pygeodesy 24.3.24__py2.py3-none-any.whl

pygeodesy/fsums.py

@@ -0,0 +1,1898 @@
+
+# -*- coding: utf-8 -*-
+
+u'''Class L{Fsum} for precision floating point summation and I{running}
+summation based on, respectively similar to Python's C{math.fsum}.
+
+Generally, an L{Fsum} instance is considered a C{float} plus a small or zero
+C{residual} value, see property L{Fsum.residual}.  However, there are several
+C{integer} L{Fsum} cases, for example the result of C{ceil}, C{floor},
+C{Fsum.__floordiv__} and methods L{Fsum.fint} and L{Fsum.fint2}.
+
+Also, L{Fsum} methods L{Fsum.pow}, L{Fsum.__ipow__}, L{Fsum.__pow__} and
+L{Fsum.__rpow__} return a (very long) C{int} if invoked with optional argument
+C{mod} set to C{None}.  The C{residual} of an C{integer} L{Fsum} may be between
+C{-1.0} and C{+1.0}, including C{INT0} if considered to be I{exact}.
+
+Set env variable C{PYGEODESY_FSUM_PARTIALS} to an empty string (or anything
+other than C{"fsum"}) for backward compatible summation of L{Fsum} partials.
+
+Set env variable C{PYGEODESY_FSUM_RESIDUAL} to a C{float} string greater
+than C{"0.0"} as the threshold to throw a L{ResidualError} in division or
+exponention of an L{Fsum} instance with a I{relative} C{residual} exceeding
+the threshold, see methods L{Fsum.RESIDUAL}, L{Fsum.pow}, L{Fsum.__ipow__}
+and L{Fsum.__itruediv__}.
+'''
+# make sure int/int division yields float quotient, see .basics
+from __future__ import division as _; del _  # PYCHOK semicolon
+
+from pygeodesy.basics import iscomplex, isint, isscalar, itemsorted, \
+                             signOf, _signOf, _xisscalar
+from pygeodesy.constants import INT0, _isfinite, isinf, isnan, _pos_self, \
+                               _0_0, _1_0, _N_1_0,  Float, Int
+from pygeodesy.errors import _OverflowError, _TypeError, _ValueError, _xError2, \
+                             _xkwds_get, _ZeroDivisionError
+from pygeodesy.interns import NN, _arg_, _COMMASPACE_, _DASH_, _DOT_, _EQUAL_, \
+                             _exceeds_, _from_, _iadd_op_, _LANGLE_, _negative_, \
+                             _NOTEQUAL_, _not_finite_, _not_scalar_, _PERCENT_, \
+                             _PLUS_, _R_, _RANGLE_, _SLASH_, _SPACE_, _STAR_, _UNDER_
+from pygeodesy.lazily import _ALL_LAZY, _getenv, _sys_version_info2
+from pygeodesy.named import _Named, _NamedTuple, _NotImplemented,  Fmt, unstr
+from pygeodesy.props import _allPropertiesOf_n, deprecated_property_RO, \
+                             Property_RO, property_RO
+# from pygeodesy.streprs import Fmt, unstr  # from .named
+# from pygeodesy.units import Float, Int  # from .constants
+
+from math import ceil as _ceil, fabs, floor as _floor  # PYCHOK used! .ltp
+
+__all__ = _ALL_LAZY.fsums
+__version__ = '24.03.18'
+
+_add_op_       = _PLUS_  # in .auxilats.auxAngle
+_eq_op_        = _EQUAL_ * 2  # _DEQUAL_
+_COMMASPACE_R_ = _COMMASPACE_ + _R_
+_exceeds_R_    = _SPACE_ + _exceeds_(_R_)
+_floordiv_op_  = _SLASH_ * 2  # _DSLASH_
+_fset_op_      = _EQUAL_
+_ge_op_        = _RANGLE_ + _EQUAL_
+_gt_op_        = _RANGLE_
+_integer_      = 'integer'
+_le_op_        = _LANGLE_ + _EQUAL_
+_lt_op_        = _LANGLE_
+_mod_op_       = _PERCENT_
+_mul_op_       = _STAR_
+_ne_op_        = _NOTEQUAL_
+_non_zero_     = 'non-zero'
+_pow_op_       = _STAR_ * 2  # _DSTAR_, in .fmath
+_sub_op_       = _DASH_      # in .auxilats.auxAngle, .fsums
+_truediv_op_   = _SLASH_
+_divmod_op_    = _floordiv_op_ + _mod_op_
+_isub_op_      = _sub_op_ + _fset_op_  # in .auxilats.auxAngle, .fsums
+
+
+def _2float(index=None, **name_value):  # in .fmath, .fstats
+    '''(INTERNAL) Raise C{TypeError} or C{ValueError} if not scalar or infinite.
+    '''
+    n, v = name_value.popitem()  # _xkwds_item2(name_value)
+    try:
+        v = float(v)
+        if _isfinite(v):
+            return v
+        E, t = _ValueError, _not_finite_
+    except Exception as e:
+        E, t = _xError2(e)
+    raise E(Fmt.INDEX(n, index), v, txt=t)
+
+
+def _2floats(xs, origin=0, sub=False):
+    '''(INTERNAL) Yield each B{C{xs}} as a C{float}.
+    '''
+    try:
+        i, x  =  origin, None
+        _fin  = _isfinite
+        _Fsum =  Fsum
+        for x in xs:
+            if isinstance(x, _Fsum):
+                for p in x._ps:
+                    yield (-p) if sub else p
+            else:
+                f = float(x)
+                if not _fin(f):
+                    raise ValueError(_not_finite_)
+                if f:
+                    yield (-f) if sub else f
+            i += 1
+    except Exception as e:
+        E, t = _xError2(e)
+        n = Fmt.SQUARE(xs=i)
+        raise E(n, x, txt=t)
+
+
+def _Powers(power, xs, origin=1):  # in .fmath
+    '''(INTERNAL) Yield each C{xs} as C{float(x**power)}.
+    '''
+    _xisscalar(power=power)
+    try:
+        i, x  =  origin, None
+        _fin  = _isfinite
+        _Fsum =  Fsum
+        _pow  =  pow  # XXX math.pow
+        for x in xs:
+            if isinstance(x, _Fsum):
+                P = x.pow(power)
+                for p in P._ps:
+                    yield p
+            else:
+                p = _pow(float(x), power)
+                if not _fin(p):
+                    raise ValueError(_not_finite_)
+                yield p
+            i += 1
+    except Exception as e:
+        E, t = _xError2(e)
+        n = Fmt.SQUARE(xs=i)
+        raise E(n, x, txt=t)
+
+
+def _1primed(xs):
+    '''(INTERNAL) 1-Prime the summation of C{xs}
+       arguments I{known} to be C{finite float}.
+    '''
+    yield _1_0
+    for x in xs:
+        if x:
+            yield x
+    yield _N_1_0
+
+
+def _psum(ps):  # PYCHOK used!
+    '''(INTERNAL) Partials summation updating C{ps}, I{overridden below}.
+    '''
+    i   =  len(ps) - 1  # len(ps) > 2
+    s   =  ps[i]
+    _2s = _2sum
+    while i > 0:
+        i -= 1
+        s, r = _2s(s, ps[i])
+        if r:  # sum(ps) became inexact
+            ps[i:] = [s, r] if s else [r]
+            if i > 0:
+                p = ps[i-1]  # round half-even
+                if (p > 0 and r > 0) or \
+                   (p < 0 and r < 0):  # signs match
+                    r *= 2
+                    t  = s + r
+                    if r == (t - s):
+                        s = t
+            break
+        ps[i:] = [s]
+    return s
+
+
+def _2scalar(other, _raiser=None):
+    '''(INTERNAL) Return B{C{other}} as C{int}, C{float} or C{as-is}.
+    '''
+    if isinstance(other, Fsum):
+        s, r = other._fint2
+        if r:
+            s, r = other._fprs2
+            if r:  # PYCHOK no cover
+                if _raiser and _raiser(r, s):
+                    raise ValueError(_stresidual(_non_zero_, r))
+                s = other  # L{Fsum} as-is
+    else:
+        s = other  # C{type} as-is
+        if isint(s, both=True):
+            s = int(s)
+    return s
+
+
+def _strcomplex(s, *args):
+    '''(INTERNAL) C{Complex} 2- or 3-arg C{pow} error C{str}.
+    '''
+    c =  iscomplex.__name__[2:]
+    n = _DASH_(len(args), _arg_)
+    t = _SPACE_(c, s, _from_, n, pow.__name__)
+    return unstr(t, *args)
+
+
+def _stresidual(prefix, residual, **name_values):
+    '''(INTERNAL) Residual error C{str}.
+    '''
+    p = _SPACE_(prefix, Fsum.residual.name)
+    t =  Fmt.PARENSPACED(p, Fmt(residual))
+    for n, v in itemsorted(name_values):
+        n =  n.replace(_UNDER_, _SPACE_)
+        p =  Fmt.PARENSPACED(n, Fmt(v))
+        t = _COMMASPACE_(t, p)
+    return t
+
+
+def _2sum(a, b):  # by .testFmath
+    '''(INTERNAL) Return C{a + b} as 2-tuple (sum, residual).
+    '''
+    s = a + b
+    if not _isfinite(s):
+        u = unstr(_2sum.__name__, a, b)
+        t = Fmt.PARENSPACED(_not_finite_, s)
+        raise _OverflowError(u, txt=t)
+    if fabs(a) < fabs(b):
+        a, b = b, a
+    return s, (b - (s - a))
+
+
+class Fsum(_Named):  # sync __methods__ with .vector3dBase.Vector3dBase
+    '''Precision floating point I{running} summation.
+
+       Unlike Python's C{math.fsum}, this class accumulates values and provides intermediate,
+       I{running} precision floating point summation.  Accumulation may continue after
+       intermediate, I{running} summuation.
+
+       @note: Accumulated values may be L{Fsum} or C{scalar} instances with C{scalar} meaning
+              type C{float}, C{int} or any C{type} convertible to a single C{float}, having
+              method C{__float__}.
+
+       @note: Handling of exceptions and C{inf}, C{INF}, C{nan} and C{NAN} differs from
+              Python's C{math.fsum}.
+
+       @see: U{Hettinger<https://GitHub.com/ActiveState/code/blob/master/recipes/Python/
+             393090_Binary_floating_point_summatiaccurate_full/recipe-393090.py>}, U{Kahan
+             <https://WikiPedia.org/wiki/Kahan_summation_algorithm>}, U{Klein
+             <https://Link.Springer.com/article/10.1007/s00607-005-0139-x>}, Python 2.6+
+             file I{Modules/mathmodule.c} and the issue log U{Full precision summation
+             <https://Bugs.Python.org/issue2819>}.
+    '''
+    _math_fsum = None
+    _n         = 0
+#   _ps        = []  # partial sums
+#   _px        = 0
+    _ratio     = None
+    _RESIDUAL  = max(float(_getenv('PYGEODESY_FSUM_RESIDUAL', _0_0)), _0_0)
+
+    def __init__(self, *xs, **name_RESIDUAL):
+        '''New L{Fsum} for precision floating point I{running} summation.
+
+           @arg xs: No, one or more initial values (each C{scalar} or an
+                    L{Fsum} instance).
+           @kwarg name_RESIDUAL: Optional C{B{name}=NN} for this L{Fsum}
+                                 (C{str}) and C{B{RESIDUAL}=None} for the
+                                 L{ResidualError} threshold.
+
+           @see: Methods L{Fsum.fadd} and L{Fsum.RESIDUAL}.
+        '''
+        if name_RESIDUAL:
+            n = _xkwds_get(name_RESIDUAL, name=NN)
+            if n:  # set name ...
+                self.name = n
+            r = _xkwds_get(name_RESIDUAL, RESIDUAL=None)
+            if r is not None:
+                self.RESIDUAL(r)  # ... for ResidualError
+#       self._n  = 0
+        self._ps = []  # [_0_0], see L{Fsum._fprs}
+        if len(xs) > 1:
+            self._facc(_2floats(xs, origin=1), up=False)  # PYCHOK yield
+        elif xs:  # len(xs) == 1
+            self._ps = [_2float(x=xs[0])]
+            self._n  = 1
+
+    def __abs__(self):
+        '''Return this instance' absolute value as an L{Fsum}.
+        '''
+        s = _fsum(self._ps_1())  # == self._cmp_0(0, ...)
+        return self._copy_n(self.__abs__) if s < 0 else \
+               self._copy_2(self.__abs__)
+
+    def __add__(self, other):
+        '''Return the C{Fsum(B{self}, B{other})}.
+
+           @arg other: An L{Fsum} or C{scalar}.
+
+           @return: The sum (L{Fsum}).
+
+           @see: Method L{Fsum.__iadd__}.
+        '''
+        f = self._copy_2(self.__add__)
+        return f._fadd(other, _add_op_)
+
+    def __bool__(self):  # PYCHOK not special in Python 2-
+        '''Return C{True} if this instance is I{exactly} non-zero.
+        '''
+        s, r = self._fprs2
+        return bool(s or r) and s != -r  # == self != 0
+
+    def __ceil__(self):  # PYCHOK not special in Python 2-
+        '''Return this instance' C{math.ceil} as C{int} or C{float}.
+
+           @return: An C{int} in Python 3+, but C{float} in Python 2-.
+
+           @see: Methods L{Fsum.__floor__} and property L{Fsum.ceil}.
+        '''
+        return self.ceil
+
+    def __cmp__(self, other):  # Python 2-
+        '''Compare this with an other instance or C{scalar}.
+
+           @return: -1, 0 or +1 (C{int}).
+
+           @raise TypeError: Incompatible B{C{other}} C{type}.
+        '''
+        s = self._cmp_0(other, self.cmp.__name__)
+        return _signOf(s, 0)
+
+    cmp = __cmp__
+
+    def __divmod__(self, other):
+        '''Return C{divmod(B{self}, B{other})} as 2-tuple C{(quotient,
+           remainder)}, an C{int} in Python 3+ or C{float} in Python 2-
+           and an L{Fsum}.
+
+           @arg other: An L{Fsum} or C{scalar} modulus.
+
+           @see: Method L{Fsum.__itruediv__}.
+        '''
+        f = self._copy_2(self.__divmod__)
+        return f._fdivmod2(other, _divmod_op_)
+
+    def __eq__(self, other):
+        '''Compare this with an other instance or C{scalar}.
+        '''
+        return self._cmp_0(other, _eq_op_) == 0
+
+    def __float__(self):
+        '''Return this instance' current precision running sum as C{float}.
+
+           @see: Methods L{Fsum.fsum} and L{Fsum.int_float}.
+        '''
+        return float(self._fprs)
+
+    def __floor__(self):  # PYCHOK not special in Python 2-
+        '''Return this instance' C{math.floor} as C{int} or C{float}.
+
+           @return: An C{int} in Python 3+, but C{float} in Python 2-.
+
+           @see: Methods L{Fsum.__ceil__} and property L{Fsum.floor}.
+        '''
+        return self.floor
+
+    def __floordiv__(self, other):
+        '''Return C{B{self} // B{other}} as an L{Fsum}.
+
+           @arg other: An L{Fsum} or C{scalar} divisor.
+
+           @return: The C{floor} quotient (L{Fsum}).
+
+           @see: Methods L{Fsum.__ifloordiv__}.
+        '''
+        f = self._copy_2(self.__floordiv__)
+        return f._floordiv(other, _floordiv_op_)
+
+    def __format__(self, *other):  # PYCHOK no cover
+        '''Not implemented.'''
+        return _NotImplemented(self, *other)
+
+    def __ge__(self, other):
+        '''Compare this with an other instance or C{scalar}.
+        '''
+        return self._cmp_0(other, _ge_op_) >= 0
+
+    def __gt__(self, other):
+        '''Compare this with an other instance or C{scalar}.
+        '''
+        return self._cmp_0(other, _gt_op_) > 0
+
+    def __hash__(self):  # PYCHOK no cover
+        '''Return this instance' C{hash}.
+        '''
+        return hash(self._ps)  # XXX id(self)?
+
+    def __iadd__(self, other):
+        '''Apply C{B{self} += B{other}} to this instance.
+
+           @arg other: An L{Fsum} or C{scalar} instance.
+
+           @return: This instance, updated (L{Fsum}).
+
+           @raise TypeError: Invalid B{C{other}}, not
+                             C{scalar} nor L{Fsum}.
+
+           @see: Methods L{Fsum.fadd} and L{Fsum.fadd_}.
+        '''
+        return self._fadd(other, _iadd_op_)
+
+    def __ifloordiv__(self, other):
+        '''Apply C{B{self} //= B{other}} to this instance.
+
+           @arg other: An L{Fsum} or C{scalar} divisor.
+
+           @return: This instance, updated (L{Fsum}).
+
+           @raise ResidualError: Non-zero residual in B{C{other}}.
+
+           @raise TypeError: Invalid B{C{other}} type.
+
+           @raise ValueError: Invalid or non-finite B{C{other}}.
+
+           @raise ZeroDivisionError: Zero B{C{other}}.
+
+           @see: Methods L{Fsum.__itruediv__}.
+        '''
+        return self._floordiv(other, _floordiv_op_ + _fset_op_)
+
+    def __imatmul__(self, other):  # PYCHOK no cover
+        '''Not implemented.'''
+        return _NotImplemented(self, other)
+
+    def __imod__(self, other):
+        '''Apply C{B{self} %= B{other}} to this instance.
+
+           @arg other: An L{Fsum} or C{scalar} modulus.
+
+           @return: This instance, updated (L{Fsum}).
+
+           @see: Method L{Fsum.__divmod__}.
+        '''
+        self._fdivmod2(other, _mod_op_ + _fset_op_)
+        return self
+
+    def __imul__(self, other):
+        '''Apply C{B{self} *= B{other}} to this instance.
+
+           @arg other: An L{Fsum} or C{scalar} factor.
+
+           @return: This instance, updated (L{Fsum}).
+
+           @raise OverflowError: Partial C{2sum} overflow.
+
+           @raise TypeError: Invalid B{C{other}} type.
+
+           @raise ValueError: Invalid or non-finite B{C{other}}.
+        '''
+        return self._fmul(other, _mul_op_ + _fset_op_)
+
+    def __int__(self):
+        '''Return this instance as an C{int}.
+
+           @see: Methods L{Fsum.int_float}, L{Fsum.__ceil__}
+                 and L{Fsum.__floor__} and properties
+                 L{Fsum.ceil} and L{Fsum.floor}.
+        '''
+        i, _ = self._fint2
+        return i
+
+    def __invert__(self):  # PYCHOK no cover
+        '''Not implemented.'''
+        # Luciano Ramalho, "Fluent Python", O'Reilly, 2nd Ed, 2022 p. 567
+        return _NotImplemented(self)
+
+    def __ipow__(self, other, *mod):  # PYCHOK 2 vs 3 args
+        '''Apply C{B{self} **= B{other}} to this instance.
+
+           @arg other: The exponent (L{Fsum} or C{scalar}).
+           @arg mod: Optional modulus (C{int} or C{None}) for the
+                     3-argument C{pow(B{self}, B{other}, B{mod})}
+                     version.
+
+           @return: This instance, updated (L{Fsum}).
+
+           @note: If B{C{mod}} is given, the result will be an C{integer}
+                  L{Fsum} in Python 3+ if this instance C{is_integer} or
+                  set to C{as_integer} if B{C{mod}} given as C{None}.
+
+           @raise OverflowError: Partial C{2sum} overflow.
+
+           @raise ResidualError: Non-zero residual in B{C{other}} and
+                                 env var C{PYGEODESY_FSUM_RESIDUAL}
+                                 set or this instance has a non-zero
+                                 residual and either B{C{mod}} is
+                                 given and non-C{None} or B{C{other}}
+                                 is a negative or fractional C{scalar}.
+
+           @raise TypeError: Invalid B{C{other}} type or 3-argument
+                             C{pow} invocation failed.
+
+           @raise ValueError: If B{C{other}} is a negative C{scalar}
+                              and this instance is C{0} or B{C{other}}
+                              is a fractional C{scalar} and this
+                              instance is negative or has a non-zero
+                              residual or B{C{mod}} is given and C{0}.
+
+           @see: CPython function U{float_pow<https://GitHub.com/
+                 python/cpython/blob/main/Objects/floatobject.c>}.
+        '''
+        return self._fpow(other, _pow_op_ + _fset_op_, *mod)
+
+    def __isub__(self, other):
+        '''Apply C{B{self} -= B{other}} to this instance.
+
+           @arg other: An L{Fsum} or C{scalar}.
+
+           @return: This instance, updated (L{Fsum}).
+
+           @raise TypeError: Invalid B{C{other}} type.
+
+           @see: Method L{Fsum.fadd}.
+        '''
+        return self._fsub(other, _isub_op_)
+
+    def __iter__(self):
+        '''Return an C{iter}ator over a C{partials} duplicate.
+        '''
+        return iter(self.partials)
+
+    def __itruediv__(self, other):
+        '''Apply C{B{self} /= B{other}} to this instance.
+
+           @arg other: An L{Fsum} or C{scalar} divisor.
+
+           @return: This instance, updated (L{Fsum}).
+
+           @raise OverflowError: Partial C{2sum} overflow.
+
+           @raise ResidualError: Non-zero residual in B{C{other}} and
+                                 env var C{PYGEODESY_FSUM_RESIDUAL} set.
+
+           @raise TypeError: Invalid B{C{other}} type.
+
+           @raise ValueError: Invalid or non-finite B{C{other}}.
+
+           @raise ZeroDivisionError: Zero B{C{other}}.
+
+           @see: Method L{Fsum.__ifloordiv__}.
+        '''
+        return self._ftruediv(other, _truediv_op_ + _fset_op_)
+
+    def __le__(self, other):
+        '''Compare this with an other instance or C{scalar}.
+        '''
+        return self._cmp_0(other, _le_op_) <= 0
+
+    def __len__(self):
+        '''Return the number of values accumulated (C{int}).
+        '''
+        return self._n
+
+    def __lt__(self, other):
+        '''Compare this with an other instance or C{scalar}.
+        '''
+        return self._cmp_0(other, _lt_op_) < 0
+
+    def __matmul__(self, other):  # PYCHOK no cover
+        '''Not implemented.'''
+        return _NotImplemented(self, other)
+
+    def __mod__(self, other):
+        '''Return C{B{self} % B{other}} as an L{Fsum}.
+
+           @see: Method L{Fsum.__imod__}.
+        '''
+        f = self._copy_2(self.__mod__)
+        return f._fdivmod2(other, _mod_op_)[1]
+
+    def __mul__(self, other):
+        '''Return C{B{self} * B{other}} as an L{Fsum}.
+
+           @see: Method L{Fsum.__imul__}.
+        '''
+        f = self._copy_2(self.__mul__)
+        return f._fmul(other, _mul_op_)
+
+    def __ne__(self, other):
+        '''Compare this with an other instance or C{scalar}.
+        '''
+        return self._cmp_0(other, _ne_op_) != 0
+
+    def __neg__(self):
+        '''Return I{a copy of} this instance, negated.
+        '''
+        return self._copy_n(self.__neg__)
+
+    def __pos__(self):
+        '''Return this instance I{as-is}, like C{float.__pos__()}.
+        '''
+        return self if _pos_self else self._copy_2(self.__pos__)
+
+    def __pow__(self, other, *mod):  # PYCHOK 2 vs 3 args
+        '''Return C{B{self}**B{other}} as an L{Fsum}.
+
+           @see: Method L{Fsum.__ipow__}.
+        '''
+        f = self._copy_2(self.__pow__)
+        return f._fpow(other, _pow_op_, *mod)
+
+    def __radd__(self, other):
+        '''Return C{B{other} + B{self}} as an L{Fsum}.
+
+           @see: Method L{Fsum.__iadd__}.
+        '''
+        f = self._copy_r2(other, self.__radd__)
+        return f._fadd(self, _add_op_)
+
+    def __rdivmod__(self, other):
+        '''Return C{divmod(B{other}, B{self})} as 2-tuple C{(quotient,
+           remainder)}.
+
+           @see: Method L{Fsum.__divmod__}.
+        '''
+        f = self._copy_r2(other, self.__rdivmod__)
+        return f._fdivmod2(self, _divmod_op_)
+
+#   def __repr__(self):
+#       '''Return the default C{repr(this)}.
+#       '''
+#       return self.toRepr(lenc=True)
+
+    def __rfloordiv__(self, other):
+        '''Return C{B{other} // B{self}} as an L{Fsum}.
+
+           @see: Method L{Fsum.__ifloordiv__}.
+        '''
+        f = self._copy_r2(other, self.__rfloordiv__)
+        return f._floordiv(self, _floordiv_op_)
+
+    def __rmatmul__(self, other):  # PYCHOK no cover
+        '''Not implemented.'''
+        return _NotImplemented(self, other)
+
+    def __rmod__(self, other):
+        '''Return C{B{other} % B{self}} as an L{Fsum}.
+
+           @see: Method L{Fsum.__imod__}.
+        '''
+        f = self._copy_r2(other, self.__rmod__)
+        return f._fdivmod2(self, _mod_op_)[1]
+
+    def __rmul__(self, other):
+        '''Return C{B{other} * B{self}} as an L{Fsum}.
+
+           @see: Method L{Fsum.__imul__}.
+        '''
+        f = self._copy_r2(other, self.__rmul__)
+        return f._fmul(self, _mul_op_)
+
+    def __round__(self, *ndigits):  # PYCHOK no cover
+        '''Return C{round(B{self}, *B{ndigits}} as an L{Fsum}.
+
+           @arg ndigits: Optional number of digits (C{int}).
+        '''
+        # <https://docs.Python.org/3.12/reference/datamodel.html?#object.__round__>
+        f = Fsum(name=self.__round__.__name__)
+        f._n  = 1
+        f._ps = [round(float(self), *ndigits)]  # can be C{int}
+        return f
+
+    def __rpow__(self, other, *mod):
+        '''Return C{B{other}**B{self}} as an L{Fsum}.
+
+           @see: Method L{Fsum.__ipow__}.
+        '''
+        f = self._copy_r2(other, self.__rpow__)
+        return f._fpow(self, _pow_op_, *mod)
+
+    def __rsub__(self, other):
+        '''Return C{B{other} - B{self}} as L{Fsum}.
+
+           @see: Method L{Fsum.__isub__}.
+        '''
+        f = self._copy_r2(other, self.__rsub__)
+        return f._fsub(self, _sub_op_)
+
+    def __rtruediv__(self, other):
+        '''Return C{B{other} / B{self}} as an L{Fsum}.
+
+           @see: Method L{Fsum.__itruediv__}.
+        '''
+        f = self._copy_r2(other, self.__rtruediv__)
+        return f._ftruediv(self, _truediv_op_)
+
+    def __str__(self):
+        '''Return the default C{str(self)}.
+        '''
+        return self.toStr(lenc=True)
+
+    def __sub__(self, other):
+        '''Return C{B{self} - B{other}} as an L{Fsum}.
+
+           @arg other: An L{Fsum} or C{scalar}.
+
+           @return: The difference (L{Fsum}).
+
+           @see: Method L{Fsum.__isub__}.
+        '''
+        f = self._copy_2(self.__sub__)
+        return f._fsub(other, _sub_op_)
+
+    def __truediv__(self, other):
+        '''Return C{B{self} / B{other}} as an L{Fsum}.
+
+           @arg other: An L{Fsum} or C{scalar} divisor.
+
+           @return: The quotient (L{Fsum}).
+
+           @see: Method L{Fsum.__itruediv__}.
+        '''
+        f = self._copy_2(self.__truediv__)
+        return f._ftruediv(other, _truediv_op_)
+
+    __trunc__ = __int__
+
+    if _sys_version_info2 < (3, 0):  # PYCHOK no cover
+        # <https://docs.Python.org/2/library/operator.html#mapping-operators-to-functions>
+        __div__     = __truediv__
+        __idiv__    = __itruediv__
+        __long__    = __int__
+        __nonzero__ = __bool__
+        __rdiv__    = __rtruediv__
+
+    def as_integer_ratio(self):
+        '''Return this instance as the ratio of 2 integers.
+
+           @return: 2-Tuple C{(numerator, denominator)} both
+                    C{int} and with positive C{denominator}.
+
+           @see: Standard C{float.as_integer_ratio} in Python 3+.
+        '''
+        n, r = self._fint2
+        if r:
+            i, d = r.as_integer_ratio()
+            n *= d
+            n += i
+        else:  # PYCHOK no cover
+            d = 1
+        return n, d
+
+    @property_RO
+    def ceil(self):
+        '''Get this instance' C{ceil} value (C{int} in Python 3+,
+           but C{float} in Python 2-).
+
+           @note: The C{ceil} takes the C{residual} into account.
+
+           @see: Method L{Fsum.int_float} and properties L{Fsum.floor},
+                 L{Fsum.imag} and L{Fsum.real}.
+        '''
+        s, r = self._fprs2
+        c = _ceil(s) + int(r) - 1
+        while r > (c - s):  # (s + r) > c
+            c += 1
+        return c
+
+    def _cmp_0(self, other, op):
+        '''(INTERNAL) Return C{scalar(self - B{other})} for 0-comparison.
+        '''
+        if isscalar(other):
+            if other:
+                s = _fsum(self._ps_1(other))
+            else:
+                s, r = self._fprs2
+                s = _signOf(s, -r)
+        elif isinstance(other, Fsum):
+            s = _fsum(self._ps_1(*other._ps))
+        else:
+            raise self._TypeError(op, other)  # txt=_invalid_
+        return s
+
+    def copy(self, deep=False, name=NN):
+        '''Copy this instance, C{shallow} or B{C{deep}}.
+
+           @return: The copy (L{Fsum}).
+         '''
+        f = _Named.copy(self, deep=deep, name=name)
+        f._n  = self._n if deep else 1
+        f._ps = list(self._ps)  # separate list
+        return f
+
+    def _copy_0(self, *xs):
+        '''(INTERNAL) Copy with/-out overriding C{partials}.
+        '''
+        # for x in xs:
+        #     assert isscalar(x)
+        f = self._Fsum(self._n + len(xs), *xs)
+        if self.name:
+            f._name = self.name  # .rename calls _update_attrs
+        return f
+
+    def _copy_2(self, which):
+        '''(INTERNAL) Copy for I{dyadic} operators.
+        '''
+        # NOT .classof due to .Fdot(a, *b) args, etc.
+        f = _Named.copy(self, deep=False, name=which.__name__)
+        # assert f._n == self._n
+        f._ps = list(self._ps)  # separate list
+        return f
+
+    def _copy_n(self, which):
+        '''(INTERNAL) Negated copy for I{monadic} C{__abs__} and C{__neg__}.
+        '''
+        if self._ps:
+            f = self._Fsum(self._n)
+            f._ps[:] = self._ps_n()
+#           f._facc_up(up=False)
+        else:
+            f = self._Fsum(self._n, _0_0)
+        f._name = which.__name__  # .rename calls _update_attrs
+        return f
+
+    def _copy_r2(self, other, which):
+        '''(INTERNAL) Copy for I{reverse-dyadic} operators.
+        '''
+        return other._copy_2(which) if isinstance(other, Fsum) else \
+               Fsum(other, name=which.__name__)  # see ._copy_2
+
+    def _copy_RESIDUAL(self, other):
+        '''(INTERNAL) Copy C{other._RESIDUAL}.
+        '''
+        R = other._RESIDUAL
+        if R is not Fsum._RESIDUAL:
+            self._RESIDUAL = R
+
+    def _copy_up(self, _fprs2=False):
+        '''(INTERNAL) Minimal, anonymous copy.
+        '''
+        f = self._Fsum(self._n, *self._ps)
+        if _fprs2:  # only the ._fprs2 2-tuple
+            Fsum._fprs2._update_from(f, self)
+        return f
+
+    def divmod(self, other):
+        '''Return C{divmod(B{self}, B{other})} as 2-tuple C{(quotient,
+           remainder)}.
+
+           @arg other: An L{Fsum} or C{scalar} divisor.
+
+           @return: 2-Tuple C{(quotient, remainder)}, with the C{quotient}
+                    an C{int} in Python 3+ or a C{float} in Python 2- and
+                    the C{remainder} an L{Fsum} instance.
+
+           @see: Method L{Fsum.__itruediv__}.
+        '''
+        f = self._copy_2(self.divmod)
+        return f._fdivmod2(other, _divmod_op_)
+
+    def _Error(self, op, other, Error, **txt):
+        '''(INTERNAL) Format an B{C{Error}} for C{{self} B{op} B{other}}.
+        '''
+        return Error(_SPACE_(self.toRepr(), op, repr(other)), **txt)
+
+    def _ErrorX(self, X, xs, **kwds):  # in .fmath
+        '''(INTERNAL) Format a caught exception.
+        '''
+        E, t = _xError2(X)
+        n = unstr(self.named3, *xs[:3], _ELLIPSIS=len(xs) > 3, **kwds)
+        return E(n, txt=t, cause=X)
+
+    def _facc(self, xs, up=True):  # from .elliptic._Defer.Fsum
+        '''(INTERNAL) Accumulate more known C{scalar}s.
+        '''
+        n, ps, _2s = 0, self._ps, _2sum
+        for x in xs:  # _iter()
+            # assert isscalar(x) and isfinite(x)
+            i = 0
+            for p in ps:
+                x, p = _2s(x, p)
+                if p:
+                    ps[i] = p
+                    i += 1
+            ps[i:] = [x]
+            n += 1
+        # assert self._ps is ps
+        if n:
+            self._n += n
+            # Fsum._px = max(Fsum._px, len(ps))
+            if up:
+                self._update()
+        return self
+
+    def _facc_(self, *xs, **up):
+        '''(INTERNAL) Accumulate all positional C{scalar}s.
+        '''
+        return self._facc(xs, **up) if xs else self
+
+#   def _facc_up(self, up=True):
+#       '''(INTERNAL) Update the C{partials}, by removing
+#          and re-accumulating the final C{partial}.
+#       '''
+#       while len(self._ps) > 1:
+#           p = self._ps.pop()
+#           if p:
+#               n = self._n
+#               self._facc_(p, up=False)
+#               self._n = n
+#               break
+#       return self._update() if up else self  # ._fpsqz()
+
+    def fadd(self, xs=()):
+        '''Add an iterable of C{scalar} or L{Fsum} instances
+           to this instance.
+
+           @arg xs: Iterable, list, tuple, etc. (C{scalar} or
+                    L{Fsum} instances).
+
+           @return: This instance (L{Fsum}).
+
+           @raise OverflowError: Partial C{2sum} overflow.
+
+           @raise TypeError: An invalid B{C{xs}} type, not C{scalar}
+                             nor L{Fsum}.
+
+           @raise ValueError: Invalid or non-finite B{C{xs}} value.
+        '''
+        if isinstance(xs, Fsum):
+            self._facc(xs._ps)
+        elif isscalar(xs):  # for backward compatibility
+            self._facc_(_2float(x=xs))  # PYCHOK no cover
+        elif xs:
+            self._facc(_2floats(xs))  # PYCHOK yield
+        return self
+
+    def fadd_(self, *xs):
+        '''Add all positional C{scalar} or L{Fsum} instances
+           to this instance.
+
+           @arg xs: Values to add (C{scalar} or L{Fsum} instances),
+                    all positional.
+
+           @return: This instance (L{Fsum}).
+
+           @raise OverflowError: Partial C{2sum} overflow.
+
+           @raise TypeError: An invalid B{C{xs}} type, not C{scalar}
+                             nor L{Fsum}.
+
+           @raise ValueError: Invalid or non-finite B{C{xs}} value.
+        '''
+        return self._facc(_2floats(xs, origin=1))  # PYCHOK yield
+
+    def _fadd(self, other, op):  # in .fmath.Fhorner
+        '''(INTERNAL) Apply C{B{self} += B{other}}.
+        '''
+        if isinstance(other, Fsum):
+            if other is self:
+                self._facc_(*other._ps)  # == ._facc(tuple(other._ps))
+            elif other._ps:
+                self._facc(other._ps)
+        elif not isscalar(other):
+            raise self._TypeError(op, other)  # txt=_invalid_
+        elif other:
+            self._facc_(other)
+        return self
+
+    fcopy   =   copy        # for backward compatibility
+    fdiv    = __itruediv__  # for backward compatibility
+    fdivmod = __divmod__    # for backward compatibility
+
+    def _fdivmod2(self, other, op):
+        '''(INTERNAL) C{divmod(B{self}, B{other})} as 2-tuple
+           (C{int} or C{float}, remainder C{self}).
+        '''
+        # result mostly follows CPython function U{float_divmod
+        # <https://GitHub.com/python/cpython/blob/main/Objects/floatobject.c>},
+        # but at least divmod(-3, 2) equals Cpython's result (-2, 1).
+        q = self._copy_up(_fprs2=True)._ftruediv(other, op).floor
+        if q:  # == float // other == floor(float / other)
+            self -= other * q
+
+        s = signOf(other)  # make signOf(self) == signOf(other)
+        if s and self.signOf() == -s:  # PYCHOK no cover
+            self += other
+            q -= 1
+
+#           t = self.signOf()
+#           if t and t != s:
+#               from pygeodesy.errors import _AssertionError
+#               raise self._Error(op, other, _AssertionError, txt=signOf.__name__)
+        return q, self  # q is C{int} in Python 3+, but C{float} in Python 2-
+
+    def _finite(self, other, op=None):
+        '''(INTERNAL) Return B{C{other}} if C{finite}.
+        '''
+        if _isfinite(other):
+            return other
+        raise ValueError(_not_finite_) if not op else \
+              self._ValueError(op, other, txt=_not_finite_)
+
+    def fint(self, raiser=True, name=NN):
+        '''Return this instance' current running sum as C{integer}.
+
+           @kwarg raiser: If C{True} throw a L{ResidualError} if the
+                          I{integer} residual is non-zero.
+           @kwarg name: Optional name (C{str}), overriding C{"fint"}.
+
+           @return: The C{integer} (L{Fsum}).
+
+           @raise ResidualError: Non-zero I{integer} residual.
+
+           @see: Methods L{Fsum.int_float} and L{Fsum.is_integer}.
+        '''
+        i, r = self._fint2
+        if r and raiser:
+            t = _stresidual(_integer_, r)
+            raise ResidualError(_integer_, i, txt=t)
+        n = name or self.fint.__name__
+        return Fsum(name=n)._fset(i, asis=True)
+
+    def fint2(self, **name):
+        '''Return this instance' current running sum as C{int} and
+           the I{integer} residual.
+
+           @kwarg name: Optional name (C{str}).
+
+           @return: An L{Fsum2Tuple}C{(fsum, residual)} with C{fsum}
+                    an C{int} and I{integer} C{residual} a C{float} or
+                    C{INT0} if the C{fsum} is considered to be I{exact}.
+        '''
+        return Fsum2Tuple(*self._fint2, **name)
+
+    @Property_RO
+    def _fint2(self):  # see ._fset
+        '''(INTERNAL) Get 2-tuple (C{int}, I{integer} residual).
+        '''
+        i =  int(self._fprs)  # int(self)
+        r = _fsum(self._ps_1(i)) if len(self._ps) > 1 else (
+               (self._ps[0] - i) if self._ps else -i)
+        return i, (r or INT0)
+
+    @deprecated_property_RO
+    def float_int(self):  # PYCHOK no cover
+        '''DEPRECATED, use method C{Fsum.int_float}.'''
+        return self.int_float()  # raiser=False
+
+    @property_RO
+    def floor(self):
+        '''Get this instance' C{floor} (C{int} in Python 3+, but
+           C{float} in Python 2-).
+
+           @note: The C{floor} takes the C{residual} into account.
+
+           @see: Method L{Fsum.int_float} and properties L{Fsum.ceil},
+                 L{Fsum.imag} and L{Fsum.real}.
+        '''
+        s, r = self._fprs2
+        f = _floor(s) + _floor(r) + 1
+        while r < (f - s):  # (s + r) < f
+            f -= 1
+        return f
+
+#   floordiv = __floordiv__  # for naming consistency
+
+    def _floordiv(self, other, op):  # rather _ffloordiv?
+        '''Apply C{B{self} //= B{other}}.
+        '''
+        q = self._ftruediv(other, op)  # == self
+        return self._fset(q.floor, asis=True)  # floor(q)
+
+    fmul = __imul__  # for backward compatibility
+
+    def _fmul(self, other, op):
+        '''(INTERNAL) Apply C{B{self} *= B{other}}.
+        '''
+        if isscalar(other):
+            f = self._mul_scalar(other, op)
+        elif not isinstance(other, Fsum):
+            raise self._TypeError(op, other)  # txt=_invalid_
+        elif len(self._ps) != 1:
+            f = self._mul_Fsum(other, op)
+        elif len(other._ps) != 1:  # len(self._ps) == 1
+            f = other._copy_up()._mul_scalar(self._ps[0], op)
+        else:  # len(other._ps) == len(self._ps) == 1
+            s = self._finite(self._ps[0] * other._ps[0])
+            return self._fset(s, asis=True, n=len(self) + 1)
+        return self._fset(f)
+
+    def fover(self, over):
+        '''Apply C{B{self} /= B{over}} and summate.
+
+           @arg over: An L{Fsum} or C{scalar} denominator.
+
+           @return: Precision running sum (C{float}).
+
+           @see: Methods L{Fsum.fsum} and L{Fsum.__itruediv__}.
+        '''
+        return float(self.fdiv(over)._fprs)
+
+    fpow = __ipow__  # for backward compatibility
+
+    def _fpow(self, other, op, *mod):
+        '''Apply C{B{self} **= B{other}}, optional B{C{mod}} or C{None}.
+        '''
+        if mod and mod[0] is not None:  # == 3-arg C{pow}
+            s = self._pow_3(other, mod[0], op)
+        elif mod and mod[0] is None and self.is_integer():
+            # return an exact C{int} for C{int}**C{int}
+            i =  self._copy_0(self._fint2[0])  # assert _fint2[1] == 0
+            x = _2scalar(other)  # C{int}, C{float} or other
+            s =  i._pow_2(x, other, op) if isscalar(x) else i._fpow(x, op)
+        else:  # pow(self, other) == pow(self, other, None)
+            p = None
+            if isinstance(other, Fsum):
+                x, r = other._fprs2
+                if r:
+                    if self._raiser(r, x):
+                        raise self._ResidualError(op, other, r)
+                    p =  self._pow_scalar(r, other, op)
+#                   p = _2scalar(p)  # _raiser = None
+            elif not isscalar(other):
+                raise self._TypeError(op, other)  # txt=_invalid_
+            else:
+                x = self._finite(other, op)
+            s = self._pow_scalar(x, other, op)
+            if p is not None:
+                s *= p
+        return self._fset(s, asis=isint(s), n=max(len(self), 1))
+
+    @Property_RO
+    def _fprs(self):
+        '''(INTERNAL) Get and cache this instance' precision
+           running sum (C{float} or C{int}), ignoring C{residual}.
+
+           @note: The precision running C{fsum} after a C{//=} or
+                  C{//} C{floor} division is C{int} in Python 3+.
+        '''
+        ps = self._ps
+        n = len(ps) - 1
+        if n > 1:
+            s = _psum(ps)
+        elif n > 0:  # len(ps) == 2
+            s, p = _2sum(*ps) if ps[1] else ps
+            ps[:] = ([p, s] if s else [p]) if p else [s]
+        elif n < 0:  # see L{Fsum.__init__}
+            s = _0_0
+            ps[:] = [s]
+        else:  # len(ps) == 1
+            s = ps[0]
+        # assert self._ps is ps
+        # assert Fsum._fprs2.name not in self.__dict__
+        return s
+
+    @Property_RO
+    def _fprs2(self):
+        '''(INTERNAL) Get and cache this instance' precision
+           running sum and residual (L{Fsum2Tuple}).
+        '''
+        s =  self._fprs
+        r = _fsum(self._ps_1(s)) if len(self._ps) > 1 else INT0
+        return Fsum2Tuple(s, r)  # name=Fsum.fsum2.__name__
+
+#   def _fpsqz(self):
+#       '''(INTERNAL) Compress, squeeze the C{partials}.
+#       '''
+#       if len(self._ps) > 2:
+#           _ = self._fprs
+#       return self
+
+    def _fset(self, other, asis=False, n=1):
+        '''(INTERNAL) Overwrite this instance with an other or a C{scalar}.
+        '''
+        if other is self:
+            pass  # from ._fmul, ._ftruediv and ._pow_scalar
+        elif isinstance(other, Fsum):
+            self._n     = other._n
+            self._ps[:] = other._ps
+            self._copy_RESIDUAL(other)
+            # use or zap the C{Property_RO} values
+            Fsum._fint2._update_from(self, other)
+            Fsum._fprs ._update_from(self, other)
+            Fsum._fprs2._update_from(self, other)
+        elif isscalar(other):
+            s = other if asis else float(other)
+            i = int(s)  # see ._fint2
+            t = i, ((s - i) or INT0)
+            self._n     =  n
+            self._ps[:] = [s]
+            # Property_RO _fint2, _fprs and _fprs2 can't be a Property:
+            # Property's _fset zaps the value just set by the @setter
+            self.__dict__.update(_fint2=t, _fprs=s, _fprs2=Fsum2Tuple(s, INT0))
+        else:  # PYCHOK no cover
+            raise self._TypeError(_fset_op_, other)  # txt=_invalid_
+        return self
+
+    def fsub(self, xs=()):
+        '''Subtract an iterable of C{scalar} or L{Fsum} instances
+           from this instance.
+
+           @arg xs: Iterable, list, tuple. etc. (C{scalar}
+                    or L{Fsum} instances).
+
+           @return: This instance, updated (L{Fsum}).
+
+           @see: Method L{Fsum.fadd}.
+        '''
+        return self._facc(_2floats(xs, sub=True)) if xs else self  # PYCHOK yield
+
+    def fsub_(self, *xs):
+        '''Subtract all positional C{scalar} or L{Fsum} instances
+           from this instance.
+
+           @arg xs: Values to subtract (C{scalar} or
+                    L{Fsum} instances), all positional.
+
+           @return: This instance, updated (L{Fsum}).
+
+           @see: Method L{Fsum.fadd}.
+        '''
+        return self._facc(_2floats(xs, origin=1, sub=True)) if xs else self  # PYCHOK yield
+
+    def _fsub(self, other, op):
+        '''(INTERNAL) Apply C{B{self} -= B{other}}.
+        '''
+        if isinstance(other, Fsum):
+            if other is self:  # or other._fprs2 == self._fprs2:
+                self._fset(_0_0, asis=True, n=len(self) * 2)  # self -= self
+            elif other._ps:
+                self._facc(other._ps_n())
+        elif not isscalar(other):
+            raise self._TypeError(op, other)  # txt=_invalid_
+        elif self._finite(other, op):
+            self._facc_(-other)
+        return self
+
+    def _Fsum(self, n, *ps):
+        '''(INTERNAL) New L{Fsum} instance.
+        '''
+        f = Fsum()
+        f._n = n
+        if ps:
+            f._ps[:] = ps
+        f._copy_RESIDUAL(self)
+        return f
+
+    def fsum(self, xs=()):
+        '''Add more C{scalar} or L{Fsum} instances and summate.
+
+           @kwarg xs: Iterable, list, tuple, etc. (C{scalar} or
+                      L{Fsum} instances).
+
+           @return: Precision running sum (C{float} or C{int}).
+
+           @see: Method L{Fsum.fadd}.
+
+           @note: Accumulation can continue after summation.
+        '''
+        f = self._facc(_2floats(xs)) if xs else self  # PYCHOK yield
+        return f._fprs
+
+    def fsum_(self, *xs):
+        '''Add all positional C{scalar} or L{Fsum} instances and summate.
+
+           @arg xs: Values to add (C{scalar} or L{Fsum} instances),
+                    all positional.
+
+           @return: Precision running sum (C{float} or C{int}).
+
+           @see: Methods L{Fsum.fsum} and L{Fsum.fsumf_}.
+        '''
+        f = self._facc(_2floats(xs, origin=1)) if xs else self  # PYCHOK yield
+        return f._fprs
+
+    def fsum2(self, xs=(), **name):
+        '''Add more C{scalar} or L{Fsum} instances and return the
+           current precision running sum and the C{residual}.
+
+           @kwarg xs: Iterable, list, tuple, etc. (C{scalar} or
+                      L{Fsum} instances).
+           @kwarg name: Optional name (C{str}).
+
+           @return: L{Fsum2Tuple}C{(fsum, residual)} with C{fsum} the
+                    current precision running sum and C{residual}, the
+                    (precision) sum of the remaining C{partials}.  The
+                    C{residual is INT0} if the C{fsum} is considered
+                    to be I{exact}.
+
+           @see: Methods L{Fsum.fint2}, L{Fsum.fsum} and L{Fsum.fsum2_}
+        '''
+        f = self._facc(_2floats(xs)) if xs else self  # PYCHOK yield
+        t = f._fprs2
+        if name:
+            t = t.dup(name=_xkwds_get(name, name=NN))
+        return t
+
+    def fsum2_(self, *xs):
+        '''Add any positional C{scalar} or L{Fsum} instances and return
+           the precision running sum and the C{differential}.
+
+           @arg xs: Values to add (C{scalar} or L{Fsum} instances),
+                    all positional.
+
+           @return: 2-Tuple C{(fsum, delta)} with the current precision
+                    running C{fsum} and C{delta}, the difference with
+                    the previous running C{fsum} (C{float}s).
+
+           @see: Methods L{Fsum.fsum_} and L{Fsum.fsum}.
+        '''
+        p, r = self._fprs2
+        if xs:
+            s, t = self._facc(_2floats(xs, origin=1))._fprs2  # PYCHOK yield
+            return s, _fsum((s, -p, r, -t))  # ((s - p) + (r - t))
+        else:  # PYCHOK no cover
+            return p, _0_0
+
+    def fsumf_(self, *xs):
+        '''Like method L{Fsum.fsum_} but only for known C{float B{xs}}.
+        '''
+        f = self._facc(xs) if xs else self  # PYCHOK yield
+        return f._fprs
+
+#   ftruediv = __itruediv__   # for naming consistency
+
+    def _ftruediv(self, other, op):
+        '''(INTERNAL) Apply C{B{self} /= B{other}}.
+        '''
+        n = _1_0
+        if isinstance(other, Fsum):
+            if other is self or other._fprs2 == self._fprs2:
+                return self._fset(_1_0, asis=True, n=len(self))
+            d, r = other._fprs2
+            if r:
+                if not d:  # PYCHOK no cover
+                    d = r
+                elif self._raiser(r, d):
+                    raise self._ResidualError(op, other, r)
+                else:
+                    d, n = other.as_integer_ratio()
+        elif isscalar(other):
+            d = other
+        else:  # PYCHOK no cover
+            raise self._TypeError(op, other)  # txt=_invalid_
+        try:
+            s = 0 if isinf(d) else (
+                d if isnan(d) else self._finite(n / d))
+        except Exception as x:
+            E, t = _xError2(x)
+            raise self._Error(op, other, E, txt=t)
+        f = self._mul_scalar(s, _mul_op_)  # handles 0, NAN, etc.
+        return self._fset(f)
+
+    @property_RO
+    def imag(self):
+        '''Get the C{imaginary} part of this instance (C{0.0}, always).
+
+           @see: Properties L{Fsum.ceil}, L{Fsum.floor} and L{Fsum.real}.
+        '''
+        return _0_0
+
+    def int_float(self, raiser=False):
+        '''Return this instance' current running sum as C{int} or C{float}.
+
+           @kwarg raiser: If C{True} throw a L{ResidualError} if the
+                          residual is non-zero.
+
+           @return: This C{integer} sum if this instance C{is_integer},
+                    otherwise return the C{float} sum if the residual
+                    is zero or if C{B{raiser}=False}.
+
+           @raise ResidualError: Non-zero residual and C{B{raiser}=True}.
+
+           @see: Methods L{Fsum.fint} and L{Fsum.fint2}.
+        '''
+        s, r = self._fint2
+        if r:
+            s, r = self._fprs2
+            if r and raiser:  # PYCHOK no cover
+                t = _stresidual(_non_zero_, r)
+                raise ResidualError(int_float=s, txt=t)
+            s = float(s)  # redundant
+        return s
+
+    def is_exact(self):
+        '''Is this instance' current running C{fsum} considered to
+           be exact? (C{bool}).
+        '''
+        return self.residual is INT0
+
+    def is_integer(self):
+        '''Is this instance' current running sum C{integer}? (C{bool}).
+
+           @see: Methods L{Fsum.fint} and L{Fsum.fint2}.
+        '''
+        _, r = self._fint2
+        return False if r else True
+
+    def is_math_fsum(self):
+        '''Return whether functions L{fsum}, L{fsum_}, L{fsum1}
+           and L{fsum1_} plus partials summation are based on
+           Python's C{math.fsum} or not.
+
+           @return: C{2} if all functions and partials summation
+                    are based on C{math.fsum}, C{True} if only
+                    the functions are based on C{math.fsum} (and
+                    partials summation is not) or C{False} if
+                    none are.
+        '''
+        f = Fsum._math_fsum
+        return 2 if _psum is f else bool(f)
+
+    def _mul_Fsum(self, other, op=_mul_op_):
+        '''(INTERNAL) Return C{B{self} * Fsum B{other}} as L{Fsum}.
+        '''
+        # assert isinstance(other, Fsum)
+        return self._copy_0()._facc(self._ps_x(op, *other._ps), up=False)
+
+    def _mul_scalar(self, factor, op):
+        '''(INTERNAL) Return C{B{self} * scalar B{factor}} as L{Fsum} or C{0}.
+        '''
+        # assert isscalar(factor)
+        if self._finite(factor, op) and self._ps:
+            if factor == _1_0:
+                return self
+            f = self._copy_0()._facc(self._ps_x(op, factor), up=False)
+        else:
+            f = self._copy_0(_0_0)
+        return f
+
+    @property_RO
+    def partials(self):
+        '''Get this instance' current partial sums (C{tuple} of C{float}s and/or C{int}s).
+        '''
+        return tuple(self._ps)
+
+    def pow(self, x, *mod):
+        '''Return C{B{self}**B{x}} as L{Fsum}.
+
+           @arg x: The exponent (L{Fsum} or C{scalar}).
+           @arg mod: Optional modulus (C{int} or C{None}) for the 3-argument
+                     C{pow(B{self}, B{other}, B{mod})} version.
+
+           @return: The C{pow(self, B{x})} or C{pow(self, B{x}, *B{mod})}
+                    result (L{Fsum}).
+
+           @note: If B{C{mod}} is given as C{None}, the result will be an
+                  C{integer} L{Fsum} provided this instance C{is_integer}
+                  or set C{integer} with L{Fsum.fint}.
+
+           @see: Methods L{Fsum.__ipow__}, L{Fsum.fint} and L{Fsum.is_integer}.
+        '''
+        f = self._copy_2(self.pow)
+        if f and isint(x) and x >= 0 and not mod:
+            f._pow_int(x, x, _pow_op_)  # f **= x
+        else:
+            f._fpow(x, _pow_op_, *mod)  # f = pow(f, x, *mod)
+        return f
+
+    def _pow_0_1(self, x, other):
+        '''(INTERNAL) Return B{C{self}**1} or C{B{self}**0 == 1.0}.
+        '''
+        return self if x else (1 if self.is_integer() and isint(other) else _1_0)
+
+    def _pow_2(self, x, other, op):
+        '''(INTERNAL) 2-arg C{pow(B{self}, scalar B{x})} embellishing errors.
+        '''
+        # assert len(self._ps) == 1 and isscalar(x)
+        b = self._ps[0]  # assert isscalar(b)
+        try:  # type(s) == type(x) if x in (_1_0, 1)
+            s = pow(b, x)  # -1**2.3 == -(1**2.3)
+            if not iscomplex(s):
+                return self._finite(s)  # 0**INF == 0.0, 1**INF==1.0
+            # neg**frac == complex in Python 3+, but ValueError in 2-
+            E, t = _ValueError, _strcomplex(s, b, x)  # PYCHOK no cover
+        except Exception as x:
+            E, t = _xError2(x)
+        raise self._Error(op, other, E, txt=t)
+
+    def _pow_3(self, other, mod, op):
+        '''(INTERNAL) 3-arg C{pow(B{self}, B{other}, int B{mod} or C{None})}.
+        '''
+        b, r = self._fprs2 if mod is None else self._fint2
+        if r and self._raiser(r, b):
+            t = _non_zero_ if mod is None else _integer_
+            E, t = ResidualError, _stresidual(t, r, mod=mod)
+        else:
+            try:  # b, other, mod all C{int}, unless C{mod} is C{None}
+                x = _2scalar(other, _raiser=self._raiser)
+                s =  pow(b, x, mod)
+                if not iscomplex(s):
+                    return self._finite(s)
+                # neg**frac == complex in Python 3+, but ValueError in 2-
+                E, t = _ValueError, _strcomplex(s, b, x, mod)  # PYCHOK no cover
+            except Exception as x:
+                E, t = _xError2(x)
+                t = _COMMASPACE_(Fmt.PARENSPACED(mod=mod), t)
+        raise self._Error(op, other, E, txt=t)
+
+    def _pow_int(self, x, other, op):
+        '''(INTERNAL) Return C{B{self} **= B{x}} for C{int B{x} >= 0}.
+        '''
+        # assert isint(x) and x >= 0
+        if len(self._ps) > 1:
+            if x > 2:
+                p = self._copy_up()
+                m = 1  # single-bit mask
+                if x & m:
+                    x -= m  # x ^= m
+                    f  = p._copy_up()
+                else:
+                    f = self._copy_0(_1_0)
+                while x:
+                    p  = p._mul_Fsum(p, op)  # p **= 2
+                    m += m  # m <<= 1
+                    if x & m:
+                        x -= m  # x ^= m
+                        f  = f._mul_Fsum(p, op)  # f *= p
+            elif x > 1:  # self**2
+                f = self._mul_Fsum(self, op)
+            else:  # self**1 or self**0
+                f = self._pow_0_1(x, other)
+        elif self._ps:  # self._ps[0]**x
+            f = self._pow_2(x, other, op)
+        else:  # PYCHOK no cover
+            # 0**pos_int == 0, but 0**0 == 1
+            f = 0 if x else 1  # like ._fprs
+        return self._fset(f, asis=isint(f), n=len(self))
+
+    def _pow_scalar(self, x, other, op):
+        '''(INTERNAL) Return C{self**B{x}} for C{scalar B{x}}.
+        '''
+        s, r = self._fprs2
+        if isint(x, both=True):
+            x = int(x)  # Fsum**int
+            y = abs(x)
+            if y > 1:
+                if r:
+                    f = self._copy_up()._pow_int(y, other, op)
+                    if x > 0:  # > 1
+                        return f
+                    # assert x < 0  # < -1
+                    s, r = f._fprs2
+                    if r:
+                        return self._copy_0(_1_0)._ftruediv(f, op)
+                    # use **= -1 for the CPython float_pow
+                    # error if s is zero, and not s = 1 / s
+                    x = -1
+#           elif y > 1:  # self**2 or self**-2
+#               f = self._mul_Fsum(self, op)
+#               if x < 0:
+#                   f = f._copy_0(_1_0)._ftruediv(f, op)
+#               return f
+            elif x < 0:  # self**-1 == 1 / self
+                if r:
+                    return self._copy_0(_1_0)._ftruediv(self, op)
+            else:  # self**1 or self**0
+                return self._pow_0_1(x, other)  # self or 0.0
+        elif not isscalar(x):  # assert ...
+            raise self._TypeError(op, other, txt=_not_scalar_)
+        elif r and self._raiser(r, s):  # non-zero residual**fractional
+            # raise self._ResidualError(op, other, r, fractional_power=x)
+            t = _stresidual(_non_zero_, r, fractional_power=x)
+            raise self._Error(op, other, ResidualError, txt=t)
+        # assert isscalar(s) and isscalar(x)
+        return self._copy_0(s)._pow_2(x, other, op)
+
+    def _ps_1(self, *less):
+        '''(INTERNAL) Yield partials, 1-primed and subtract any C{less}.
+        '''
+        yield _1_0
+        for p in self._ps:
+            if p:
+                yield p
+        for p in less:
+            if p:
+                yield -p
+        yield _N_1_0
+
+    def _ps_n(self):
+        '''(INTERNAL) Yield partials, negated.
+        '''
+        for p in self._ps:
+            if p:
+                yield -p
+
+    def _ps_x(self, op, *factors):  # see .fmath.Fhorner
+        '''(INTERNAL) Yield all C{partials} times each B{C{factor}},
+           in total, up to C{len(partials) * len(factors)} items.
+        '''
+        ps = self._ps
+        if len(ps) < len(factors):
+            ps, factors = factors, ps
+        _f = _isfinite
+        for f in factors:
+            for p in ps:
+                p *= f
+                if _f(p):
+                    yield p
+                else:  # PYCHOK no cover
+                    self._finite(p, op)  # throw ValueError
+
+    @property_RO
+    def real(self):
+        '''Get the C{real} part of this instance (C{float}).
+
+           @see: Methods L{Fsum.__float__} and L{Fsum.fsum}
+                 and properties L{Fsum.ceil}, L{Fsum.floor},
+                 L{Fsum.imag} and L{Fsum.residual}.
+        '''
+        return float(self._fprs)
+
+    @property_RO
+    def residual(self):
+        '''Get this instance' residual (C{float} or C{int}), the
+           C{sum(partials)} less the precision running sum C{fsum}.
+
+           @note: If the C{residual is INT0}, the precision running
+                  C{fsum} is considered to be I{exact}.
+
+           @see: Methods L{Fsum.fsum}, L{Fsum.fsum2} and L{Fsum.is_exact}.
+        '''
+        return self._fprs2.residual
+
+    def _raiser(self, r, s):
+        '''(INTERNAL) Does the ratio C{r / s} exceed threshold?
+        '''
+        self._ratio = t = fabs((r / s) if s else r)
+        return t > self._RESIDUAL
+
+    def RESIDUAL(self, *threshold):
+        '''Get and set this instance' I{ratio} for raising L{ResidualError}s,
+           overriding the default from env variable C{PYGEODESY_FSUM_RESIDUAL}.
+
+           @arg threshold: If C{scalar}, the I{ratio} to exceed for raising
+                           L{ResidualError}s in division and exponention, if
+                           C{None} restore the default set with env variable
+                           C{PYGEODESY_FSUM_RESIDUAL} or if omitted, keep the
+                           current setting.
+
+           @return: The previous C{RESIDUAL} setting (C{float}).
+
+           @raise ValueError: Negative B{C{threshold}}.
+
+           @note: A L{ResidualError} is thrown if the non-zero I{ratio}
+                  C{residual} / C{fsum} exceeds the B{C{threshold}}.
+        '''
+        r = self._RESIDUAL
+        if threshold:
+            t = threshold[0]
+            t = Fsum._RESIDUAL if t is None else (
+                      float(t) if isscalar(t) else (  # for backward ...
+                          _0_0 if bool(t) else _1_0))  # ... compatibility
+            if t < 0:
+                u = _DOT_(self, unstr(self.RESIDUAL, *threshold))
+                raise _ValueError(u, RESIDUAL=t, txt=_negative_)
+            self._RESIDUAL = t
+        return r
+
+    def _ResidualError(self, op, other, residual):
+        '''(INTERNAL) Non-zero B{C{residual}} etc.
+        '''
+        t = _stresidual(_non_zero_, residual, ratio=self._ratio,
+                                            RESIDUAL=self._RESIDUAL)
+        t =  t.replace(_COMMASPACE_R_, _exceeds_R_)
+        return self._Error(op, other, ResidualError, txt=t)
+
+    def signOf(self, res=True):
+        '''Determine the sign of this instance.
+
+           @kwarg res: If C{True} consider, otherwise
+                       ignore the residual (C{bool}).
+
+           @return: The sign (C{int}, -1, 0 or +1).
+        '''
+        s, r = self._fprs2 if res else (self._fprs, 0)
+        return _signOf(s, -r)
+
+    def toRepr(self, **prec_sep_fmt_lenc):  # PYCHOK signature
+        '''Return this C{Fsum} instance as representation.
+
+           @kwarg prec_sep_fmt_lenc: Optional keyword arguments for
+                       method L{Fsum2Tuple.toRepr} plus C{B{lenc}=True}
+                       (C{bool}) to in-/exclude the current C{[len]}
+                       of this L{Fsum} enclosed in I{[brackets]}.
+
+           @return: This instance (C{repr}).
+        '''
+        return self._toT(self._fprs2.toRepr, **prec_sep_fmt_lenc)
+
+    def toStr(self, **prec_sep_fmt_lenc):  # PYCHOK signature
+        '''Return this C{Fsum} instance as string.
+
+           @kwarg prec_sep_fmt_lenc: Optional keyword arguments for
+                       method L{Fsum2Tuple.toStr} plus C{B{lenc}=True}
+                       (C{bool}) to in-/exclude the current C{[len]}
+                       of this L{Fsum} enclosed in I{[brackets]}.
+
+           @return: This instance (C{str}).
+        '''
+        return self._toT(self._fprs2.toStr, **prec_sep_fmt_lenc)
+
+    def _toT(self, toT, fmt=Fmt.g, lenc=True, **kwds):
+        '''(INTERNAL) Helper for C{toRepr} and C{toStr}.
+        '''
+        n = self.named3
+        if lenc:
+            n = Fmt.SQUARE(n, len(self))
+        return _SPACE_(n, toT(fmt=fmt, **kwds))
+
+    def _TypeError(self, op, other, **txt):  # PYCHOK no cover
+        '''(INTERNAL) Return a C{TypeError}.
+        '''
+        return self._Error(op, other, _TypeError, **txt)
+
+    def _update(self):  # see ._fset
+        '''(INTERNAL) Zap all cached C{Property_RO} values.
+        '''
+        Fsum._fint2._update(self)
+        Fsum._fprs ._update(self)
+        Fsum._fprs2._update(self)
+        return self
+
+    def _ValueError(self, op, other, **txt):  # PYCHOK no cover
+        '''(INTERNAL) Return a C{ValueError}.
+        '''
+        return self._Error(op, other, _ValueError, **txt)
+
+    def _ZeroDivisionError(self, op, other, **txt):  # PYCHOK no cover
+        '''(INTERNAL) Return a C{ZeroDivisionError}.
+        '''
+        return self._Error(op, other, _ZeroDivisionError, **txt)
+
+_allPropertiesOf_n(3, Fsum, Property_RO)  # PYCHOK assert, see Fsum._fset, -._update
+
+
+def _Float_Int(arg, **name_Error):
+    '''(INTERNAL) Unit of L{Fsum2Tuple} items.
+    '''
+    U = Int if isint(arg) else Float
+    return U(arg, **name_Error)
+
+
+class Fsum2Tuple(_NamedTuple):
+    '''2-Tuple C{(fsum, residual)} with the precision running C{fsum}
+       and the C{residual}, the sum of the remaining partials.  Each
+       item is either C{float} or C{int}.
+
+       @note: If the C{residual is INT0}, the C{fsum} is considered
+              to be I{exact}, see method L{Fsum2Tuple.is_exact}.
+    '''
+    _Names_ = ( Fsum.fsum.__name__, Fsum.residual.name)
+    _Units_ = (_Float_Int,         _Float_Int)
+
+    @Property_RO
+    def Fsum(self):
+        '''Get this L{Fsum2Tuple} as an L{Fsum}.
+        '''
+        f = Fsum(name=self.name)
+        return f._copy_0(*(s for s in reversed(self) if s))
+
+    def is_exact(self):
+        '''Is this L{Fsum2Tuple} considered to be exact? (C{bool}).
+        '''
+        return self.Fsum.is_exact()
+
+    def is_integer(self):
+        '''Is this L{Fsum2Tuple} C{integer}? (C{bool}).
+        '''
+        return self.Fsum.is_integer()
+
+
+class ResidualError(_ValueError):
+    '''Error raised for an operation involving a L{pygeodesy.sums.Fsum}
+       instance with a non-zero C{residual}, I{integer} or otherwise.
+
+       @see: Module L{pygeodesy.fsums} and method L{Fsum.RESIDUAL}.
+    '''
+    pass
+
+
+try:
+    from math import fsum as _fsum  # precision IEEE-754 sum, Python 2.6+
+
+    # make sure _fsum works as expected (XXX check
+    # float.__getformat__('float')[:4] == 'IEEE'?)
+    if _fsum((1, 1e101, 1, -1e101)) != 2:  # PYCHOK no cover
+        del _fsum  # nope, remove _fsum ...
+        raise ImportError  # ... use _fsum below
+
+    Fsum._math_fsum = _sum = _fsum  # PYCHOK exported
+
+    if _getenv('PYGEODESY_FSUM_PARTIALS', _fsum.__name__) == _fsum.__name__:
+        _psum = _fsum  # PYCHOK redef
+
+except ImportError:
+    _sum = sum  # Fsum(NAN) exception fall-back
+
+    def _fsum(xs):
+        '''(INTERNAL) Precision summation, Python 2.5-.
+        '''
+        return Fsum(name=_fsum.__name__)._facc(xs, up=False)._fprs
+
+
+def fsum(xs, floats=False):
+    '''Precision floating point summation based on or like Python's C{math.fsum}.
+
+       @arg xs: Iterable, list, tuple, etc. of values (C{scalar} or
+                L{Fsum} instances).
+       @kwarg floats: Optionally, use C{B{floats}=True} iff I{all}
+                      B{C{xs}} are known to be C{float}.
+
+       @return: Precision C{fsum} (C{float}).
+
+       @raise OverflowError: Partial C{2sum} overflow.
+
+       @raise TypeError: Non-scalar B{C{xs}} value.
+
+       @raise ValueError: Invalid or non-finite B{C{xs}} value.
+
+       @note: Exceptions and I{non-finite} handling may differ if not
+              based on Python's C{math.fsum}.
+
+       @see: Class L{Fsum} and methods L{Fsum.fsum} and L{Fsum.fadd}.
+    '''
+    return _fsum(xs if floats else _2floats(xs)) if xs else _0_0  # PYCHOK yield
+
+
+def fsum_(*xs, **floats):
+    '''Precision floating point summation of all positional arguments.
+
+       @arg xs: Values to be added (C{scalar} or L{Fsum} instances),
+                all positional.
+       @kwarg floats: Optionally, use C{B{floats}=True} iff I{all}
+                      B{C{xs}} are known to be C{float}.
+
+       @return: Precision C{fsum} (C{float}).
+
+       @see: Function C{fsum}.
+    '''
+    return _fsum(xs if _xkwds_get(floats, floats=False) else
+                _2floats(xs, origin=1)) if xs else _0_0  # PYCHOK yield
+
+
+def fsumf_(*xs):
+    '''Precision floating point summation L{fsum_}C{(*xs, floats=True)}.
+    '''
+    return _fsum(xs) if xs else _0_0
+
+
+def fsum1(xs, floats=False):
+    '''Precision floating point summation of a few arguments, 1-primed.
+
+       @arg xs: Iterable, list, tuple, etc. of values (C{scalar} or
+                L{Fsum} instances).
+       @kwarg floats: Optionally, use C{B{floats}=True} iff I{all}
+                      B{C{xs}} are known to be C{float}.
+
+       @return: Precision C{fsum} (C{float}).
+
+       @see: Function C{fsum}.
+    '''
+    return _fsum(_1primed(xs if floats else _2floats(xs))) if xs else _0_0  # PYCHOK yield
+
+
+def fsum1_(*xs, **floats):
+    '''Precision floating point summation of a few arguments, 1-primed.
+
+       @arg xs: Values to be added (C{scalar} or L{Fsum} instances),
+                all positional.
+       @kwarg floats: Optionally, use C{B{floats}=True} iff I{all}
+                      B{C{xs}} are known to be C{float}.
+
+       @return: Precision C{fsum} (C{float}).
+
+       @see: Function C{fsum}
+    '''
+    return _fsum(_1primed(xs if _xkwds_get(floats, floats=False) else
+                         _2floats(xs, origin=1))) if xs else _0_0  # PYCHOK yield
+
+
+def fsum1f_(*xs):
+    '''Precision floating point summation L{fsum1_}C{(*xs, floats=True)}.
+    '''
+    return _fsum(_1primed(xs)) if xs else _0_0
+
+
+# **) MIT License
+#
+# Copyright (C) 2016-2024 -- mrJean1 at Gmail -- All Rights Reserved.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included
+# in all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+# OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+# OTHER DEALINGS IN THE SOFTWARE.