snappy 3.2__cp310-cp310-macosx_10_12_x86_64.whl

snappy/decorated_isosig.py

@@ -0,0 +1,710 @@
+"""
+We decorate Regina's triangulation isomorphism signature (isosig) to
+record the peripheral structure of a cusped manifold M, that is, the
+cusp labels and the peripheral curves on each cusp. The basic idea is
+to store these relative to SnapPy's combinatorial defaults for the
+triangulation created from the bare isosig.
+
+Specifically, if M has n cusps, we append a permutation on {0,...,n-1}
+as well as n change-of-basis matrices, represented as a sequence of 5n
+integers and encoded as a string of isosig characters. This decoration
+string is appended to the isosig string, after first appending a
+separator character which is not a valid isosig character.  To save
+space, the permutation may be omitted when it is equal to the identity
+permutation; this is indicated by the fact that the length of the
+decoration is 4n rather than 5n.
+
+A simple valid decorated isosig for a two-cusped manifold is::
+
+    eLPkbdcddhgggb_abBaaBBaaB
+
+Here, the bare isosig is what precedes the underscore; what follows is
+an encoded version of the 5n integers mentioned above.  This decorated
+isosig is equivalent to
+
+    eLPkbdcddhgggb_BaaBBaaB
+
+where the permutation part has been elided since the permutation is
+the identity.
+
+In practice, one can extract the isosig and decorator from a decorated
+isosig, say named di, as follows:
+
+isosig, decorator = di.split('_')
+
+Note: An isosig is an invariant of a triangulation of an *unoriented*
+manifold.  For an amphicheiral manifold M, it can happen that
+Manifold(M.triangulation_isosig(decorated=False)) has the opposite
+orientation from M itself.
+The decoration implicitly embeds the preferred orientation of
+M in the sign of the determinant of the change-of-basis matrices.
+
+Note: If the triangulation has combinatorial symmetries, there can be
+multiple change-of-basis matrices that yield combinatorially
+isomorphic pairs (triangulation, peripheral curves).  In such cases,
+the decoration is the lexicographically first one.
+
+Caveat: We pick the decoration with the lexicographically smallest
+encoding with the following consequence: If we have more 26 cusps, the
+lexicographically smallest permutation might not have the smallest encoding
+and thus might not be the one picked.
+
+Caveat: We drop the trivial permutation from the encoding. Pairs (a,b) of string
+come with the lexicographic ordering. We also obtain a (partial) ordering by
+ordering by a + b. These two orderings are not the same.
+In particular, there is a combinatorial isomorphism from
+Triangulation('L6n1') to
+Triangulation(Triangulation('L6n1').triangulation_isosig(decorated = False))
+that acts on the cusp by the identity perm and, thus, we would expect it to
+be preferred. However,
+Triangulation("L6n1(0,0)(0,0)(0,0)").triangulation_isosig() results in
+'gLMzQbcdefffaelaaai_acbBaabCbbabbBC' which does not use the identity perm.
+
+Caveat: There are de-facto two canonical choices of peripheral curves.
+When calling
+>>> T = Triangulation('ovLMvvPQQQccddlmnijklmnmnlgvfamtvfblhaumx'),
+the SnapPea kernel picks peripheral curves and then orients the manifold
+(see data_to_triangulation in kernel_code/kernel/triangulation.c)
+>>> T.set_peripheral_curves('combinatorial')
+is now calling the same SnapPea kernel code to pick peripheral curve but
+on the oriented manifold. This can result in different peripheral curves.
+
+Note that the encoding and decoding needs to use the same of the two
+canonical choices of peripheral curves.
+
+For the decoding, there is a difference based on whether the isosig is
+decorated because set_peripheral_from_decoration calls
+manifold.set_peripheral_curves('combinatorial').
+
+We need to account for that in the encoding: we need to use
+set_peripheral_curves('combinatorial') on the "target" triangulation if
+we anticipate a decoration. And if we called
+set_peripheral_curves('combinatorial'), we need to make sure we have a
+decoration (see force_decoration).
+"""
+
+import re
+import string
+
+from .matrix import make_vector
+
+# Used between the base isosig and the decorated version.
+separator = '_'
+
+# Pattern matching decorated isosigs
+
+base64_pat = r'([a-zA-Z0-9\+\-]+)'
+separator_pat = '[%s]{1}' % separator
+base64_opt_pat = r'([a-zA-Z0-9\+\-]*)'
+isosig_pattern = re.compile(base64_pat + separator_pat + base64_opt_pat + '$')
+
+# We store lists of integers as (non-RFC4648) base64 strings.
+
+base64_letters = string.ascii_letters + '0123456789+-'
+base64_lower = string.ascii_lowercase + '01234+'
+base64_upper = string.ascii_uppercase + '56789-'
+in_one = string.ascii_lowercase[:16] + string.ascii_lowercase[1:16].upper()
+
+int_to_letter = dict(enumerate(base64_letters))
+letter_to_int = {a: i for i, a in enumerate(base64_letters)}
+
+
+def encode_nonnegative_int(x):
+    """
+    Regina's base64 encoding scheme for nonnegative integers,
+    described in the appendix to http://arxiv.org/abs/1110.6080
+    """
+    assert x >= 0
+    six_bits = []
+    while True:
+        low_six = x & 63
+        six_bits.append(low_six)
+        x = x >> 6
+        if x == 0:
+            break
+    return ''.join([int_to_letter[b] for b in six_bits])
+
+
+def decode_nonnegative_int(s):
+    return sum( letter_to_int[a] << 6*i for i, a in enumerate(s))
+
+
+def encode_int(x):
+    """
+    Encodes an integer in the range [-2**90 + 1, 2**90 - 1] with a "stop"
+    at the end so a concatenation of such encodings is easily decodable.
+    The basic format is:
+
+    If x in [0...15], encode as a single letter in [a...p].
+    If x in [-15...-1] encode as a single letter in [P...B].
+
+    Otherwise, the first letter specifies the length of
+    encode_nonnegative_int(abs(x)) as well as sign(x), followed by the
+    encoding of abs(x).
+    """
+    if 0 <= x < 16:
+        return base64_letters[x]
+    if -15 <= x < 0:
+        return base64_letters[abs(x) + 26]
+    encoded_xabs = encode_nonnegative_int(abs(x))
+    L = len(encoded_xabs)
+    try:
+        if x > 0:
+            return base64_lower[16 + L] + encoded_xabs
+        if x < 0:
+            return base64_upper[16 + L] + encoded_xabs
+    except IndexError:
+        raise ValueError('The given integer is too large to encode')
+
+
+def encode_integer_list(L):
+    return ''.join(map(encode_int, L))
+
+
+def decode_integer_list(encoded):
+    ans = []
+    while len(encoded):
+        s = encoded[0]
+        sign = 1 if s in base64_lower else -1
+        if s in in_one:
+            ans.append(sign*letter_to_int[s.lower()])
+            encoded = encoded[1:]
+        else:
+            if sign == 1:
+                L = base64_lower.index(s) - 16
+            else:
+                L = base64_upper.index(s) - 16
+            current, encoded = encoded[1:L+1], encoded[L+1:]
+            ans.append(sign*decode_nonnegative_int(current))
+    return ans
+
+###############################################################################
+# Some helper functions
+
+def det(A):
+    return A[0][0]*A[1][1] - A[0][1]*A[1][0]
+
+
+def inverse_perm(L):
+    ans = len(L)*[None]
+    for i, x in enumerate(L):
+        ans[x] = i
+    return ans
+
+
+def as_two_by_two_matrices(L):
+    assert len(L) % 4 == 0
+    return [[(L[i], L[i+1]), (L[i+2], L[i+3])] for i in range(0, len(L), 4)]
+
+def first_non_zero_entry_in_column(matrix, col):
+    e = matrix[0, col] 
+    if e != 0:
+        return e
+    return matrix[1, col]
+
+def sgn_column(matrix, col):
+    """
+    Returns +1 or -1 depending on the sign of the first non-zero entry
+    in the column of the given matrix.
+    """
+    if first_non_zero_entry_in_column(matrix, col) > 0:
+        return +1
+    else:
+        return -1
+
+def apply_peripheral_curve_flips(
+        matrix, slope, manifold_orientable, isomorphism_orientation):
+    """
+    Flips peripheral curves (as encoded by matrix) to bring them into
+    canonical form and updates slope accordingly.
+    """
+
+    # Determine whether to flip meridian
+    f0 = sgn_column(matrix, 0)
+
+    # Determine whether to flip longitude
+    if manifold_orientable:
+        # We conform the matrix such that the first non-zero entry in the
+        # first column and the determinant are always positive
+        f1 = f0 * isomorphism_orientation
+    else:
+        # We conform the matrix such that the first non-zero entry in each
+        # column is always positive
+        f1 = sgn_column(matrix, 1)
+
+    flips = [ f0, f1 ]
+
+    for col, flip in enumerate(flips):
+        for row in range(2):
+            matrix[row,col] *= flip
+        slope[col] *= flip
+
+def pack_matrices(matrices):
+    """
+    Multiplies the columns of each matrix by the entries in flips and
+    packs all the matrices into one array, column-major.
+    """
+
+    return  [ matrix[row,col]
+              for matrix in matrices
+              for col in range(2)
+              for row in range(2) ]
+
+
+def supress_minus_zero(x):
+    if x == 0:
+        return 0
+    else:
+        return x
+
+
+def is_trivial_perm(perm):
+    return all(i == p for i, p in enumerate(perm))
+
+
+def key_prefer_pos(x):
+    """
+    Intended for key argument to min.
+
+    Prefers positive and then absolute value.
+    """
+    
+    if x >= 0:
+        return (0,  x)
+    else:
+        return (1, -x)
+
+def key_slope(slope):
+    """
+    Intended for key argument to min.
+
+    Prefers positive and small denominator.
+    """
+
+    slope_m, slope_l = slope
+    return (key_prefer_pos(slope_l), key_prefer_pos(slope_m))
+
+def key_decoration_info(info):
+    encoded, slopes = info
+    return [encoded, [key_slope(slope) for slope in slopes]]
+
+def normalized_slope(slope):
+    """
+    Returns slope or -slope preferring positive and small denominator.
+
+    Equivalent to min([slope, -slope], key=key_slope)
+    """
+
+    slope_m, slope_l = slope
+    if slope_l < 0:
+        return -slope
+    if slope_l > 0:
+        return slope
+    if slope_m < 0:
+        return -slope
+    return slope
+
+def candidate_decoration_info(
+        isomorphism,
+        slopes,
+        manifold_orientable,
+        ignore_cusp_ordering,
+        ignore_curves,
+        ignore_curve_orientations,
+        ignore_filling_orientations,
+        ignore_orientation):
+
+    matrices = isomorphism.cusp_maps()
+    isomorphism_orientation = matrices[0].det()
+
+    # Do not consider orientation-reversing isomorphisms if
+    # ignore_orientation isn't specified.
+    if manifold_orientable and not ignore_orientation:
+        if isomorphism_orientation < 0:
+            return None
+
+    # Make a copy as vectors so that we can modify in place and
+    # apply matrices.
+    # Outside of SnapPy, SimpleVector is not making a copy of the
+    # tuple and we need a list to modify things in place.
+    slopes = [ make_vector([slope_m, slope_l])
+               for slope_m, slope_l in slopes ]
+
+    # Permutation of cusps
+    perm = inverse_perm(isomorphism.cusp_images())
+
+    if ignore_cusp_ordering:
+        # If we do not include the permutation in the encoding,
+        # we need to apply it to the matrices
+        matrices = [ matrices[i] for i in perm ]
+        slopes = [ slopes[i] for i in perm ]
+
+    if ignore_curves:
+        slopes = [ matrix * slope
+                   for matrix, slope in zip(matrices, slopes) ]
+    else:
+        if ignore_curve_orientations:
+            for matrix, slope in zip(matrices, slopes):
+                apply_peripheral_curve_flips(
+                    matrix, slope, manifold_orientable, isomorphism_orientation)
+
+    encoded = ''
+
+    if not ignore_cusp_ordering:
+        #
+        # Force decoration to fix a very subtle bug!
+        #
+        # Recall that we call N.set_peripheral_curves('combinatorial')
+        # in decorated_isosig below if either ignore_cusp_ordering or
+        # ignore_curves is False.
+        #
+        # On the decoding site, we thus also need to call
+        # manifold.set_peripheral_curves('combinatorial') in those cases.
+        # This happens in set_peripheral_from_decoration and thus only
+        # if the isosig returned here has a decoration, that is "encoded"
+        # is not empty.
+        #
+        # Assume that ignore_cusp_ordering is False and ignore_curves is
+        # True and the permutation happens to be the identity.
+        #
+        # We need to make sure not to have an empty "encoded" in this case.
+        force_decoration = ignore_curves
+        
+        if force_decoration or not is_trivial_perm(perm):
+            # Encode permutation
+            encoded += encode_integer_list(perm)
+
+    if not ignore_curves:
+        # Encode the matrices
+        encoded += encode_integer_list(pack_matrices(matrices))
+
+    if ignore_filling_orientations:
+        slopes = [ normalized_slope(slope)
+                   for slope in slopes ]
+
+    return encoded, slopes
+
+# main two functions
+
+def decorated_isosig(manifold, triangulation_class,
+                     ignore_cusp_ordering=False,
+                     ignore_curves=False,
+                     ignore_curve_orientations=False,
+                     ignore_filling_orientations=False,
+                     ignore_orientation=True):
+
+    isosig = manifold._undecorated_triangulation_isosig(
+        ignore_orientation=ignore_orientation)
+
+    # Do not decorate if no cusps
+    if manifold.num_cusps() == 0:
+        return isosig
+
+    N = triangulation_class(isosig, remove_finite_vertices=False)
+    if not (ignore_cusp_ordering and ignore_curves):
+        # Note that data_to_triangulation determines the peripheral
+        # curves before orienting the manifold.
+        # Thus, we get different peripheral curves when calling
+        # N.set_peripheral_curves.
+        # For backwards compatibility (see set_peripheral_from_decoration),
+        # we need to keep calling N.set_peripheral_curves here unless there is
+        # no decoration.
+        N.set_peripheral_curves('combinatorial')
+
+    manifold_orientable = manifold.is_orientable()
+    slopes = manifold.cusp_info('filling')
+
+    # Try all combinatorial isomorphisms and pick
+    # lexicographically smallest info.
+    encoded, slopes = min(
+        (info
+         for isomorphism in manifold.isomorphisms_to(N)
+         if (
+             info := candidate_decoration_info(
+                 isomorphism,
+                 slopes,
+                 manifold_orientable=manifold_orientable,
+                 ignore_cusp_ordering=ignore_cusp_ordering,
+                 ignore_curves=ignore_curves,
+                 ignore_curve_orientations=ignore_curve_orientations,
+                 ignore_filling_orientations=ignore_filling_orientations,
+                 ignore_orientation=ignore_orientation)
+            ) is not None),
+        key = key_decoration_info)
+
+    ans = isosig
+
+    if encoded:
+        ans += separator + encoded
+
+    if not all(manifold.cusp_info('complete?')):
+        for slope_m, slope_l in slopes:
+            ans += '(%g,%g)' % (supress_minus_zero(slope_m),
+                                supress_minus_zero(slope_l))
+
+    return ans
+
+def set_peripheral_from_decoration(manifold, decoration):
+    """
+    The manifold is assumed to already have a triangulation created
+    from the "bare" isosig.
+    """
+    dec = decode_integer_list(decoration)
+    manifold.set_peripheral_curves('combinatorial')
+    n = manifold.num_cusps()
+    k = len(dec)
+
+    if k not in [ n, 4 * n, 5 * n]:
+        raise ValueError("Decoration has unexpected length.")
+
+    if k == n or k == 5 * n:
+        if k == n:
+            manifold._reindex_cusps(dec)
+        else:
+            manifold._reindex_cusps(dec[:n])
+    if k == 4 * n or k == 5 * n:
+        if k == 4 * n:
+            cobs = as_two_by_two_matrices(dec)
+        else:
+            cobs = as_two_by_two_matrices(dec[n:])
+        if det(cobs[0]) < 0 and manifold.is_orientable():
+            manifold.reverse_orientation()
+            cobs = [[(-a, b), (-c, d)] for [(a, b), (c,d)] in cobs]
+        manifold.set_peripheral_curves(cobs)
+
+# Testing code
+
+
+def is_identity(A):
+    return A[0, 0] == A[1, 1] == 1 and A[1, 0] == A[0, 1] == 0
+
+
+def preserves_peripheral_curves(h):
+    perm = h.cusp_images()
+    each_cusp = all(is_identity(A) for A in h.cusp_maps())
+    return perm == sorted(perm) and each_cusp
+
+
+def same_peripheral_curves(M, N):
+    for h in M.isomorphisms_to(N):
+        if preserves_peripheral_curves(h):
+            return True
+    return False
+
+
+asymmetric = ['v3372', 't10397', 't10448', 't11289', 't11581',
+              't11780', 't11824', 't12685', 'o9_34328', 'o9_35609', 'o9_35746',
+              'o9_36591', 'o9_37290', 'o9_37552', 'o9_38147', 'o9_38375',
+              'o9_38845', 'o9_39220', 'o9_41039', 'o9_41063', 'o9_41329',
+              'o9_43248']
+
+
+def main_test():
+    import snappy
+    censuses = [snappy.OrientableClosedCensus[:100],
+                snappy.OrientableCuspedCensus(filter='tets<7'),
+                snappy.NonorientableClosedCensus,
+                snappy.NonorientableCuspedCensus,
+                snappy.CensusKnots(),
+                snappy.HTLinkExteriors(filter='cusps>3 and volume<14'),
+                [snappy.Manifold(name) for name in asymmetric]]
+    tests = 0
+    for census in censuses:
+        for M in census:
+            isosig = decorated_isosig(M, snappy.Triangulation)
+            N = snappy.Triangulation(isosig)
+            assert same_peripheral_curves(M, N), M
+            assert isosig == decorated_isosig(N, snappy.Triangulation), M
+            assert M.homology() == N.homology()
+            tests += 1
+    print('Tested decorated isosig encode/decode on %d triangulations' % tests)
+
+
+def test_integer_list_encoder(tries=1000, length=100, max_entry=2**90):
+    import random
+    tests = 0
+    for i in range(tries):
+        entries = [random.randrange(-max_entry, max_entry) for i in range(length)]
+        entries += [random.randrange(-15, 16) for i in range(length)]
+        random.shuffle(entries)
+        assert decode_integer_list(encode_integer_list(entries)) == entries
+        tests += 1
+    print('Tested encode/decode on %d lists of integers' % tests)
+
+
+def test_link_invariant():
+    import snappy
+
+    # DT codes of the same link but with different orientations of the
+    # components
+
+    dt_codes = [
+        [(-14, -46, -40, -28, -60, -70), (-32, -34, -38, -4, -52, -50, -48), (-44, -42, -64, -2, -16, -58), (-56, -54, -8), (-36, -26, -24, -22, -20, -6, -18), (-10, -66), (-72, -30, -62), (-12, -68)],
+        [(-14, -46, -40, -28, -60, -70), (-36, -34, -30, -4, -52, -50, -48), (-42, -44, -58, -16, -2, -64), (-56, -54, -8), (-32, -26, -24, -22, -20, -6, -18), (-10, -66), (-72, -38, -62), (-12, -68)],
+        [(14, 70, 64, 50, 36, 24), (18, 2), (26, 16, 72), (46, 44, 22, 6, 48, 54), (52, 62, 60, 58, 56, 12, 34), (68, 66, 32, 10, 42, 40, 38), (28, 30, 8), (20, 4)],
+        [(-14, -46, -40, -28, -60, -70), (-32, -34, -38, -4, -52, -50, -48), (-44, -42, -64, -2, -16, -58), (-56, -54, -8), (-36, -26, -24, -22, -20, -6, -18), (-10, -68), (-30, -72, -62), (-12, -66)],
+        [(14, 70, 64, 50, 36, 24), (2, 18), (34, 16, 72), (42, 40, 54, 38, 6, 22), (62, 52, 26, 12, 56, 58, 60), (68, 66, 28, 10, 44, 46, 48), (32, 30, 8), (20, 4)],
+        [(-14, -46, -40, -28, -60, -70), (-34, -36, -58, -56, -54, -4, -30), (-42, -44, -48, -26, -2, -64), (-50, -52, -8), (-16, -32, -24, -6, -22, -20, -18), (-68, -10), (-38, -72, -62), (-66, -12)],
+        [(-14, -46, -40, -28, -60, -70), (-34, -36, -58, -56, -54, -4, -30), (-42, -44, -48, -26, -2, -64), (-50, -52, -8), (-16, -32, -24, -6, -22, -20, -18), (-10, -66), (-72, -38, -62), (-68, -12)],
+        [(14, 70, 64, 50, 36, 24), (2, 18), (16, 34, 72), (42, 40, 54, 38, 6, 20), (62, 52, 26, 12, 56, 58, 60), (68, 66, 28, 10, 44, 46, 48), (32, 30, 8), (4, 22)],
+        [(-14, -46, -40, -28, -60, -70), (-32, -34, -38, -4, -52, -50, -48), (-44, -42, -64, -2, -16, -58), (-56, -54, -8), (-36, -26, -24, -22, -20, -6, -18), (-66, -10), (-72, -30, -62), (-68, -12)]
+        ]
+
+    # Get complement for each dt_code and complement with opposite orientation
+    # for each dt_code
+    mfds = [ snappy.Manifold('DT%s' % dt_code) for dt_code in (dt_codes + dt_codes) ]
+    for mfd in mfds[:len(dt_codes)]:
+        mfd.reverse_orientation()
+
+    isometry_signatures = [ mfd.isometry_signature(of_link=True)
+                            for mfd in mfds ]
+
+    # All the links only differ in orientation of complement or components,
+    # should get the same isometry_signature
+    assert len(set(isometry_signatures)) == 1
+
+    M = snappy.Manifold(isometry_signatures[0])
+    N = snappy.Manifold(M.isometry_signature(of_link=True))
+
+    # Instantiating a manifold from its decorated isometry_signature should
+    # eventually yield to a fixed point
+    assert same_peripheral_curves(M, N)
+
+    # More sanity checks
+    assert isometry_signatures[0] == M.isometry_signature(of_link=True)
+    assert isometry_signatures[0] == N.isometry_signature(of_link=True)
+
+    for mfd in mfds:
+        assert mfd.is_isometric_to(M, True)[0].extends_to_link()
+        assert mfd.is_isometric_to(N, True)[0].extends_to_link()
+
+    print("Tested that decorated isometry_signature is a link invariant")
+
+
+def helper_are_isometric(M, N):
+    for i in range(100):
+        try:
+            if M.is_isometric_to(N):
+                return
+        except:
+            pass
+        M.randomize()
+        N.randomize()
+
+    raise Exception("Could not find isometry")
+
+
+def helper_test_by_dehn_filling(M):
+    from snappy import Manifold
+
+    M_filled = M.filled_triangulation()
+
+    for ignore_cusp_ordering in [ False, True]:
+        for ignore_curve_orientations in [ False, True]:
+            isosig = M.triangulation_isosig(
+                    decorated=True,
+                    ignore_cusp_ordering=ignore_cusp_ordering,
+                    ignore_curve_orientations=ignore_curve_orientations)
+            N = Manifold(isosig)
+            N_filled = N.filled_triangulation()
+
+            helper_are_isometric(M, N)
+
+
+def test_by_dehn_filling():
+    import random
+
+    from snappy import OrientableCuspedCensus
+
+    count = 0
+
+    for M in OrientableCuspedCensus(cusps=3):
+        for i in range(20):
+            unfilled = random.randint(0, 2)
+            for c in range(3):
+                if c != unfilled:
+                    fillings = [(1,0), (0,1), (11,12), (-13,16),
+                                (9,-11), (8, 9), (1,7), (13, 14),
+                                (14,-15), (17, -18)]
+                    M.dehn_fill(fillings[random.randint(0, len(fillings) - 1)],
+                                c)
+
+            if 'positive' in M.solution_type():
+                count += 1
+                helper_test_by_dehn_filling(M)
+
+    print("Tested %d randomly Dehn filled manifolds" % count)
+
+def test_slope_transformations():
+    """
+    Tests that slopes are transformed so that the filled
+    manifold is the same.
+    """
+
+    import snappy
+    M = snappy.ManifoldHP("L14n63023(-5,1)(5,1)(10,1)")
+    oriented_isosig = M.triangulation_isosig(
+        decorated=False, ignore_orientation=False)
+    isosig = M.triangulation_isosig(
+        decorated=False)
+    Mop = M.copy()
+    Mop.reverse_orientation()
+    reverse_oriented_isosig = Mop.triangulation_isosig(
+        decorated=False, ignore_orientation=False)
+
+    if oriented_isosig != 'vLLvvLLMALQQzQQceillmnppqrlmrqtruututiivimllaelaqxrvdoxqltt':
+        raise AssertionError()
+    if isosig != 'vLLvLLPwPQLAMPQcefikkmnplkopqrsttutuuiixvimqlippawidlabavth':
+        raise AssertionError()
+
+    # The canonical orientation (used to compute the unoriented isosig)
+    # is the reverse of the actual orientation:
+    if reverse_oriented_isosig != isosig:
+        raise AssertionError()
+    if oriented_isosig == isosig:
+        raise ValueError()
+
+    isom_sig_pos = M.isometry_signature(ignore_orientation = False)
+    if isom_sig_pos != 'KLALvLwLLwMQLQPAMzMzMPzMPcbbeghnklntpqpqvrswtuvxyzABCDEFEGHIJJhhkofnaocnmrlsiaowxfcsaxhxhxhxhjhhhhs':
+        raise AssertionError()
+    isom_sig_neg = Mop.isometry_signature(ignore_orientation = False)
+    if isom_sig_neg != 'KLAMvMvvAwLvQPPPQMPzMPzMPcbbdegilopoouqtryvuxvwxzzBACDEFEGHIJJhhkhhohahrscaagwxkkgbvwpuxwqxqxwxxxxr':
+        raise AssertionError()
+
+    # It is not just the triangulation that is chiral, the manifold itself is:
+    if isom_sig_pos == isom_sig_neg:
+        raise ValueError()
+
+    # So we expect the oriented isometry signature to flip when neither the isomorphism
+    # signature nor its decoration capture the orientation.
+    for ignore_cusp_ordering in [False, True]:
+        for ignore_curves in [False, True]:
+            for ignore_curve_orientations in [False, True]:
+                for ignore_filling_orientations in [False, True]:
+                    for ignore_orientation in [False, True]:
+                        isosig = M.triangulation_isosig(
+                            ignore_cusp_ordering = ignore_cusp_ordering,
+                            ignore_curves = ignore_curves,
+                            ignore_curve_orientations = ignore_curve_orientations,
+                            ignore_filling_orientations = ignore_filling_orientations,
+                            ignore_orientation = ignore_orientation)
+                        isom_sig = (
+                            snappy.ManifoldHP(isosig)
+                                 .isometry_signature(ignore_orientation = False))
+                        does_ignore_orientation = (
+                            ignore_orientation and
+                            (ignore_curve_orientations or ignore_curves))
+                        expected_isom_sig = (
+                            isom_sig_neg
+                            if does_ignore_orientation
+                            else isom_sig_pos)
+                        if isom_sig != expected_isom_sig:
+                            raise AssertionError()
+    print("Tested slope transformations")
+
+def run_tests():
+    test_integer_list_encoder()
+    main_test()
+    test_link_invariant()
+    test_by_dehn_filling()
+    test_slope_transformations()