Rename sequence

docs/functional.rst

@@ -57,8 +57,8 @@ Encodings
 
     multiset
     multibind
-    sequence
-    distinct_sequence
+    bundle_sequence
+    bind_sequence
     hash_table
     cross_product
     ngrams

docs/structures.rst

@@ -14,8 +14,8 @@ This module provides class-based implementations of HDC data structures.
     Memory
     Multiset
     HashTable
-    Sequence
-    DistinctSequence
+    BundleSequence
+    BindSequence
     Graph
     Tree
     FiniteStateAutomata

torchhd/functional.py

@@ -23,8 +23,8 @@
     "multiset",
     "multibind",
     "cross_product",
-    "sequence",
-    "distinct_sequence",
+    "bundle_sequence",
+    "bind_sequence",
     "ngrams",
     "hash_table",
     "map_range",
@@ -621,13 +621,13 @@ def hamming_similarity(input: Tensor, others: Tensor) -> LongTensor:
     return torch.sum(input == others, dim=-1, dtype=torch.long)
 
 
-def multiset(
-    input: Tensor,
-) -> Tensor:
+def multiset(input: Tensor) -> Tensor:
     r"""Multiset of input hypervectors.
 
     Bundles all the input hypervectors together.
 
+    Aliased as ``torchhd.functional.multibundle``.
+
     .. math::
 
         \bigoplus_{i=0}^{n-1} V_i
@@ -662,6 +662,9 @@ def multiset(
     return torch.sum(input, dim=-2, dtype=input.dtype)
 
 
+multibundle = multiset
+
+
 def multibind(input: Tensor) -> Tensor:
     r"""Binding of multiple hypervectors.
 
@@ -747,7 +750,7 @@ def ngrams(input: Tensor, n: int = 3) -> Tensor:
         \bigoplus_{i=0}^{m - n} \bigotimes_{j = 0}^{n - 1} \Pi^{n - j - 1}(V_{i + j})
 
     .. note::
-        For :math:`n=1` use :func:`~torchhd.functional.multiset` instead and for :math:`n=m` use :func:`~torchhd.functional.distinct_sequence` instead.
+        For :math:`n=1` use :func:`~torchhd.functional.multiset` instead and for :math:`n=m` use :func:`~torchhd.functional.bind_sequence` instead.
 
     Args:
         input (Tensor): The value hypervectors.
@@ -812,7 +815,7 @@ def hash_table(keys: Tensor, values: Tensor) -> Tensor:
     return multiset(bind(keys, values))
 
 
-def sequence(input: Tensor) -> Tensor:
+def bundle_sequence(input: Tensor) -> Tensor:
     r"""Bundling-based sequence.
 
     The first value is permuted :math:`n-1` times, the last value is not permuted.
@@ -837,7 +840,7 @@ def sequence(input: Tensor) -> Tensor:
                 [ 1.,  1.,  1.],
                 [-1., -1., -1.],
                 [ 1.,  1.,  1.]])
-        >>> functional.sequence(x)
+        >>> functional.bundle_sequence(x)
         tensor([-1.,  3.,  1.])
 
     """
@@ -851,7 +854,7 @@ def sequence(input: Tensor) -> Tensor:
     return multiset(permuted)
 
 
-def distinct_sequence(input: Tensor) -> Tensor:
+def bind_sequence(input: Tensor) -> Tensor:
     r"""Binding-based sequence.
 
     The first value is permuted :math:`n-1` times, the last value is not permuted.
@@ -876,7 +879,7 @@ def distinct_sequence(input: Tensor) -> Tensor:
                 [ 1., -1., -1.],
                 [ 1., -1., -1.],
                 [-1., -1., -1.]])
-        >>> functional.distinct_sequence(x)
+        >>> functional.bind_sequence(x)
         tensor([-1.,  1.,  1.])
 
     """

torchhd/structures.py

@@ -8,8 +8,8 @@
     "Memory",
     "Multiset",
     "HashTable",
-    "Sequence",
-    "DistinctSequence",
+    "BundleSequence",
+    "BindSequence",
     "Graph",
     "Tree",
     "FiniteStateAutomata",
@@ -134,7 +134,7 @@ def __delitem__(self, key: Tensor) -> None:
 
 
 class Multiset:
-    """Hypervector-based multiset data structure.
+    """Hypervector multiset data structure.
 
     Creates an empty multiset of dim dimensions or from an input tensor.
 
@@ -278,7 +278,7 @@ def from_tensor(cls, input: Tensor):
 
 
 class HashTable:
-    """Hypervector-based hash table data structure.
+    """Hypervector hash table data structure.
 
     Creates an empty hash table of dim dimensions or a hash table from an input tensor.
 
@@ -437,8 +437,8 @@ def from_tensors(cls, keys: Tensor, values: Tensor):
         return cls(value, size=keys.size(-2))
 
 
-class Sequence:
-    """Hypervector-based (bundling-based) sequence data structure
+class BundleSequence:
+    """Hypervector bundling-based sequence data structure
 
     Creates an empty sequence of dim dimensions or from an input tensor.
 
@@ -453,11 +453,11 @@ class Sequence:
 
     Examples::
 
-        >>> S = structures.Sequence(10000)
+        >>> S = structures.BundleSequence(10000)
 
         >>> letters = list(string.ascii_lowercase)
         >>> letters_hv = functional.random_hv(len(letters), 10000)
-        >>> S = structures.Sequence(letters_hv[0], size=1)
+        >>> S = structures.BundleSequence(letters_hv[0], size=1)
 
     """
 
@@ -559,21 +559,21 @@ def replace(self, index: int, old: Tensor, new: Tensor) -> None:
         rotated_new = functional.permute(new, shifts=self.size - index - 1)
         self.value = functional.bundle(self.value, rotated_new)
 
-    def concat(self, seq: "Sequence") -> "Sequence":
+    def concat(self, seq: "BundleSequence") -> "BundleSequence":
         """Concatenates the current sequence with the given one.
 
         Args:
             seq (Sequence): Sequence to be concatenated with the current one.
 
         Examples::
 
-            >>> S1 = structures.Sequence(dimensions=10000)
+            >>> S1 = structures.BundleSequence(dimensions=10000)
             >>> S2 = S.concat(S1)
 
         """
         value = functional.permute(self.value, shifts=len(seq))
         value = functional.bundle(value, seq.value)
-        return Sequence(value, size=len(self) + len(seq))
+        return BundleSequence(value, size=len(self) + len(seq))
 
     def __getitem__(self, index: int) -> Tensor:
         """Gets the approximate value from given index.
@@ -615,37 +615,37 @@ def clear(self) -> None:
     def from_tensor(cls, input: Tensor):
         """Creates a sequence from hypervectors.
 
-        See: :func:`~torchhd.functional.sequence`.
+        See: :func:`~torchhd.functional.bundle_sequence`.
 
         Args:
-            input (Tensor): Tensor representing a sequence.
+            input (Tensor): Tensor containing hypervectors that form the sequence.
 
         Examples::
             >>> letters_hv = functional.random_hv(len(letters), 10000)
-            >>> S = structures.Sequence.from_tensor(letters_hv)
+            >>> S = structures.BundleSequence.from_tensor(letters_hv)
 
         """
-        value = functional.sequence(input)
+        value = functional.bundle_sequence(input)
         return cls(value, size=input.size(-2))
 
 
-class DistinctSequence:
-    """Hypervector-based distinct (binding-based) sequence data structure.
+class BindSequence:
+    """Hypervector binding-based sequence data structure.
 
     Creates an empty sequence of dim dimensions or from an input tensor.
 
     Args:
-        dimensions (int): number of dimensions of the distinct sequence.
+        dimensions (int): number of dimensions of the sequence.
         dtype (``torch.dtype``, optional): the desired data type of returned tensor. Default: if ``None``, uses a global default (see ``torch.set_default_tensor_type()``).
         device (``torch.device``, optional):  the desired device of returned tensor. Default: if ``None``, uses the current device for the default tensor type (see torch.set_default_tensor_type()). ``device`` will be the CPU for CPU tensor types and the current CUDA device for CUDA tensor types.
 
     Args:
-        input (Tensor): tensor representing a distinct sequence.
-        size (int, optional): the length of the distinct sequence provided as input. Default: ``0``.
+        input (Tensor): tensor representing a binding-based sequence.
+        size (int, optional): the length of the sequence provided as input. Default: ``0``.
 
     Examples::
 
-        >>> DS = structures.DistinctSequence(10000)
+        >>> DS = structures.BindSequence(10000)
     """
 
     @overload
@@ -739,7 +739,7 @@ def replace(self, index: int, old: Tensor, new: Tensor) -> None:
 
         Examples::
 
-            >>> DS1 = structures.DistinctSequence(dimensions=10000)
+            >>> DS1 = structures.BindSequence(dimensions=10000)
             >>> DS.concat(DS1)
 
         """
@@ -775,17 +775,17 @@ def clear(self) -> None:
     def from_tensor(cls, input: Tensor):
         """Creates a sequence from tensor.
 
-        See: :func:`~torchhd.functional.distinct_sequence`.
+        See: :func:`~torchhd.functional.bind_sequence`.
 
         Args:
-            input (Tensor): Tensor representing a sequence.
+            input (Tensor): Tensor containing hypervectors that form the sequence.
 
         Examples::
             >>> letters_hv = functional.random_hv(len(letters), 10000)
-            >>> DS = structures.DistinctSequence.from_tensor(letters_hv)
+            >>> DS = structures.BindSequence.from_tensor(letters_hv)
 
         """
-        value = functional.distinct_sequence(input)
+        value = functional.bind_sequence(input)
         return cls(value, size=input.size(-2))
 
 

torchhd/tests/structures/test_distinct_sequence.py

@@ -8,17 +8,17 @@
 letters = list(string.ascii_lowercase)
 
 
-class TestDistinctSequence:
+class TestBindSequence:
     def test_creation_dim(self):
-        S = structures.DistinctSequence(10000)
+        S = structures.BindSequence(10000)
         assert torch.equal(S.value, torch.ones(10000))
 
     def test_creation_tensor(self):
         generator = torch.Generator()
         generator.manual_seed(seed)
         hv = functional.random_hv(len(letters), 10000, generator=generator)
 
-        S = structures.DistinctSequence(hv[0])
+        S = structures.BindSequence(hv[0])
         assert torch.equal(S.value, hv[0])
 
     def test_generator(self):
@@ -36,23 +36,23 @@ def test_append(self):
         generator = torch.Generator()
         generator.manual_seed(seed)
         hv = functional.random_hv(len(letters), 10000, generator=generator)
-        S = structures.DistinctSequence(10000)
+        S = structures.BindSequence(10000)
         S.append(hv[0])
         assert functional.cosine_similarity(S.value, hv)[0] > 0.5
 
     def test_appendleft(self):
         generator = torch.Generator()
         generator.manual_seed(seed)
         hv = functional.random_hv(len(letters), 10000, generator=generator)
-        S = structures.DistinctSequence(10000)
+        S = structures.BindSequence(10000)
         S.appendleft(hv[0])
         assert functional.cosine_similarity(S.value, hv)[0] > 0.5
 
     def test_pop(self):
         generator = torch.Generator()
         generator.manual_seed(seed)
         hv = functional.random_hv(len(letters), 10000, generator=generator)
-        S = structures.DistinctSequence(10000)
+        S = structures.BindSequence(10000)
         S.append(hv[0])
         S.append(hv[1])
         S.pop(hv[1])
@@ -68,7 +68,7 @@ def test_popleft(self):
         generator = torch.Generator()
         generator.manual_seed(seed)
         hv = functional.random_hv(len(letters), 10000, generator=generator)
-        S = structures.DistinctSequence(10000)
+        S = structures.BindSequence(10000)
         S.appendleft(hv[0])
         S.appendleft(hv[1])
         S.popleft(hv[1])
@@ -84,7 +84,7 @@ def test_replace(self):
         generator = torch.Generator()
         generator.manual_seed(seed)
         hv = functional.random_hv(len(letters), 10000, generator=generator)
-        S = structures.DistinctSequence(10000)
+        S = structures.BindSequence(10000)
         S.append(hv[0])
         assert functional.cosine_similarity(S.value, hv)[0] > 0.5
         S.replace(0, hv[0], hv[1])
@@ -94,7 +94,7 @@ def test_length(self):
         generator = torch.Generator()
         generator.manual_seed(seed)
         hv = functional.random_hv(len(letters), 10000, generator=generator)
-        S = structures.DistinctSequence(10000)
+        S = structures.BindSequence(10000)
         S.append(hv[0])
         S.append(hv[0])
         S.append(hv[0])
@@ -113,7 +113,7 @@ def test_clear(self):
         generator = torch.Generator()
         generator.manual_seed(seed)
         hv = functional.random_hv(len(letters), 10000, generator=generator)
-        S = structures.DistinctSequence(10000)
+        S = structures.BindSequence(10000)
         S.append(hv[0])
         S.append(hv[0])
         S.append(hv[0])