summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorNao Pross <np@0hm.ch>2024-04-15 15:15:50 +0200
committerNao Pross <np@0hm.ch>2024-04-15 15:19:16 +0200
commit67c7570edadbbb0fc5f3d35ee61b61c5507dae0d (patch)
tree99259dab18897f9f0184c7d0102080f5a073cc1a
parentMore review comments in expression mixins (diff)
downloadpolymatrix-67c7570edadbbb0fc5f3d35ee61b61c5507dae0d.tar.gz
polymatrix-67c7570edadbbb0fc5f3d35ee61b61c5507dae0d.zip
Introduce index types and new PolyMatrix API
-rw-r--r--polymatrix/polymatrix/impl.py6
-rw-r--r--polymatrix/polymatrix/init.py6
-rw-r--r--polymatrix/polymatrix/mixins.py123
-rw-r--r--polymatrix/polymatrix/typing.py74
-rw-r--r--polymatrix/polymatrix/utils/mergemonomialindices.py9
-rw-r--r--polymatrix/polymatrix/utils/multiplypolynomial.py8
-rw-r--r--polymatrix/polymatrix/utils/sortmonomialindices.py4
7 files changed, 170 insertions, 60 deletions
diff --git a/polymatrix/polymatrix/impl.py b/polymatrix/polymatrix/impl.py
index 27b6260..152d5fa 100644
--- a/polymatrix/polymatrix/impl.py
+++ b/polymatrix/polymatrix/impl.py
@@ -2,16 +2,16 @@ import dataclassabc
from polymatrix.polymatrix.abc import PolyMatrix
from polymatrix.polymatrix.mixins import BroadcastPolyMatrixMixin
-from polymatrix.polymatrix.typing import PolynomialData
+from polymatrix.polymatrix.typing import PolyMatrixDict, PolyDict
@dataclassabc.dataclassabc(frozen=True)
class PolyMatrixImpl(PolyMatrix):
- data: dict[tuple[int, int], PolynomialData] # FIXME: use polymatrix.typing
+ data: PolyMatrixDict
shape: tuple[int, int]
@dataclassabc.dataclassabc(frozen=True)
class BroadcastPolyMatrixImpl(BroadcastPolyMatrixMixin):
- data: tuple[tuple[int], float] # FIXME: use polymatrix.typing
+ data: PolyDict
shape: tuple[int, int]
diff --git a/polymatrix/polymatrix/init.py b/polymatrix/polymatrix/init.py
index c4912c7..704ebe5 100644
--- a/polymatrix/polymatrix/init.py
+++ b/polymatrix/polymatrix/init.py
@@ -1,10 +1,10 @@
from polymatrix.polymatrix.impl import BroadcastPolyMatrixImpl, PolyMatrixImpl
-from polymatrix.polymatrix.typing import PolynomialData
+from polymatrix.polymatrix.typing import PolyMatrixDict, PolyDict
# FIXME: use polymatrix.typing
def init_poly_matrix(
- data: dict[tuple[int, int], PolynomialData],
+ data: PolyMatrixDict,
shape: tuple[int, int],
):
return PolyMatrixImpl(
@@ -16,7 +16,7 @@ def init_poly_matrix(
# NP: cosider renaming to scalar cf. comment on broadcast_poly_matrix and BroadcastPolymatrix
# FIXME: use polymatrix.typing
def init_broadcast_poly_matrix(
- data: PolynomialData,
+ data: PolyDict,
shape: tuple[int, int],
):
return BroadcastPolyMatrixImpl(
diff --git a/polymatrix/polymatrix/mixins.py b/polymatrix/polymatrix/mixins.py
index d4381e7..ddd00fc 100644
--- a/polymatrix/polymatrix/mixins.py
+++ b/polymatrix/polymatrix/mixins.py
@@ -1,68 +1,111 @@
import abc
import typing
-from polymatrix.polymatrix.typing import PolynomialData
+from typing import Iterable, cast
+from typing_extensions import override
+
+from polymatrix.polymatrix.typing import PolyDict, PolyMatrixDict, MatrixIndex
+from polymatrix.utils.deprecation import deprecated
class PolyMatrixMixin(abc.ABC):
- # NP: is this still in use? cf. Filter operation, that breaks the
- # NP: assumption that shape can be computed in advance
- # FIXME: classmethod as properties are no longer allowed, change to abstractmethod
@property
- @abc.abstractclassmethod
+ @abc.abstractmethod
def shape(self) -> tuple[int, int]: ...
- # NP: the "data" is an iteration over the entries of the matrix
- # FIXME: rename to something meaningful like entries()
- # NP: the fact that it returns the inner dictionary directly leaks the abstraction,
- # NP: as discussed in in meeting #6 IMHO it is better to return indices and query like in mdpoly
- def gen_data(self) -> typing.Generator[tuple[tuple[int, int], PolynomialData], None, None]:
- for row in range(self.shape[0]):
- for col in range(self.shape[1]):
- polynomial = self.get_poly(row, col)
- if polynomial is None:
- continue
-
- yield (row, col), polynomial
-
-
- # NP: get a polynomial at a given entry
- # NP: why is this an abstract class method if it uses self?
- # FIXME: change to abstractmethod
- @abc.abstractclassmethod
- def get_poly(self, row: int, col: int) -> PolynomialData | None: ...
-
-
-# NP: As far as I can tell the dict is the only way you ever store a polymatrix
-# NP: (cf. PolynomialData), why separate interface for PolyMatrixAsDictMixin?
-# NP: Also applies to BroadcastPolyMatrixMixin class below.
+ @deprecated("Replaced by PolyMatrixMixin.entries()")
+ def gen_data(self) -> typing.Generator[tuple[tuple[int, int], PolyDict], None, None]:
+ yield from self.entries()
+
+ @abc.abstractmethod
+ @deprecated("Replaced by PolyMatrixMixin.at(row, col).")
+ def get_poly(self, row: int, col: int) -> PolyDict | None:
+ # NOTE: in the new API when there is nothing at an entry the at()
+ # function returns an empty dictionary. This is because emtpy
+ # dicionaries are falsy so you can do
+ #
+ # `if p.at(row, col):`
+ #
+ # like it was possible with get_poly. However this is better because
+ # id does not make use of a null (None) type.
+ return self.at(row, col) or None
+
+ # --- New API ---
+
+ @abc.abstractmethod
+ def at(self, row: int, col: int) -> PolyDict:
+ """ Return the polynomial at the entry (row, col).
+
+ .. code:: py
+ p.at(row, col) # if you have row and col
+ p.at(*entry) # if you have a MatrixIndex
+ """
+
+ def entries(self) -> Iterable[tuple[MatrixIndex, PolyDict]]:
+ """
+ Iterate over the non-zero entries of the polynomial matrix.
+
+ TODO: docstrings, use like this (`p` is a polymatrix)
+
+ .. code:: py
+ for entry, poly in p.entries():
+ for monomial, coeff in poly.terms():
+ # do something with coefficients
+ """
+ nrows, ncols = self.shape
+ for row in range(nrows):
+ for col in range(ncols):
+ if poly := self.at(row, col):
+ yield MatrixIndex(row, col), poly
+
+
class PolyMatrixAsDictMixin(
PolyMatrixMixin,
abc.ABC,
):
+ """
+ Polynomial matrix, stored as a dictionary.
+ """
+
@property
@abc.abstractmethod
- def data(self) -> dict[tuple[int, int], PolynomialData]: ...
+ def data(self) -> PolyMatrixDict:
+ """" Get the dictionary. """
- # overwrites the abstract method of `PolyMatrixMixin`
- # NP: Returning none is not great design, from math perspective it would
- # NP: make more sense to return zero
- def get_poly(self, row: int, col: int) -> PolynomialData | None:
+
+ @override
+ def get_poly(self, row: int, col: int) -> PolyDict | None:
if (row, col) in self.data:
- return self.data[row, col]
+ return self.data[cast(MatrixIndex, (row, col))]
+ return None
+
+ # --- New API ---
+
+ @override
+ def at(self, row: int, col: int) -> PolyDict:
+ """ See :py:meth:`PolyMatrixMixin.at`. """
+ return self.data.get(MatrixIndex(row, col)) or PolyDict.empty()
-# NP: What does broadcast mean here? Is this a scalar multivariate polynomial
-# NP: that does not require the (row, col) indexing?
-# NP: If yes, consider new name ScalarPolynomial, ScalarPolyMatrix?
class BroadcastPolyMatrixMixin(
PolyMatrixMixin,
abc.ABC,
):
+ """
+ TODO: docstring, similar to numpy broadcasting.
+ https://numpy.org/doc/stable/user/basics.broadcasting.html
+ """
@property
@abc.abstractmethod
- def data(self) -> PolynomialData: ...
+ def data(self) -> PolyDict: ...
# overwrites the abstract method of `PolyMatrixMixin`
- def get_poly(self, col: int, row: int) -> PolynomialData | None:
+ def get_poly(self, col: int, row: int) -> PolyDict | None:
+ return self.data or None
+
+ # --- New API ---
+
+ @override
+ def at(self, row: int, col: int) -> PolyDict:
+ """ See :py:meth:`PolyMatrixMixin.at`. """
return self.data
diff --git a/polymatrix/polymatrix/typing.py b/polymatrix/polymatrix/typing.py
index e304e5c..43d7210 100644
--- a/polymatrix/polymatrix/typing.py
+++ b/polymatrix/polymatrix/typing.py
@@ -1,9 +1,73 @@
-# monomial x1**2 x2
-# with indices {x1: 0, x2: 1}
-# is represented as ((0, 2), (1, 1))
+from __future__ import annotations
+from typing import NamedTuple, Iterable
-# NP: consider renaming Data to Index (data does not mean anything to me)
-# NP: introduce index classes?
+# TODO: remove these types, they are here for backward compatiblity
MonomialData = tuple[tuple[int, int], ...]
PolynomialData = dict[MonomialData, int | float]
PolynomialMatrixData = dict[tuple[int, int], PolynomialData]
+
+
+class VariableIndex(NamedTuple):
+ """
+ Index for a variable raised to an integral power.
+ """
+ variable: int # index in ExpressionState object
+ power: int
+
+
+ def __lt__(self, other):
+ """ Variables indices can be sorted with respect to variable index. """
+ return self.variable < other.variable
+
+
+class MonomialIndex(tuple[VariableIndex]):
+ """
+ Index for a monomial, i.e. a product of `VariableIndex`.
+ """
+
+ @staticmethod
+ def empty() -> MonomialIndex:
+ return MonomialIndex(tuple())
+
+
+
+class PolyDict(dict[MonomialIndex, int | float]):
+ """ Polynomial, stored as a dictionary. """
+
+ # NOTE: this class directly inherits from dict instead of
+ # collections.UserDict, hence it is not possible to override any of the
+ # __dunder__ methods. If you like to change behaviour of one of these
+ # methods change to inherit from UserDict, this could incur a slight
+ # performance cost, because dict is straight CPython and thus faster
+
+ @staticmethod
+ def empty() -> PolyDict:
+ return PolyDict({})
+
+
+ def terms(self) -> Iterable[tuple[MonomialIndex, int | float]]:
+ """ Iterate over terms with a non-zero coefficient. """
+ # This is an alias for readability
+ yield from self.items()
+
+ def monomials(self) -> Iterable[MonomialIndex]:
+ """ Get monomials that compose the polynomial. """
+ # This is an alias for readability
+ yield from self.keys()
+
+ def coefficients(self) -> Iterable[int | float]:
+ """ Get the coefficients of the polynomial. """
+ # This is an alias for readability
+ yield from self.values()
+
+
+class MatrixIndex(NamedTuple):
+ """ Index to represent an entry of a matrix. """
+ row: int
+ col: int
+
+
+class PolyMatrixDict(dict[MatrixIndex, PolyDict]):
+ """ Matrix whose entries are polynomials, stored as a dictionary. """
+
+ # NOTE: Same as NOTE in PolyDict.
diff --git a/polymatrix/polymatrix/utils/mergemonomialindices.py b/polymatrix/polymatrix/utils/mergemonomialindices.py
index ec8f309..9ff2a63 100644
--- a/polymatrix/polymatrix/utils/mergemonomialindices.py
+++ b/polymatrix/polymatrix/utils/mergemonomialindices.py
@@ -1,12 +1,12 @@
-from polymatrix.polymatrix.typing import MonomialData
+from polymatrix.polymatrix.typing import MonomialIndex
from polymatrix.polymatrix.utils.sortmonomialindices import sort_monomial_indices
# NP: this is a werid case of the product
# NP: merge is not intutive IMHO, find better name
def merge_monomial_indices(
- monomials: tuple[MonomialData, ...],
-) -> MonomialData:
+ monomials: tuple[MonomialIndex, ...],
+) -> MonomialIndex:
"""
(x1**2 x2, x2**2) -> x1**2 x2**3
@@ -17,9 +17,8 @@ def merge_monomial_indices(
((1, 2),) # x2**2
) -> ((0, 2), (1, 3)) # x1**1 x2**3
"""
-
if len(monomials) == 0:
- return tuple()
+ return MonomialIndex.empty()
elif len(monomials) == 1:
return monomials[0]
diff --git a/polymatrix/polymatrix/utils/multiplypolynomial.py b/polymatrix/polymatrix/utils/multiplypolynomial.py
index 4b54813..5b7a62c 100644
--- a/polymatrix/polymatrix/utils/multiplypolynomial.py
+++ b/polymatrix/polymatrix/utils/multiplypolynomial.py
@@ -2,14 +2,14 @@ import itertools
import math
from polymatrix.polymatrix.utils.mergemonomialindices import merge_monomial_indices
-from polymatrix.polymatrix.typing import PolynomialData
+from polymatrix.polymatrix.typing import PolyDict
# NP: compute index of product of polynomials
def multiply_polynomial(
- left: PolynomialData,
- right: PolynomialData,
- result: PolynomialData,
+ left: PolyDict,
+ right: PolyDict,
+ result: PolyDict,
) -> None:
"""
Multiplies two polynomials `left` and `right` and adds the result to the mutable polynomial `result`.
diff --git a/polymatrix/polymatrix/utils/sortmonomialindices.py b/polymatrix/polymatrix/utils/sortmonomialindices.py
index ff4ce59..69d121a 100644
--- a/polymatrix/polymatrix/utils/sortmonomialindices.py
+++ b/polymatrix/polymatrix/utils/sortmonomialindices.py
@@ -1,7 +1,11 @@
from polymatrix.polymatrix.typing import MonomialData
+from polymatrix.utils.deprecation import deprecated
+
# NP: sort mononial indices with respect to variable index in state object
+
+@deprecated("With new index types you can use sorted() on MonomialIndex")
def sort_monomial_indices(
monomial: MonomialData,
) -> MonomialData: