZZ_pX_conv_modulus(ans.unit, right.unit, self.prime_pow.get_context_capdiv(srprec).x)
+                self.prime_pow.restore_context_capdiv(srprec)
+                # conv_modulus should have restored the context, so we don't need to again.
+                ZZ_pX_add(ans.unit, ans.unit, self.unit)
+            else:
+                ans = self._new_c(-rrprec)
+                ZZ_pX_conv_modulus(ans.unit, self.unit, self.prime_pow.get_context_capdiv(rrprec).x)
+                self.prime_pow.restore_context_capdiv(rrprec)
+                # conv_modulus should have restored the context, so we don't need to again.
+                ZZ_pX_add(ans.unit, ans.unit, right.unit)
+            ans.ordp = self.ordp
+        elif self.ordp < right.ordp:
+            tmpL = right.ordp - self.ordp
+            if tmpL >= srprec:
+                return self
+            if srprec <= tmpL + rrprec:
+                ans = self._new_c(-srprec)
+            else:
+                ans = self._new_c(-tmpL - rrprec)
+            ans.ordp = self.ordp
+            ZZ_pX_conv_modulus(ans.unit, right.unit, self.prime_pow.get_context_capdiv(ans.relprec).x)
+            ans._internal_lshift(tmpL)
+            if srprec <= tmpL + rrprec:
+                ZZ_pX_add(ans.unit, ans.unit, self.unit)
+            else:
+                ZZ_pX_conv_modulus(tmpP, self.unit, self.prime_pow.get_context_capdiv(ans.relprec).x)
+                ZZ_pX_add(ans.unit, ans.unit, tmpP)
+            # if self is normalized, then the valuations are actually different so the sum will be normalized.
+            if self.relprec > 0:
+                ans.relprec = -ans.relprec
+        else:
+            tmpL = self.ordp - right.ordp
+            if tmpL >= rrprec:
+                return right
+            if rrprec <= tmpL + srprec:
+                ans = self._new_c(-rrprec)
+            else:
+                ans = self._new_c(-tmpL - srprec)
+            ans.ordp = right.ordp
+            ZZ_pX_conv_modulus(ans.unit, self.unit, self.prime_pow.get_context_capdiv(ans.relprec).x)
+            ans._internal_lshift(tmpL)
+            if rrprec <= tmpL + srprec:
+                ZZ_pX_add(ans.unit, ans.unit, right.unit)
+            else:
+                ZZ_pX_conv_modulus(tmpP, right.unit, self.prime_pow.get_context_capdiv(ans.relprec).x)
+                ZZ_pX_add(ans.unit, ans.unit, tmpP)
+            # if right is normalized, then the valuations are actually different so the sum will be normalized.
+            if right.relprec > 0:
+                ans.relprec = -ans.relprec
+        return ans
+
+    cpdef _sub_(self, right):
+        """
+        Return the difference of two elements.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: a = W(329)
+            sage: b = W(111)
+            sage: a - b  # indirect doctest
+            3 + 3*w^5 + w^7 + 2*w^9 + 3*w^10 + 4*w^11 + 2*w^13 + 2*w^14 + w^15 + 4*w^16 + 2*w^18 + 3*w^19 + 2*w^20 + 3*w^21 + w^22 + w^24 + O(w^25)
+            sage: W(218)
+            3 + 3*w^5 + w^7 + 2*w^9 + 3*w^10 + 4*w^11 + 2*w^13 + 2*w^14 + w^15 + 4*w^16 + 2*w^18 + 3*w^19 + 2*w^20 + 3*w^21 + w^22 + w^24 + O(w^25)
+            sage: a - O(w^14)
+            4 + 3*w^10 + 2*w^12 + O(w^14)
+            sage: a - 0
+            4 + 3*w^10 + 2*w^12 + w^14 + 2*w^15 + w^16 + 3*w^17 + 3*w^18 + w^19 + 2*w^21 + 4*w^22 + w^23 + 4*w^24 + O(w^25)
+            sage: O(w^14) - a
+            1 + 4*w^5 + 3*w^7 + w^9 + w^10 + 2*w^11 + w^12 + w^13 + O(w^14)
+        """
+        # For now, a simple implementation
+        return self + (-right)
+
+    cpdef _mul_(self, _right):
+        """
+        Return the product of two elements.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: a = W(329)
+            sage: b = W(111)
+            sage: a*b  # indirect doctest
+            4 + 3*w^5 + w^7 + 2*w^9 + 4*w^11 + 3*w^12 + 2*w^13 + w^14 + 2*w^15 + 3*w^16 + 4*w^17 + 4*w^18 + 2*w^19 + 2*w^21 + 4*w^22 + 2*w^23 + w^24 + O(w^25)
+            sage: a * 0
+            0
+            sage: a * O(w^14)
+            O(w^14)
+        """
+        cdef pAdicZZpXCRElement right = <pAdicZZpXCRElement>_right
+        cdef ZZ_pX_c modulus_corrected
+        cdef pAdicZZpXCRElement ans
+        if self._is_exact_zero():
+            return self
+        if right._is_exact_zero():
+            return right
+        self._normalize()
+        right._normalize()
+        if self.relprec <= right.relprec:
+            ans = self._new_c(self.relprec)
+        else:
+            ans = self._new_c(right.relprec)
+        ans.ordp = self.ordp + right.ordp
+        check_ordp(ans.ordp)
+        if ans.relprec == 0:
+            return ans
+        if self.relprec == right.relprec:
+            self.prime_pow.restore_context_capdiv(ans.relprec)
+            sig_on()
+            ZZ_pX_MulMod_pre(ans.unit, self.unit, right.unit, self.prime_pow.get_modulus_capdiv(ans.relprec)[0])
+            sig_off()
+        elif self.relprec < right.relprec:
+            sig_on()
+            ZZ_pX_conv_modulus(modulus_corrected, right.unit, self.prime_pow.get_context_capdiv(ans.relprec).x)
+            ZZ_pX_MulMod_pre(ans.unit, self.unit, modulus_corrected, self.prime_pow.get_modulus_capdiv(ans.relprec)[0])
+            sig_off()
+        else:
+            sig_on()
+            ZZ_pX_conv_modulus(modulus_corrected, self.unit, self.prime_pow.get_context_capdiv(ans.relprec).x)
+            ZZ_pX_MulMod_pre(ans.unit, right.unit, modulus_corrected, self.prime_pow.get_modulus_capdiv(ans.relprec)[0])
+            sig_off()
+        return ans
+
+    cpdef _div_(self, right):
+        """
+        Return the quotient of two elements.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: W(14) / W(125)  # indirect doctest
+            4*w^-15 + w^-13 + 3*w^-11 + 2*w^-10 + 3*w^-9 + 4*w^-8 + 4*w^-7 + 3*w^-6 + 2*w^-5 + 4*w^-4 + 3*w^-3 + 2*w^-2 + 4*w^-1 + 2 + w^2 + w^4 + 4*w^5 + w^6 + w^7 + 3*w^9 + O(w^10)
+            sage: 1 / w
+            w^-1 + O(w^24)
+            sage: W.<w> = R.ext(x^25 - 165*x + 5)
+            sage: a = (1 + w)^25 - 1
+            sage: b = (1 + w)^5 - 1
+            sage: c = (1 + w)^20 + (1 + w)^15 + (1 + w)^10 + (1 + w)^5 + 1
+            sage: d = a / b; d == c
+            True
+            sage: d.precision_absolute()
+            120
+            sage: c.precision_absolute()
+            125
+            sage: 1 / a == ~a
+            True
+        """
+        # for now, a simple implementation
+        return self * (~right)
+
+    def __copy__(self):
+        """
+        Return a copy of this element.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: b = W(45, 17); b
+            4*w^5 + 3*w^7 + w^9 + w^10 + 2*w^11 + w^12 + w^13 + 3*w^14 + w^16 + O(w^17)
+            sage: c = copy(b); c
+            4*w^5 + 3*w^7 + w^9 + w^10 + 2*w^11 + w^12 + w^13 + 3*w^14 + w^16 + O(w^17)
+            sage: c is b
+            False
+        """
+        cdef pAdicZZpXCRElement ans = self._new_c(self.relprec)
+        ans.ordp = self.ordp
+        ans.unit = self.unit
+        return ans
+
+    def _integer_(self, Z=None):
+        r"""
+        Return an integer congruent to this element modulo
+        `\pi^{\mathrm{self.absolute_precision()}`, if possible.
+
+        EXAMPLES::
+
+            sage: ZZ(ZqCR(125,names='a')(-1))  # indirect doctest
+            95367431640624
+            sage: R = Zp(5); S.<x> = ZZ[]; f = x^5 + 25*x^3 - 5; W.<w> = R.ext(f)
+            sage: ZZ(W(-1))
+            95367431640624
+            sage: ZZ(W(0))
+            0
+            sage: ZZ(W(0,7))
+            0
+            sage: ZZ(w)
+            Traceback (most recent call last):
+            ...
+            ValueError: this element not well approximated by an integer
+            sage: ZZ(W(5))  # todo: this should be different...
+            381469726562505
+        """
+        cdef Integer ans
+        cdef ZZ_c tmp_z
+        if self._is_exact_zero() or self.relprec == 0:
+            ans = PY_NEW(Integer)
+            return ans
+        if self.ordp < 0:
+            self._normalize()
+            if self.ordp < 0:
+                raise ValueError("this element has negative valuation")
+        cdef ntl_ZZ_pX f = <ntl_ZZ_pX>self._ntl_rep_abs()[0]
+        if f.degree() > 0:
+            raise ValueError("this element not well approximated by an integer")
+        ans = PY_NEW(Integer)
+        tmp_z = ZZ_p_rep(ZZ_pX_ConstTerm(f.x))
+        ZZ_to_mpz(ans.value, &tmp_z)
+        return ans
+
+    def is_zero(self, absprec=None):
+        r"""
+        Return whether the valuation of this element is at least
+        ``absprec``.  If ``absprec`` is ``None``, checks if this element
+        is indistinguishable from zero.
+
+        If this element is an inexact zero of valuation less than ``absprec``,
+        raises a :exc:`PrecisionError`.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: O(w^189).is_zero()
+            True
+            sage: W(0).is_zero()
+            True
+            sage: a = W(675)
+            sage: a.is_zero()
+            False
+            sage: a.is_zero(7)
+            True
+            sage: a.is_zero(21)
+            False
+        """
+        cdef bint ans
+        cdef long aprec
+        self._normalize()
+        if self._is_exact_zero():
+            ans = True
+        elif absprec is None:
+            ans = (self.relprec == 0)
+        else:
+            if not isinstance(absprec, Integer):
+                absprec = Integer(absprec)
+            if mpz_fits_slong_p((<Integer>absprec).value) == 0:
+                if mpz_sgn((<Integer>absprec).value) < 0:
+                    ans = True
+                elif self.relprec == 0:
+                    raise PrecisionError("not enough precision to determine if element is zero")
+                else:
+                    ans = False
+            else:
+                aprec = mpz_get_si((<Integer>absprec).value)
+                if self.relprec == 0 and aprec > self.ordp:
+                    raise PrecisionError("not enough precision to determine if element is zero")
+                else:
+                    ans = (self.ordp >= aprec)
+        return ans
+
+    cpdef ntl_ZZ_pX _ntl_rep_unnormalized(self):
+        """
+        Return an ``ntl_ZZ_pX`` holding the current unit part of this element.
+
+        The element is not normalized before this operation, so the polynomial
+        returned may not actually be a unit.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: a = W(566); b = W(209)
+            sage: c = a + b; c._ntl_rep_unnormalized()
+            [775]
+            sage: c
+            w^10 + 4*w^12 + 2*w^14 + w^15 + 2*w^16 + 4*w^17 + w^18 + w^20 + 2*w^21 + 3*w^22 + w^23 + w^24 + O(w^25)
+            sage: c._ntl_rep_unnormalized()
+            [106 60 114 35 112]
+        """
+        if self.relprec == 0:
+            raise ValueError("self == 0")
+        self.prime_pow.restore_context_capdiv(self.relprec)
+        cdef ntl_ZZ_pX ans = ntl_ZZ_pX.__new__(ntl_ZZ_pX)
+        ans.c = self.prime_pow.get_context_capdiv(self.relprec)
+        ans.x = self.unit
+        return ans
+
+    cpdef ntl_ZZ_pX _ntl_rep(self):
+        """
+        Return an ``ntl_ZZ_pX`` that holds the unit part of this element.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: a = W(566); b = W(209)
+            sage: c = a + b; c._ntl_rep()
+            [106 60 114 35 112]
+            sage: c
+            w^10 + 4*w^12 + 2*w^14 + w^15 + 2*w^16 + 4*w^17 + w^18 + w^20 + 2*w^21 + 3*w^22 + w^23 + w^24 + O(w^25)
+            sage: c._ntl_rep()
+            [106 60 114 35 112]
+        """
+        self._normalize()
+        return self._ntl_rep_unnormalized()
+
+    cpdef _ntl_rep_abs(self):
+        """
+        Return a pair ``(f, k)`` where ``f`` is an ``ntl_ZZ_pX`` and ``k`` is a
+        nonpositive integer such that ``self = f(self.parent.gen())*p^k``.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: a = W(566); b = W(209)
+            sage: c = a + b; c._ntl_rep_abs()
+            ([775], 0)
+            sage: c
+            w^10 + 4*w^12 + 2*w^14 + w^15 + 2*w^16 + 4*w^17 + w^18 + w^20 + 2*w^21 + 3*w^22 + w^23 + w^24 + O(w^25)
+            sage: c._ntl_rep_abs()
+            ([775], 0)
+            sage: (~c)._ntl_rep_abs()
+            ([121], -2)
+            sage: ~c
+            w^-10 + w^-8 + 4*w^-6 + 4*w^-5 + 3*w^-3 + 4*w^-2 + 3*w^-1 + 4 + 4*w + 2*w^2 + 4*w^3 + 3*w^4 + O(w^5)
+            sage: ~c * 25
+            1 + 4*w^5 + 3*w^7 + w^9 + 4*w^10 + 2*w^11 + 3*w^12 + w^13 + 4*w^14 + O(w^15)
+            sage: W(121)
+            1 + 4*w^5 + 3*w^7 + w^9 + 4*w^10 + 2*w^11 + 3*w^12 + w^13 + 4*w^14 + 2*w^16 + 3*w^17 + 3*w^18 + 4*w^19 + 4*w^20 + 3*w^21 + w^22 + w^23 + 4*w^24 + O(w^25)
+        """
+        self._normalize()
+        if self.ordp == 0:
+            return self._ntl_rep(), Integer(0)
+        cdef ntl_ZZ_pContext_class ctx
+        cdef long little_shift, ppow
+        if self.ordp > 0:
+            ctx = self.prime_pow.get_context_capdiv(self.ordp + self.relprec)
+        else:
+            little_shift = ((-self.ordp) % self.prime_pow.e)
+            if little_shift != 0:
+                little_shift = self.prime_pow.e - little_shift
+            ctx = self.prime_pow.get_context_capdiv(self.relprec + little_shift)
+        ctx.restore_c()
+        cdef pAdicZZpXCRElement dummy = pAdicZZpXCRElement.__new__(pAdicZZpXCRElement)
+        cdef ntl_ZZ_pX ans = ntl_ZZ_pX.__new__(ntl_ZZ_pX)
+        cdef Integer ans_k = PY_NEW(Integer)
+        dummy.unit = self.unit
+        dummy.prime_pow = self.prime_pow
+        if self.ordp > 0:
+            dummy.relprec = self.ordp + self.relprec
+            dummy._internal_lshift(self.ordp)
+            ans.x = dummy.unit
+        else:
+            ppow = (self.ordp - little_shift) / self.prime_pow.e
+            mpz_set_si(ans_k.value, ppow)
+            dummy.ordp = 0 # _pshift_self wants ordp set
+            dummy.relprec = self.relprec + little_shift
+            # self = x^(self.prime_pow.e * ppow) * x^(little_shift) * self.unit
+            # so we want to _internal_lshift dummy.unit by little_shift
+            dummy._internal_lshift(little_shift)
+            # and then write
+            # self = p^(ppow) * (x^e/p)^(ppow) * dummy.unit
+            # so we need to multiply dummy.unit by (p/x^e)^(-ppow) in the Eisenstein case
+            # which we can do by _pshift_self
+            dummy._pshift_self(-ppow)
+            ans.x = dummy.unit
+        ans.c = ctx
+        return ans, ans_k
+
+    def _polynomial_list(self, pad=False):
+        """
+        Return the coefficient list for a polynomial over the base ring
+        yielding this element.
+
+        INPUT:
+
+        - ``pad`` -- whether to pad the result with zeros of the appropriate precision
+
+        EXAMPLES::
+
+            sage: R.<x> = ZZ[]
+            sage: W.<w> = Qp(5).extension(x^3 - 5)
+            sage: (1 + w + O(w^11))._polynomial_list()
+            [1 + O(5^4), 1 + O(5^4)]
+            sage: (1 + w + O(w^11))._polynomial_list(pad=True)
+            [1 + O(5^4), 1 + O(5^4), O(5^3)]
+            sage: W(0)._polynomial_list()
+            []
+            sage: W(0)._polynomial_list(pad=True)
+            [0, 0, 0]
+            sage: W(O(w^7))._polynomial_list()
+            []
+            sage: W(O(w^7))._polynomial_list(pad=True)
+            [O(5^3), O(5^2), O(5^2)]
+            sage: T.<a> = Qp(5).extension(x^2 - 5)
+            sage: T(1/5)._polynomial_list()
+            [5^-1 + O(5^19)]
+            sage: T(a^-800)._polynomial_list()
+            [5^-400 + O(5^-380)]
+            sage: T(O(a^-2))._polynomial_list()
+            []
+        """
+        R = self.base_ring()
+        if self.is_zero():
+            L = []
+            k = 0
+        else:
+            f, k = self._ntl_rep_abs()
+            L = [Integer(c) for c in f.list()]
+        if pad:
+            n = self.parent().degree()
+            L.extend([R.zero()] * (n - len(L)))
+        if self._is_exact_zero():
+            return L
+        prec = self.relprec + self.ordp
+        e = self.parent().e()
+        if e == 1:
+            return [R(c, prec-k) << k for c in L]
+        else:
+            return [R(c, (((prec - i - 1) // e) + 1) - k) << k for i, c in enumerate(L)]
+
+    def polynomial(self, var='x'):
+        """
+        Return a polynomial over the base ring that yields this element
+        when evaluated at the generator of the parent.
+
+        INPUT:
+
+        - ``var`` -- string, the variable name for the polynomial
+
+        EXAMPLES::
+
+            sage: S.<x> = ZZ[]
+            sage: W.<w> = Zp(5).extension(x^2 - 5)
+            sage: (w + W(5, 7)).polynomial()
+            (1 + O(5^3))*x + 5 + O(5^4)
+        """
+        R = self.base_ring()
+        S = R[var]
+        return S(self._polynomial_list())
+
+    cdef ZZ_p_c _const_term(self) noexcept:
+        """
+        Return the constant term of ``self.unit``.
+
+        Note: this may be divisible by `p` if ``self`` is not normalized.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: a = W(566)
+            sage: a._const_term_test()  # indirect doctest
+            566
+        """
+        return ZZ_pX_ConstTerm((<pAdicZZpXCRElement>self).unit)
+
+    def is_equal_to(self, right, absprec=None):
+        """
+        Return whether this element is equal to ``right`` modulo ``self.uniformizer()^absprec``.
+
+        If ``absprec`` is ``None``, checks whether this element is equal to ``right``
+        modulo the lower of their two precisions.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: a = W(47); b = W(47 + 25)
+            sage: a.is_equal_to(b)
+            False
+            sage: a.is_equal_to(b, 7)
+            True
+        """
+        # Should be sped up later
+        return (self - right).is_zero(absprec)
+
+#    def lift(self):
+#        """
+#        Return an element of a number field defined by the same polynomial as
+#        ``self``'s parent that is congruent to self modulo an appropriate ideal.
+
+#        Not currently implemented.
+#        """
+#        raise NotImplementedError
+
+    cpdef pAdicZZpXCRElement lift_to_precision(self, absprec=None):
+        """
+        Return a ``pAdicZZpXCRElement`` congruent to this element but with
+        absolute precision at least ``absprec``.
+
+        INPUT:
+
+        - ``absprec`` -- (default: ``None``) the absolute precision of
+          the result.  If ``None``, lifts to the maximum precision
+          allowed.
+
+        .. NOTE::
+
+            If setting ``absprec`` that high would violate the
+            precision cap, raises a precision error.  If ``self`` is an
+            inexact zero and ``absprec`` is greater than the maximum
+            allowed valuation, raises an error.
+
+            Note that the new digits will not necessarily be zero.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: a = W(345, 17); a
+            4*w^5 + 3*w^7 + w^9 + 3*w^10 + 2*w^11 + 4*w^12 + w^13 + 2*w^14 + 2*w^15 + O(w^17)
+            sage: b = a.lift_to_precision(19); b
+            4*w^5 + 3*w^7 + w^9 + 3*w^10 + 2*w^11 + 4*w^12 + w^13 + 2*w^14 + 2*w^15
+             + w^17 + 2*w^18 + O(w^19)
+            sage: c = a.lift_to_precision(24); c
+            4*w^5 + 3*w^7 + w^9 + 3*w^10 + 2*w^11 + 4*w^12 + w^13 + 2*w^14 + 2*w^15
+             + w^17 + 2*w^18 + 4*w^19 + 4*w^20 + 2*w^21 + 4*w^23 + O(w^24)
+            sage: a._ntl_rep()
+            [19 35 118 60 121]
+            sage: b._ntl_rep()
+            [19 35 118 60 121]
+            sage: c._ntl_rep()
+            [19 35 118 60 121]
+            sage: a.lift_to_precision().precision_relative() == W.precision_cap()
+            True
+        """
+        cdef pAdicZZpXCRElement ans
+        cdef long aprec, rprec
+        self._normalize()
+        if self._is_exact_zero():
+            return self
+        if absprec is not None and not isinstance(absprec, Integer):
+            absprec = Integer(absprec)
+        if absprec is None:
+            if self.relprec == 0:
+                # return an exact zero
+                ans = self._new_c(0)
+                ans._set_exact_zero()
+                return ans
+            aprec = self.prime_pow.ram_prec_cap + self.ordp
+        elif mpz_fits_slong_p((<Integer>absprec).value) == 0:
+            if mpz_sgn((<Integer>absprec).value) < 0 or self.relprec == self.prime_pow.ram_prec_cap:
+                return self
+            else:
+                if self.relprec == 0:
+                    raise ValueError("absprec larger than maximum allowable valuation")
+                else:
+                    raise PrecisionError("precision higher than allowed by the precision cap")
+        else:
+            aprec = mpz_get_si((<Integer>absprec).value)
+        if aprec <= self.ordp + self.relprec:
+            return self
+        if self.relprec == 0:
+            if self.ordp >= aprec:
+                return self
+            elif aprec >= maxordp:
+                raise ValueError("absprec larger than maximum allowable valuation")
+            else:
+                ans = self._new_c(0)
+                ans._set_inexact_zero(aprec)
+                return ans
+        # Now we're done handling all the special cases.
+        rprec = aprec - self.ordp
+        if rprec > self.prime_pow.ram_prec_cap:
+            raise PrecisionError("precision higher than allowed by the precision cap")
+        ans = self._new_c(rprec)
+        ans.ordp = self.ordp
+        ZZ_pX_conv_modulus(ans.unit, self.unit, self.prime_pow.get_context_capdiv(rprec).x)
+        return ans
+
+    def expansion(self, n=None, lift_mode='simple'):
+        """
+        Return a list giving a series representation of ``self``.
+
+        - If ``lift_mode == 'simple'`` or ``'smallest'``, the returned
+          list will consist of integers (in the Eisenstein case) or a
+          list of lists of integers (in the unramified case).  ``self``
+          can be reconstructed as a sum of elements of the list times
+          powers of the uniformiser (in the Eisenstein case), or as a
+          sum of powers of the `p` times polynomials in the generator
+          (in the unramified case).
+
+          + If ``lift_mode == 'simple'``, all integers will be in the interval
+            `[0,p-1]`.
+
+          + If ``lift_mode == 'smallest'`` they will be in the
+            interval `[(1-p)/2, p/2]`.
+
+        - If ``lift_mode == 'teichmuller'``, returns a list of
+          ``pAdicZZpXCRElements``, all of which are Teichmuller
+          representatives and such that ``self`` is the sum of that list
+          times powers of the uniformizer.
+
+        Note that zeros are truncated from the returned list if
+        ``self.parent()`` is a field, so you must use the
+        ``valuation`` function to fully reconstruct ``self``.
+
+        INPUT:
+
+        - ``n`` -- integer (default: ``None``); if given, returns the
+          corresponding entry in the expansion
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: y = W(775, 19); y
+            w^10 + 4*w^12 + 2*w^14 + w^15 + 2*w^16 + 4*w^17 + w^18 + O(w^19)
+            sage: (y>>9).expansion()
+            [0, 1, 0, 4, 0, 2, 1, 2, 4, 1]
+            sage: (y>>9).expansion(lift_mode='smallest')
+            [0, 1, 0, -1, 0, 2, 1, 2, 0, 1]
+            sage: w^10 - w^12 + 2*w^14 + w^15 + 2*w^16 + w^18 + O(w^19)
+            w^10 + 4*w^12 + 2*w^14 + w^15 + 2*w^16 + 4*w^17 + w^18 + O(w^19)
+            sage: g = x^3 + 3*x + 3
+            sage: A.<a> = R.ext(g)
+            sage: y = 75 + 45*a + 1200*a^2; y
+            4*a*5 + (3*a^2 + a + 3)*5^2 + 4*a^2*5^3 + a^2*5^4 + O(5^6)
+            sage: E = y.expansion(); E
+            5-adic expansion of 4*a*5 + (3*a^2 + a + 3)*5^2 + 4*a^2*5^3 + a^2*5^4 + O(5^6)
+            sage: list(E)
+            [[], [0, 4], [3, 1, 3], [0, 0, 4], [0, 0, 1], []]
+            sage: list(y.expansion(lift_mode='smallest'))
+            [[], [0, -1], [-2, 2, -2], [1], [0, 0, 2], []]
+            sage: 5*((-2*5 + 25) + (-1 + 2*5)*a + (-2*5 + 2*125)*a^2)
+            4*a*5 + (3*a^2 + a + 3)*5^2 + 4*a^2*5^3 + a^2*5^4 + O(5^6)
+            sage: list(W(0).expansion())
+            []
+            sage: list(W(0,4).expansion())
+            []
+            sage: list(A(0,4).expansion())
+            []
+
+        TESTS:
+
+        We check that :issue:`24949` is fixed::
+
+            sage: R = Zp(2)
+            sage: S.<x> = R[]
+            sage: A.<a> = R.extension(x^10 + 2)
+            sage: u = a^4 + a^5
+            sage: v = a^2 + a^3
+            sage: w = u - v^2
+            sage: w.expansion(4)
+            0
+        """
+        self._normalize()
+        if lift_mode == 'teichmuller':
+            zero = self.parent()(0)
+        elif self.prime_pow.e == 1:
+            zero = []
+        else:
+            zero = Integer(0)
+        if isinstance(n, slice):
+            return self.slice(n.start, n.stop, n.step)
+        elif n is not None:
+            if self._is_exact_zero() or n < self.ordp:
+                return zero
+            elif n >= self.ordp + self.relprec:
+                raise PrecisionError
+        if self.relprec == 0: # cannot have n = None
+            return []
+        if lift_mode == 'simple':
+            ulist = self.ext_p_list(pos=True)
+        elif lift_mode == 'smallest':
+            ulist = self.ext_p_list(pos=False)
+        elif lift_mode == 'teichmuller':
+            if n is None:
+                ulist = self.teichmuller_expansion()
+            else:
+                return self.teichmuller_expansion(n)
+        else:
+            raise ValueError("lift mode must be one of 'simple', 'smallest' or 'teichmuller'")
+        if n is not None:
+            try:
+                return ulist[n - self.ordp]
+            except IndexError:
+                return zero
+        if self.prime_pow.in_field == 0 and self.ordp > 0:
+            ulist = [zero] * self.ordp + ulist
+        return ulist
+
+    def matrix_mod_pn(self):
+        r"""
+        Return the matrix of right multiplication by the element on
+        the power basis `1, x, x^2, \ldots, x^{d-1}` for this
+        extension field.  Thus the *rows* of this matrix give the
+        images of each of the `x^i`.  The entries of the matrices are
+        :class:`IntegerMod` elements, defined modulo `p^{N / e}` where `N` is
+        the absolute precision of this element (unless this element is
+        zero to arbitrary precision; in that case the entries are
+        integer zeros.)
+
+        Raises an error if this element has negative valuation.
+
+        EXAMPLES::
+
+            sage: R = ZpCR(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: a = (3+w)^7
+            sage: a.matrix_mod_pn()
+            [2757  333 1068  725 2510]
+            [  50 1507  483  318  725]
+            [ 500   50 3007 2358  318]
+            [1590 1375 1695 1032 2358]
+            [2415  590 2370 2970 1032]
+
+        TESTS:
+
+        Check that :issue:`13617` has been fixed::
+
+            sage: W.zero().matrix_mod_pn()
+            [0 0 0 0 0]
+            [0 0 0 0 0]
+            [0 0 0 0 0]
+            [0 0 0 0 0]
+            [0 0 0 0 0]
+        """
+        if self.valuation_c() < 0:
+            raise ValueError("self must be integral")
+        n = self.prime_pow.deg
+        from sage.matrix.constructor import matrix
+        if self._is_exact_zero():
+            from sage.rings.integer_ring import IntegerRing
+            return matrix(IntegerRing(), n, n)
+        R = IntegerModRing(self.prime_pow.pow_Integer(self.prime_pow.capdiv(self.ordp + self.relprec)))
+        L = []
+        cdef ntl_ZZ_pX cur = <ntl_ZZ_pX>self._ntl_rep_abs()[0]
+        cur.c.restore_c()
+        cdef ZZ_pX_Modulus_c* m = self.prime_pow.get_modulus_capdiv(self.ordp + self.relprec)
+        cdef ZZ_pX_c x
+        ZZ_pX_SetX(x)
+        cdef Py_ssize_t i
+        zero = int(0)
+        for i in range(n):
+            curlist = cur.list()
+            L.extend(curlist + [zero] * (n - len(curlist)))
+            ZZ_pX_MulMod_pre(cur.x, cur.x, x, m[0])
+        return matrix(R, n, n,  L)
+
+#     def matrix(self, base=None):
+#         """
+#         If base is None, return the matrix of right multiplication by
+#         the element on the power basis `1, x, x^2, \ldots, x^{d-1}`
+#         for this extension field.  Thus the \emph{rows} of this matrix
+#         give the images of each of the `x^i`.
+
+#         If base is not None, then base must be either a field that
+#         embeds in the parent of self or a morphism to the parent of
+#         self, in which case this function returns the matrix of
+#         multiplication by self on the power basis, where we view the
+#         parent field as a field over base.
+
+#         INPUT:
+#             base -- field or morphism
+#         """
+#         raise NotImplementedError
+
+#     def multiplicative_order(self, prec=None):
+#         """
+#         Return the multiplicative order of ``self``, ie the smallest
+#         positive `n` so that there is an exact `p`-adic element congruent
+#         to ``self`` modulo ``self``'s precision that is an `n`-th root of unity.
+
+#         Note: unlike the case for Qp and Zp, it is possible to have
+#         non-teichmuller elements with finite orders.  This can happen
+#         only if (p-1) divides the ramification index (see the
+#         documentation on __pow__).
+
+#         INPUT:
+
+#         - self -- a `p`-adic element
+#         - ``prec`` -- integer
+
+#         OUTPUT: integer; the multiplicative order of ``self``
+#         """
+#         raise NotImplementedError
+
+    def teichmuller_expansion(self, n=None):
+        r"""
+        Return a list [`a_0`, `a_1`,..., `a_n`] such that:
+
+        - `a_i^q = a_i`
+        - ``self.unit_part()`` = `\sum_{i = 0}^n a_i \pi^i`, where `\pi` is a
+          uniformizer of ``self.parent()``
+        - if `a_i \ne 0`, the absolute precision of `a_i` is
+          ``self.precision_relative() - i``
+
+        INPUT:
+
+        - ``n`` -- integer (default: ``None``); if given, returns the
+          corresponding entry in the expansion
+
+        EXAMPLES::
+
+            sage: R.<a> = ZqCR(5^4,4)
+            sage: E = a.teichmuller_expansion(); E
+            5-adic expansion of a + O(5^4) (teichmuller)
+            sage: list(E)
+            [a + (2*a^3 + 2*a^2 + 3*a + 4)*5 + (4*a^3 + 3*a^2 + 3*a + 2)*5^2
+               + (4*a^2 + 2*a + 2)*5^3 + O(5^4),
+             (3*a^3 + 3*a^2 + 2*a + 1) + (a^3 + 4*a^2 + 1)*5 + (a^2 + 4*a + 4)*5^2 + O(5^3),
+             (4*a^3 + 2*a^2 + a + 1) + (2*a^3 + 2*a^2 + 2*a + 4)*5 + O(5^2),
+             (a^3 + a^2 + a + 4) + O(5)]
+            sage: sum([c * 5^i for i, c in enumerate(E)])
+            a + O(5^4)
+            sage: all(c^625 == c for c in E)
+            True
+
+            sage: S.<x> = ZZ[]
+            sage: f = x^3 - 98*x + 7
+            sage: W.<w> = ZpCR(7,3).ext(f)
+            sage: b = (1+w)^5; L = b.teichmuller_expansion(); L
+            [1 + O(w^9), 5 + 5*w^3 + w^6 + 4*w^7 + O(w^8), 3 + 3*w^3 + O(w^7),
+             3 + 3*w^3 + O(w^6), O(w^5), 4 + 5*w^3 + O(w^4), 3 + O(w^3),
+             6 + O(w^2), 6 + O(w)]
+            sage: sum([w^i*L[i] for i in range(9)]) == b
+            True
+            sage: all(L[i]^(7^3) == L[i] for i in range(9))
+            True
+
+            sage: L = W(3).teichmuller_expansion(); L
+            [3 + 3*w^3 + w^7 + O(w^9), O(w^8), O(w^7), 4 + 5*w^3 + O(w^6),
+             O(w^5), O(w^4), 3 + O(w^3), 6 + O(w^2)]
+            sage: sum([w^i*L[i] for i in range(len(L))])
+            3 + O(w^9)
+        """
+        cdef long rp = self.relprec
+        cdef pAdicZZpXCRElement v
+        if n is None:
+            L = []
+            if rp == 0:
+                return L
+        elif self._is_exact_zero() or n < self.ordp:
+            return self.parent()(0)
+        elif n >= self.ordp + rp:
+            raise PrecisionError
+        else:
+            v = self._new_c(rp)
+        cdef pAdicZZpXCRElement u = self.unit_part()
+        cdef long goal
+        if n is not None: goal = rp - n + self.ordp
+        while u.relprec > 0:
+            v = self._new_c(rp)
+            self.prime_pow.teichmuller_set_c(&v.unit, &u.unit, rp)
+            v.ordp = 0
+            if n is None:
+                L.append(v)
+            elif rp == goal:
+                return v
+            if rp == 1: break
+            ZZ_pX_sub(u.unit, u.unit, v.unit)
+            u.relprec = -u.relprec
+            u._normalize()
+            if u.relprec == 0: break
+            rp -= 1
+            u.ordp -= 1
+            while u.ordp > 0:
+                if n is None:
+                    v = self._new_c(0)
+                    v._set_inexact_zero(rp)
+                    L.append(v)
+                elif rp == goal:
+                    v = self._new_c(0)
+                    v._set_inexact_zero(rp)
+                    return v
+                rp -= 1
+                u.ordp -= 1
+        if n is None:
+            return L
+        else:
+            v = self._new_c(0)
+            v._set_inexact_zero(rp)
+            return v
+
+    def _teichmuller_set_unsafe(self):
+        """
+        Set this element to the Teichmuller representative with the
+        same residue.
+
+        .. WARNING::
+
+            This function modifies the element, which is not safe.
+            Elements are supposed to be immutable.
+
+        EXAMPLES::
+
+            sage: R = Zp(7,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 77*x^3 - 98*x^2 - 7
+            sage: W.<w> = R.ext(f)
+            sage: y = W.teichmuller(3, 15); y  # indirect doctest
+            3 + 4*w^5 + 2*w^8 + 6*w^10 + w^11 + 6*w^12 + 5*w^13 + 4*w^14 + O(w^15)
+
+            sage: y^7 == y
+            True
+            sage: g = x^3 + 3*x^2 + 4
+            sage: A.<a> = R.ext(g)
+            sage: b = A.teichmuller(1 + 2*a - a^2); b
+            (6*a^2 + 2*a + 1) + (5*a + 3)*7 + (5*a + 5)*7^2 + (4*a^2 + 4*a + 2)*7^3 + (2*a + 1)*7^4 + O(7^5)
+            sage: b^343 == b
+            True
+
+        TESTS:
+
+        We check that :issue:`8239` is resolved::
+
+            sage: K.<a> = Qq(25)
+            sage: K.teichmuller(K(2/5))
+            Traceback (most recent call last):
+            ...
+            ValueError: cannot set negative valuation element to Teichmuller representative
+        """
+        self._normalize()
+        if self.ordp > 0:
+            self._set_exact_zero()
+        elif self.ordp < 0:
+            raise ValueError("cannot set negative valuation element to Teichmuller representative")
+        elif self.relprec == 0:
+            raise ValueError("not enough precision known")
+        else:
+            self.prime_pow.teichmuller_set_c(&self.unit, &self.unit, self.relprec)
+
+#    def padded_list(self, n, lift_mode='simple'):
+#        """
+#        Return a list of coefficients of pi starting with `pi^0` up to
+#        `pi^n` exclusive (padded with zeros if needed)
+
+#        """
+#        raise NotImplementedError
+
+    def precision_absolute(self):
+        """
+        Return the absolute precision of this element, i.e., the power of the
+        uniformizer modulo which this element is defined.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: a = W(75, 19); a
+            3*w^10 + 2*w^12 + w^14 + w^16 + w^17 + 3*w^18 + O(w^19)
+            sage: a.valuation()
+            10
+            sage: a.precision_absolute()
+            19
+            sage: a.precision_relative()
+            9
+            sage: a.unit_part()
+            3 + 2*w^2 + w^4 + w^6 + w^7 + 3*w^8 + O(w^9)
+            sage: (a.unit_part() - 3).precision_absolute()
+            9
+        """
+        cdef Integer ans
+        if self.ordp == maxordp:
+            return infinity
+        else:
+            ans = PY_NEW(Integer)
+            if self.relprec > 0:
+                mpz_set_si(ans.value, self.relprec + self.ordp)
+            else:
+                mpz_set_si(ans.value, -self.relprec + self.ordp)
+            return ans
+
+    def precision_relative(self):
+        """
+        Return the relative precision of this element, i.e., the power of the
+        uniformizer modulo which the unit part of ``self`` is defined.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: a = W(75, 19); a
+            3*w^10 + 2*w^12 + w^14 + w^16 + w^17 + 3*w^18 + O(w^19)
+            sage: a.valuation()
+            10
+            sage: a.precision_absolute()
+            19
+            sage: a.precision_relative()
+            9
+            sage: a.unit_part()
+            3 + 2*w^2 + w^4 + w^6 + w^7 + 3*w^8 + O(w^9)
+        """
+        self._normalize()
+        cdef Integer ans = PY_NEW(Integer)
+        mpz_set_ui(ans.value, self.relprec)
+        return ans
+
+#    def residue(self, n=1):
+#        """
+#        Reduces this element modulo pi^n.
+#        """
+#        raise NotImplementedError
+
+    cdef long valuation_c(self) noexcept:
+        """
+        Return the valuation of this element.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: a = W(75, 19); a
+            3*w^10 + 2*w^12 + w^14 + w^16 + w^17 + 3*w^18 + O(w^19)
+            sage: a.valuation()  # indirect doctest
+            10
+            sage: a.precision_absolute()
+            19
+            sage: a.precision_relative()
+            9
+            sage: a.unit_part()
+            3 + 2*w^2 + w^4 + w^6 + w^7 + 3*w^8 + O(w^9)
+        """
+        self._normalize()
+        return self.ordp
+
+    cpdef pAdicZZpXCRElement unit_part(self):
+        """
+        Return the unit part of this element, ie ``self / uniformizer^(self.valuation())``.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: a = W(75, 19); a
+            3*w^10 + 2*w^12 + w^14 + w^16 + w^17 + 3*w^18 + O(w^19)
+            sage: a.valuation()
+            10
+            sage: a.precision_absolute()
+            19
+            sage: a.precision_relative()
+            9
+            sage: a.unit_part()
+            3 + 2*w^2 + w^4 + w^6 + w^7 + 3*w^8 + O(w^9)
+
+        TESTS:
+
+        We check that :issue:`13616` is resolved::
+
+            sage: z = (1+w)^5
+            sage: y = z - 1
+            sage: t = y - y
+            sage: t.unit_part()
+            O(w^0)
+        """
+        self._normalize()
+        cdef pAdicZZpXCRElement ans = self._new_c(self.relprec)
+        ans.ordp = 0
+        if self.relprec != 0:
+            ans.unit = self.unit
+        return ans
+
+    cdef ext_p_list(self, bint pos):
+        """
+        Return a list of integers (in the Eisenstein case) or a list
+        of lists of integers (in the unramified case).  ``self`` can be
+        reconstructed as a sum of elements of the list times powers of
+        the uniformiser (in the Eisenstein case), or as a sum of
+        powers of `p` times polynomials in the generator (in the
+        unramified case).
+
+        If ``pos`` is ``True``, all integers will be in the interval `[0,p-1]`,
+        otherwise they will be in the interval `[(1-p)/2, p/2]`.
+
+        Note that zeros are truncated from the returned list, so you
+        must use the method :meth:`valuation` to completely recover ``self``.
+
+        EXAMPLES::
+
+            sage: R = Zp(5,5)
+            sage: S.<x> = R[]
+            sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+            sage: W.<w> = R.ext(f)
+            sage: y = W(775, 19); y
+            w^10 + 4*w^12 + 2*w^14 + w^15 + 2*w^16 + 4*w^17 + w^18 + O(w^19)
+            sage: y._ext_p_list(True)
+            [1, 0, 4, 0, 2, 1, 2, 4, 1]
+            sage: y._ext_p_list(False)
+            [1, 0, -1, 0, 2, 1, 2, 0, 1]
+            sage: w^10 - w^12 + 2*w^14 + w^15 + 2*w^16 + w^18 + O(w^19)
+            w^10 + 4*w^12 + 2*w^14 + w^15 + 2*w^16 + 4*w^17 + w^18 + O(w^19)
+            sage: g = x^3 + 3*x + 3
+            sage: A.<a> = R.ext(g)
+            sage: y = 75 + 45*a + 1200*a^2; y
+            4*a*5 + (3*a^2 + a + 3)*5^2 + 4*a^2*5^3 + a^2*5^4 + O(5^6)
+            sage: y._ext_p_list(True)
+            [[0, 4], [3, 1, 3], [0, 0, 4], [0, 0, 1]]
+            sage: y._ext_p_list(False)
+            [[0, -1], [-2, 2, -2], [1], [0, 0, 2]]
+            sage: 5*((-2*5 + 25) + (-1 + 2*5)*a + (-2*5 + 2*125)*a^2)
+            4*a*5 + (3*a^2 + a + 3)*5^2 + 4*a^2*5^3 + a^2*5^4 + O(5^6)
+        """
+        self._normalize()
+        return self.ext_p_list_precs(pos, self.relprec)
+
+
+def make_ZZpXCRElement(parent, unit, ordp, relprec, version):
+    """
+    Unpickling.
+
+    EXAMPLES::
+
+        sage: R = Zp(5,5)
+        sage: S.<x> = R[]
+        sage: f = x^5 + 75*x^3 - 15*x^2 + 125*x - 5
+        sage: W.<w> = R.ext(f)
+        sage: y = W(775, 19); y
+        w^10 + 4*w^12 + 2*w^14 + w^15 + 2*w^16 + 4*w^17 + w^18 + O(w^19)
+        sage: loads(dumps(y))  # indirect doctest
+        w^10 + 4*w^12 + 2*w^14 + w^15 + 2*w^16 + 4*w^17 + w^18 + O(w^19)
+
+        sage: from sage.rings.padics.padic_ZZ_pX_CR_element import make_ZZpXCRElement
+        sage: make_ZZpXCRElement(W, y._ntl_rep(), 3, 9, 0)
+        w^3 + 4*w^5 + 2*w^7 + w^8 + 2*w^9 + 4*w^10 + w^11 + O(w^12)
+    """
+    cdef pAdicZZpXCRElement ans
+    cdef ZZ_pX_c poly
+    if version == 0:
+        ans = pAdicZZpXCRElement(parent, [], empty = True)
+        if relprec == 0:
+            ans._set_inexact_zero(mpz_get_si((<Integer>ordp).value))
+        else:
+            ans.prime_pow.restore_context_capdiv(mpz_get_si((<Integer>relprec).value))
+            poly = (<ntl_ZZ_pX>unit).x
+            ans._set(&poly, mpz_get_si((<Integer>ordp).value), mpz_get_si((<Integer>relprec).value))
+        return ans
+    else:
+        raise ValueError("unknown unpickling version")