Class NDMatrix
- Namespace
- NumSharp
- Assembly
- NumSharp.dll
NumPy's numpy.matrix: a two-dimensional array whose * is the matrix product, whose rows and
columns stay two-dimensional when indexed, and which adds T, H,
I, A and A1. Create one with matrix(NDArray, DType, bool),
asmatrix(NDArray, DType) or bmat(NDArray[][]). Use it to port NumPy code
written against np.matrix; for new code NumPy recommends plain arrays and np.matmul.
public class NDMatrix : NDArray, IIndex, ICloneable, IDisposable, IEnumerable
- Inheritance
-
NDMatrix
- Implements
- Inherited Members
- Extension Methods
Remarks
Members and operators follow the static type. The members and operators of this class are chosen
by the compiler when the expression is typed NDMatrix. The same object seen through an
NDArray reference (a cast, a variable typed NDArray) is a plain 2-D array to
the operators, the indexers and the NDArray members: * is elementwise there, an
integer index drops an axis and a.sum(0) has N elements.
The np.* functions follow the object. A public function that receives a matrix object
returns the class and the shape NumPy returns for a numpy.matrix operand: np.sum(m, 0) is a
1 x N matrix, np.negative(m), np.transpose(m) and np.linalg.inv(m) are matrices,
np.ravel(m) is a plain array, np.stack raises shape too large to be a matrix.. The
result is typed NDArray; cast it ((NDMatrix)np.sum(m, 0)) to reach the matrix
members, no copy is made. A function whose C# return type is a typed array
(isclose(NDArray, NDArray, double, double, bool)) or a struct of arrays (np.gradient, np.meshgrid) returns plain
arrays. The mechanisms are declared on the computations and woven into the functions (NumSharp.NDSubclass).
Always two-dimensional. Every matrix result goes through NumPy's finalize rule
(NumSharp.Shape.AsMatrix()): a 1-d result becomes a row (1, N), a 0-d result (1, 1), and a
result with more than two axes longer than 1 raises shape too large to be a matrix.. Two NumPy
quirks are kept: indexing with a new axis (m[Slice.NewAxis]) and an in-place
resize can leave a matrix object with another rank.
Scalars. Where NumPy returns a scalar, a member returns a 0-d NDArray when the result
is always a scalar (m[0, 0], m.sum()) and a 0-d NDMatrix when it depends on the
arguments (m["0, 0"]). A C# number converts implicitly to a 0-d NDMatrix, so
m * 2, m + 1.5 and m[mask] = 0 compile. A 0-d matrix is a scalar: * scales by
it, and so does every 0-d NDArray that a computation returned (m * m.sum(),
m * np.linalg.norm(a), m * a[0, 0]). The exception is NumPy's: an explicit 0-d ARRAY
(np.array(2), np.asarray(2); NumSharp.NDArray.IsExplicitZeroDimArray) is an ndarray to
matrix.__mul__, so m * np.array(2) is the product with a 1 x 1 matrix and raises
shapes (2,3) and (1,1) not aligned unless m has one column. A 0-d array built another way
(np.zeros(()), x.reshape(())) still scales, where NumPy raises.
Lifetime. A matrix is an NDArray: it holds one counted reference on its buffer and is released with Dispose() or an NDScope. A matrix that views another array (asmatrix(NDArray, DType), T, a row) keeps that array's buffer alive.
Constructors
NDMatrix(UnmanagedStorage)
Wraps storage as a matrix WITHOUT applying the finalize rule and without copying.
Callers must pass a storage object that no other array wraps: an in-place shape change of either
array would otherwise show through the other.
public NDMatrix(UnmanagedStorage storage)
Parameters
storageUnmanagedStorageThe storage to wrap. Its shape becomes the matrix's shape as it is (two axes, or zero axes for the scalar stand-in; see the file header).
Remarks
The constructor a subclass of the matrix chains to (: base(storage)). To build a matrix from data, use
asmatrix(NDArray, DType) or np.matrix: they apply NumPy's finalize rule, which this
constructor skips, so a storage of three axes here gives a three-axis matrix NumPy could never hold.
Exceptions
- NullReferenceException
storageisnull.
Fields
__name__
NumPy's class name (matrix.name), for its error texts; NumSharp.Build's weave reads it to
find the array subclasses and name them in the generated method dispatch.
protected const string __name__ = "matrix"
Field Value
Properties
A
NumPy's matrix.A: this matrix as a plain NDArray of the same shape. Use it to
leave the matrix semantics: m.A * m.A is elementwise and m.A[0] is 1-d.
public NDArray A { get; }
Property Value
- NDArray
A new array object over the same memory (a view; writes reach the matrix). It keeps the matrix's buffer alive while it lives.
Remarks
A cast ((NDArray)m) gives the plain semantics too, without a new object.
A1
NumPy's matrix.A1: the elements as a 1-d NDArray in C order (m.A.ravel()).
public NDArray A1 { get; }
Property Value
- NDArray
A view when the matrix is C-contiguous (writes reach the matrix), otherwise a copy, exactly as
raveldecides.
H
NumPy's matrix.H: the conjugate transpose. For a matrix that is not complex it is the plain
transpose.
public NDMatrix H { get; }
Property Value
- NDMatrix
A new matrix holding the conjugated elements for a complex matrix; for every other dtype the same VIEW as T, so writes reach this matrix there and not for a complex one.
I
NumPy's matrix.I: the multiplicative inverse of a square matrix (np.linalg.inv), the
Moore-Penrose pseudo-inverse of any other (np.linalg.pinv).
public NDMatrix I { get; }
Property Value
- NDMatrix
A new matrix. An integer or bool matrix gives float64; float32 stays float32.
Exceptions
- LinAlgError
A square matrix is singular.
- TypeError
The dtype is not supported by
np.linalg(float16, decimal, char).- NotSupportedException
The matrix is not square and no LAPACK backend is installed: the pseudo-inverse needs a singular value decomposition, which NumSharp.Core does not carry (reference
NumSharp.Interop.OpenBLAS).- ValueError
The matrix object is not two-dimensional (possible only after the rank quirks named on the class): NumPy's
M, N = self.shapefails to unpack.
this[long, long]
The element at (row, column) as a 0-d NDArray,
NumPy's scalar (NumPy's m[i, j] returns a scalar, not a matrix). Assigning stores one
element. Two int arguments bind here too (they convert to long).
public NDArray this[long row, long column] { get; set; }
Parameters
rowlongThe row; negative counts from the end.
columnlongThe column; negative counts from the end.
Property Value
- NDArray
NumPy's scalar: a read-only 0-d COPY of the element, which does not change when the matrix does (
matrix.getitemreturnsndarray.getitem's item unwrapped). Convert it with a cast ((double)m[0, 1]); write an element through the setter.
Remarks
The one C# form beside the general indexer, kept because its TYPE differs: the general indexer is typed NDMatrix (a row, a column, a matrix view), while this element access is typed as the scalar it returns, and it checks both integers inline, so the hot element read allocates no index array. C# prefers it for two integer arguments (a conversion to long is better than one to object).
Exceptions
- IndexError
An index is out of bounds, or the matrix object does not have two axes.
- ValueError
(set) The matrix is read-only, or the value is an array with one or more axes (NumPy's
setting an array element with a sequence.).- TypeError
(set) An array with one or more axes into a complex matrix (NumPy's
only 0-dimensional arrays can be converted to Python scalars).
this[object[]]
NumPy's matrix.getitem(index) as the one C# indexer: integers, slices, slice notation, index arrays,
boolean masks and sequences in any mix, under the matrix rule (file header): a 1-d selection comes back as a
row, or as a column when the second entry is a scalar. A boolean matrix from a comparison is an index too:
m[m > 5] is a (1, K) row.
public NDMatrix this[params object[] key] { get; set; }
Parameters
keyobject[]The index tuple (NumPy's index tuple; a Slice[] or
object[]passed alone is the tuple itself, a single C# tuple entry is spread). A slice-notation string may hold several entries (m["1:3, 0"],m["..., 0"]); an index with a new axis returns a matrix object with three axes, as NumPy does.
Property Value
- NDMatrix
A view for an index of integers and slices (
m[0]andm[0, ":"]are rows,m[":", 0]a column), a new matrix when an index array or a mask is involved; the 0-d scalar stand-in when the selection has no axis left (prefer this[long, long], which is typed as a scalar).
Remarks
The C# setter type. The value assigned through this indexer is an NDMatrix or a C# number (which converts implicitly to a 0-d matrix). Assign any other value (an NDArray, a list) with __setitem__(object, object).
Static, not virtual. This indexer HIDES this[object[]] instead of overriding it: the matrix rule applies through an NDMatrix reference, and through an NDArray reference a matrix is indexed as a plain 2-d array, which the library's own code relies on (an NDArray member reached through the matrix object computes on its plain view, NumSharp.PlainReceiverAttribute).
Dispatch. An index of integers, slices and slice strings takes the view route (one rule for the row /
column shape, NumPy's index errors, and the source's read-only flag kept on the view): the body of the old
typed this[int], this[long], this[string] and this[params Slice[]]. Any other index
runs the plain array's advanced indexing and is shaped by the matrix rule (the old this[NDArray<bool>]
and this[params NDArray<int>[]] took the same route).
Exceptions
- ArgumentNullException
keyor (set) the value isnull.- IndexError
An index is out of bounds, too many entries, a second ellipsis, or index arrays that do not broadcast together (NumPy's texts, checked before the storage is touched).
- ValueError
(set) The matrix is read-only (checked first), or the value does not fit the selection (a 1-d value assigned to a COLUMN raises, as in NumPy: a column is
(N, 1)).
ReprClassName
The class name NumPy's repr() prints before the opening parenthesis: array for an
ndarray, the subclass name for a subclass (type(arr).name in
_array_repr_implementation). Override it in a subclass that NumPy prints under another name;
the continuation lines and a wrapped dtype= suffix are aligned to this name's length, so it
changes the layout of the whole repr, not the first word alone.
protected override string ReprClassName { get; }
Property Value
- string
The name without the parenthesis. Never
nullor empty.
Remarks
A subclass declared outside NumSharp.dll overrides it as protected override; NumPy prints a
user subclass under its own name, which is what such an override should return.
Shape
The shape. Assigning reshapes IN PLACE to the given dimensions and folds them onto two axes; see shape.
public Shape Shape { get; set; }
Property Value
- Shape
The shape descriptor (dimensions, strides, flags).
Exceptions
- AttributeError
(set) The new dimensions need a copy.
- IncorrectShapeException
(set) The new size differs from the matrix's size.
- ValueError
(set) More than two of the new dimensions are longer than 1; the matrix keeps its old shape.
T
NumPy's matrix.T: the transpose, as a matrix.
public NDMatrix T { get; }
Property Value
- NDMatrix
A view with the two axes exchanged (writes reach this matrix).
__array_priority__
NumPy's matrix.array_priority, 10.0: above an ndarray's and a chararray's 0.0, so the class wrap of a
ufunc result and the subtype of a sequence function (np.concatenate) are a matrix whenever one operand
is. The operators of this class resolve a mixed operation at compile time; the public functions read it at
run time.
public override double __array_priority__ { get; }
Property Value
imag
The imaginary part, as a matrix. For a complex matrix a VIEW of the imaginary components; for every other dtype a READ-ONLY matrix of zeros. Assigning stores into the imaginary parts of a complex matrix.
public NDMatrix imag { get; set; }
Property Value
- NDMatrix
The imaginary part.
Exceptions
- TypeError
(set) The matrix is not complex (
array does not have imaginary part to set).- ValueError
(set) The matrix is read-only, or the value does not broadcast.
mT
NumPy's m.mT: the matrix transpose; for a matrix the same as T.
public NDMatrix mT { get; }
Property Value
- NDMatrix
A view.
Exceptions
- ArgumentException
The matrix object has fewer than two axes (the 0-d scalar stand-in).
real
The real part, as a matrix. For a complex matrix a VIEW of the real components (writes reach the matrix); for every other dtype the matrix ITSELF. Assigning stores into the real parts.
public NDMatrix real { get; set; }
Property Value
- NDMatrix
The real part.
Exceptions
- ValueError
(set) The matrix is read-only, or the value does not broadcast.
shape
The dimensions. Assigning reshapes the matrix IN PLACE and folds the new shape back onto two axes,
as NumPy does: m.shape = (6,) gives (1, 6).
public long[] shape { get; set; }
Property Value
- long[]
A copy of the dimensions (
(rows, columns)).
Exceptions
- AttributeError
(set) The new dimensions need a copy (a non-contiguous matrix).
- IncorrectShapeException
(set) The new size differs from the matrix's size.
- ValueError
(set) More than two of the new dimensions are longer than 1 (
shape too large to be a matrix.); the matrix keeps its old shape.
Methods
Clone()
A deep copy that is a matrix again. It HIDES Clone() and does not override it,
on purpose: library code that clones an array it holds as NDArray must get a plain
array back. Which np.* function returns a matrix is decided by its rule
(the mechanism declared on its computation, NumSharp.NDSubclass), never by whether its body happens to
clone the operand.
public NDMatrix Clone()
Returns
- NDMatrix
A new matrix owning a copy of the data.
Equals(object)
Value equality, as for NDArray: true when every element compares equal after
broadcasting. Declared because this class declares ==.
public override bool Equals(object obj)
Parameters
objobjectThe object to compare with.
Returns
- bool
truewhen all elements are equal;falsefornull.
GetEnumerator()
Iterates the rows. Each item is a (1, N) NDMatrix VIEW, as in NumPy, where
iterating an NDArray of the same shape yields 1-d rows.
public IEnumerator GetEnumerator()
Returns
- IEnumerator
An enumerator over the rows; nothing for a matrix with no rows.
Remarks
Each row holds a counted reference on the buffer: dispose the rows in a hot loop, or iterate inside an NDScope.
Exceptions
- TypeError
The matrix is the 0-d scalar stand-in (
iteration over a 0-d array).
GetHashCode()
Always throws: an array is mutable and therefore unhashable, as in NumPy.
public override int GetHashCode()
Returns
- int
Never returns.
Exceptions
- NotSupportedException
Always.
NanMax(NDAxis?, NDArray, bool, object, NDArray)
np.nanmax of a matrix: NumPy's nanmax sends any operand that is not a base ndarray down its branch "slow, but
safe for subclasses of ndarray" (NanExtreme(NDAxis?, NDArray, bool, object, NDArray, bool)), which ends in this class's own max. A 0-d
matrix is a NumPy scalar (a C# number converted to the class), computed as a base ndarray.
protected override NDArray NanMax(NDAxis? axis, NDArray @out, bool keepdims, object initial, NDArray where)
Parameters
axisNDAxis?The axis to reduce, or
nullfor every axis.outNDArrayNumPy's
out=, ornull: a matrix gets the method's answer, a plain array the values and then the method's AttributeError.keepdimsboolThe caller's keepdims;
trueis matrix.max's TypeError.initialobjectThe caller's initial value; any value is matrix.max's TypeError.
whereNDArrayThe caller's mask; any mask is matrix.max's TypeError.
Returns
Remarks
NoInlining | NoOptimization, like every
override of a NumSharp.SubclassAwareAttribute member: the test-time caller audit stops its stack walk at
this frame, and optimized code may tail-call NanExtreme(NDAxis?, NDArray, bool, object, NDArray, bool) and take the frame off the stack, which
makes np.nanmax look like a library caller (the chararray override's failure on PR #633's CI). The guard
test SubclassAuditFrameTests holds every override to it.
Exceptions
- TypeError
See NanExtreme(NDAxis?, NDArray, bool, object, NDArray, bool).
- AttributeError
See NanExtreme(NDAxis?, NDArray, bool, object, NDArray, bool).
- ValueError
See NanExtreme(NDAxis?, NDArray, bool, object, NDArray, bool).
- AxisError
axisis out of bounds.
NanMin(NDAxis?, NDArray, bool, object, NDArray)
np.nanmin of a matrix: NumPy's branch for subclasses (NanExtreme(NDAxis?, NDArray, bool, object, NDArray, bool)), ending in this class's own
min; a 0-d matrix is a NumPy scalar, computed as a base ndarray.
protected override NDArray NanMin(NDAxis? axis, NDArray @out, bool keepdims, object initial, NDArray where)
Parameters
axisNDAxis?The axis to reduce, or
nullfor every axis.outNDArrayNumPy's
out=, ornull.keepdimsboolThe caller's keepdims;
trueis matrix.min's TypeError.initialobjectThe caller's initial value; any value is matrix.min's TypeError.
whereNDArrayThe caller's mask; any mask is matrix.min's TypeError.
Returns
Remarks
The frame stays on the stack for the caller audit, as for NanMax(NDAxis?, NDArray, bool, object, NDArray).
Exceptions
- TypeError
See NanExtreme(NDAxis?, NDArray, bool, object, NDArray, bool).
- AttributeError
See NanExtreme(NDAxis?, NDArray, bool, object, NDArray, bool).
- ValueError
See NanExtreme(NDAxis?, NDArray, bool, object, NDArray, bool).
- AxisError
axisis out of bounds.
__array_finalize__(NDArray, bool, NewSubtypeZeroDim)
NumPy's creation of a result as a matrix (Py_TYPE(self), then matrix.array_finalize): a matrix
view of created under the finalize rule, or WITHOUT it for a result NumPy made by indexing
the matrix with a new axis (np.atleast_3d(m) is the 3-d matrix object m[:, :, newaxis], whose
_getitem flag switches the rule off).
protected override NDArray __array_finalize__(NDArray created, bool keepRank, NewSubtypeZeroDim zeroDim)
Parameters
createdNDArrayThe result computed plain.
keepRankboolSkip the finalize rule (the indexing case).
zeroDimNewSubtypeZeroDimUnused: a matrix's 0-d result is the finalize rule's 1 x 1 matrix.
Returns
- NDArray
createdwhen it is a matrix, else a new matrix view of it.
Exceptions
- ValueError
More than two axes longer than 1, unless
keepRank.
__array_wrap__(NDArray)
NumPy's array_wrap for a matrix, which is ndarray's (array.view(type(self))) followed by
matrix.array_finalize: a matrix view of array under the finalize rule (a 1-d result
is a 1 x N row). An array that already is a matrix comes back itself (a matrix out=, an operand handed back).
protected override NDArray __array_wrap__(NDArray array)
Parameters
arrayNDArrayThe result to wrap.
Returns
- NDArray
arraywhen it is a matrix, else a new matrix view of it (never an owner).
Exceptions
- ValueError
More than two axes longer than 1 (
shape too large to be a matrix.).
__getitem__(object)
NumPy's matrix.getitem(index): the same as m[index] (see this[object[]]).
public NDMatrix __getitem__(object index)
Parameters
indexobjectThe index as Python passes it: one entry (an integer, slice notation, a Slice, an index array, a mask) or the whole index tuple — an
object[]/ Slice[] or a C# tuple. A typed C# array other than Slice[] (int[],bool[]) is ONE entry, NumPy's list.
Returns
- NDMatrix
The selection under the matrix rule: a row for one integer, the 0-d scalar stand-in for two.
Remarks
One member, NumPy's signature (it replaced the int, long, string and
params int[] forms; coordinates are a tuple: m.getitem((1, 2))).
Exceptions
- ArgumentNullException
indexisnull.- IndexError
An index is out of bounds, more entries than axes, or an invalid index.
__imul__(object)
Python's m *= other: computes m * other and stores it IN this matrix
(self[:] = self * other). C#'s own m *= x rebinds the variable to a new matrix instead;
call this method when other references to the matrix must see the product.
public NDMatrix __imul__(object other)
Parameters
otherobjectA matrix, an array, a sequence or a number.
Returns
- NDMatrix
This matrix.
Remarks
The product is cast to this matrix's dtype with an unsafe cast, as NumPy's assignment does: an int matrix times 2.5 keeps truncated integers.
Exceptions
- ArgumentNullException
otherisnull.- ArgumentException
The inner dimensions differ.
- ValueError
This matrix is read-only, or the product's shape is not this matrix's shape (
could not broadcast input array from shape (2,3) into shape (2,2)).- NotSupportedException
The operand holds strings or other objects.
__ipow__(object)
Python's m **= other: computes the power and stores it IN this matrix
(self[:] = self ** other).
public NDMatrix __ipow__(object other)
Parameters
otherobjectThe exponent (see __pow__(object)).
Returns
- NDMatrix
This matrix.
Remarks
The result is cast to this matrix's dtype with an unsafe cast: the inverse of an integer matrix is truncated.
Exceptions
- LinAlgError
The matrix is not square, or a negative power of a singular matrix.
- TypeError
The exponent is not an integer.
- ValueError
This matrix is read-only.
__iter__()
Python's iter(m): the same as GetEnumerator().
public IEnumerator __iter__()
Returns
- IEnumerator
An enumerator over the rows as
(1, N)matrices.
Exceptions
- TypeError
The matrix is the 0-d scalar stand-in.
__mul__(object)
Python's m.mul(other): the same as m * other.
public NDMatrix __mul__(object other)
Parameters
otherobjectA matrix, an array, a sequence or a number.
Returns
- NDMatrix
The product as a new matrix.
Exceptions
- ArgumentNullException
otherisnull.- ArgumentException
The inner dimensions differ.
- ValueError
The operand is ragged or cannot be seen as a matrix.
- NotSupportedException
The operand holds strings or other objects.
__pow__(int)
NumPy's m ** n: the matrix power (np.linalg.matrix_power), m multiplied by itself
n times. Zero gives the identity, a negative power inverts first.
public NDMatrix __pow__(int n)
Parameters
nintThe integer power.
Returns
- NDMatrix
This matrix ITSELF for
n == 1(as NumPy); a new matrix otherwise. A negative power of an integer matrix is float64.
Exceptions
- LinAlgError
The matrix is not square (
Last 2 dimensions of the array must be square), or a negative power of a singular matrix (Singular matrix).- TypeError
A negative power of a dtype
np.linalgdoes not support.
__pow__(object)
NumPy's m ** other for an exponent of any kind: an integer, a bool or a 0-d integer
array is the power; anything else is refused.
public NDMatrix __pow__(object other)
Parameters
otherobjectThe exponent.
Returns
- NDMatrix
The power; see __pow__(int).
Exceptions
- LinAlgError
The matrix is not square (tested BEFORE the exponent, as NumPy does), or a negative power of a singular matrix.
- TypeError
The exponent is not an integer (
exponent must be an integer).- OverflowException
The exponent does not fit int.
__rmul__(object)
Python's m.rmul(other): other * m, with other on the LEFT.
public NDMatrix __rmul__(object other)
Parameters
otherobjectAn array, a sequence or a number.
Returns
- NDMatrix
The product as a new matrix.
Exceptions
- ArgumentNullException
otherisnull.- ArgumentException
The inner dimensions differ.
- ValueError
The operand is ragged, or the product cannot be a matrix.
- NotSupportedException
The operand holds strings or other objects.
__rpow__(object)
Python's other ** m. NumPy's matrix does not define it (rpow returns
NotImplemented), so a number base is an error; an ARRAY base is ndarray's own elementwise power,
wrapped as a matrix.
public NDMatrix __rpow__(object other)
Parameters
otherobjectThe base.
Returns
- NDMatrix
np.power(other, m)as a matrix whenotheris an array.
Exceptions
- TypeError
otheris not an array: Python'sunsupported operand type(s) for ** or pow(): 'int' and 'matrix'.- ValueError
The broadcast result has more than two axes longer than 1.
__setitem__(object, object)
Python's m[index] = value for ANY value: an NDArray, a C# array, a list or tuple,
or a number. Use it where the indexer does not accept the value (it takes an NDMatrix or a
number).
public void __setitem__(object index, object value)
Parameters
indexobjectOne index entry, or the whole index tuple as an
object[]or a C# tuple.valueobjectThe value. A 1-d value assigned to a COLUMN raises, as in NumPy: a column is
(N, 1).
Remarks
One member (it replaced the int, long and string index forms). It HIDES
__setitem__(object, object): numpy.matrix inherits ndarray's setitem, whose
view assignment goes through the subclass's own getitem, so the matrix rule applies here.
Exceptions
- ArgumentNullException
valueisnull.- IndexError
The index is invalid for this matrix.
- ValueError
The matrix is read-only, the value does not broadcast to the selection, or an array is assigned to a single element.
_collapse(NDArray, NDAxis?)
NumPy's matrix._collapse(axis) (self[0, 0] if axis is None else self) after a
keepdims=True reduction, applied to the plain reduction (keepdims=False) NumPy's function computed: the
scalar without an axis, a column (N, 1) along the last axis, a row or the matrix itself otherwise.
protected override NDArray _collapse(NDArray result, NDAxis? axis)
Parameters
resultNDArrayThe plain reduction, unaliased (a matrix result, or
null, comes back itself).axisNDAxis?The caller's axis:
nullfor None (the scalar), an int or a one-axis tuple (row or column for the 1-d plain result), or any other tuple (the kept-dims matrix:(1, 1)for(0, 1), the whole matrix for()).
Returns
- NDArray
The 0-d result, a matrix view of it, or
result.
Exceptions
- ValueError
More than two axes longer than 1.
all(int, NDArray)
NumPy's matrix.all(axis) for a given axis: whether every element along the axis is
true, as a bool matrix that keeps the reduced axis with length 1.
public NDMatrix all(int axis, NDArray @out = null)
Parameters
axisintThe axis to test along.
outNDArrayMust be
null(NumPy'smatrix.allcannot take one; see any(int, NDArray)).
Returns
- NDMatrix
The bool matrix.
Remarks
The C#-typed form of all(NDAxis?, NDArray) for an int axis (C# prefers it there, an exact match): the answer is always a matrix, so the declared type says so.
Exceptions
all(NDAxis?, NDArray)
NumPy's matrix.all(axis=None, out=None): ndarray.all(self, axis, out,
keepdims=True)._collapse(axis) - a bool scalar without an axis, otherwise a bool matrix that keeps the
reduced axes with length 1.
public NDArray all(NDAxis? axis = null, NDArray @out = null)
Parameters
axisNDAxis?The axis, a tuple of axes (
(0, 1)gives the 1 x 1 matrix,()the truth values as a matrix), ornullto test every element.outNDArrayMust be
null.
Returns
- NDArray
A 0-d bool array when
axisisnull; otherwise the bool matrix.
Remarks
The target of np.all's generated matrix dispatch ([NumPyMethod]), which hands it
np.all's own NDAxis by name and type: a tuple reaches it since API-parity batch 4
(2026-11-12), where the int? axis it took gave a tuple call no matrix arm (a plain result).
Exceptions
- AxisError
An axis is out of bounds.
- ValueError
A tuple names an axis twice (
duplicate value in 'axis').- TypeError
outis notnull.
any(int, NDArray)
NumPy's matrix.any(axis) for a given axis: whether any element along the axis is true,
as a bool matrix that keeps the reduced axis with length 1.
public NDMatrix any(int axis, NDArray @out = null)
Parameters
axisintThe axis to test along.
outNDArrayMust be
null. NumPy'smatrix.anyhands an out to the argument slot ofdtype, so every call with one fails.
Returns
- NDMatrix
The bool matrix.
Remarks
The C#-typed form of any(NDAxis?, NDArray) for an int axis (C# prefers it there, an exact match): the answer is always a matrix, so the declared type says so.
Exceptions
- AxisError
axisis out of bounds.- TypeError
outis notnull(NumPy'sCannot construct a dtype from an array).
any(NDAxis?, NDArray)
NumPy's matrix.any(axis=None, out=None): ndarray.any(self, axis, out,
keepdims=True)._collapse(axis) - a bool scalar without an axis, otherwise a bool matrix that keeps the
reduced axes with length 1.
public NDArray any(NDAxis? axis = null, NDArray @out = null)
Parameters
axisNDAxis?The axis, a tuple of axes (
(0, 1)gives the 1 x 1 matrix,()the truth values as a matrix), ornullto test every element.outNDArrayMust be
null(see any(int, NDArray)).
Returns
- NDArray
A 0-d bool array when
axisisnull; otherwise the bool matrix.
Remarks
The target of np.any's generated matrix dispatch ([NumPyMethod]), which hands it
np.any's own NDAxis by name and type: a tuple reaches it since API-parity batch 4
(2026-11-12), where the int? axis it took gave a tuple call no matrix arm (a plain result).
Exceptions
- AxisError
An axis is out of bounds.
- ValueError
A tuple names an axis twice (
duplicate value in 'axis').- TypeError
outis notnull.
argmax(int, NDArray)
NumPy's matrix.argmax(axis, out) for a given axis: the index of the first maximum along the
axis, as a row (1, N) for axis 0 and a column (M, 1) for axis 1. The flat form
m.argmax() is the inherited member and returns a long.
public NDMatrix argmax(int axis, NDArray @out = null)
Parameters
axisint0 or 1. NumPy's matrix rejects every other value, also -1 and -2.
outNDArrayMust be
nullin practice: NumPy's method fails for every out (see the exceptions).
Returns
- NDMatrix
The int64 indices as a matrix.
Exceptions
- AxisError
axisis out of bounds.- ValueError
axisis a valid negative axis (unsupported axis), the reduced axis is empty, oroutdoes not have the 1-d result's shape (output array does not match result of np.argmax.; a matrix never has it).- AttributeError
outis a plain array of the right shape: NumPy writes the indices into it and then fails onout._align.
argmax(NDAxis?, NDArray)
NumPy's matrix.argmax(axis=None, out=None) as the generated dispatch of np.argmax calls it
(_wrapfunc): along an axis the aligned index matrix of argmax(int, NDArray); without one
_align(None), the flat index as a 0-d array.
public NDArray argmax(NDAxis? axis, NDArray @out = null)
Parameters
axisNDAxis?The axis (0 or 1), or
nullfor the flat index. Deliberately without a default: anm.argmax()call keeps binding the inherited flat member.outNDArrayMust be
nullin practice: NumPy's method fails on every out (see the exceptions).
Returns
- NDArray
The 0-d flat index without an axis; otherwise the int64 index matrix.
Exceptions
- AxisError
axisis out of bounds.- ValueError
An unsupported (negative) axis, an empty reduction, or a misshapen out.
- AttributeError
outis a plain array NumPy's reduction accepted:_aligndoes not exist on it.
argmin(int, NDArray)
NumPy's matrix.argmin(axis, out) for a given axis; see
argmax(int, NDArray).
public NDMatrix argmin(int axis, NDArray @out = null)
Parameters
Returns
- NDMatrix
The int64 indices as a matrix.
Exceptions
- AxisError
axisis out of bounds.- ValueError
An unsupported axis, an empty reduction, or a misshapen
out.- AttributeError
outis a plain array of the right shape.
argmin(NDAxis?, NDArray)
NumPy's matrix.argmin(axis=None, out=None) as np.argmin's generated dispatch calls it; see
argmax(NDAxis?, NDArray).
public NDArray argmin(NDAxis? axis, NDArray @out = null)
Parameters
axisNDAxis?The axis (0 or 1), or
nullfor the flat index (no default, see argmax).outNDArrayMust be
nullin practice.
Returns
- NDArray
The 0-d flat index without an axis; otherwise the int64 index matrix.
Exceptions
- AxisError
axisis out of bounds.- ValueError
An unsupported axis, an empty reduction, or a misshapen out.
- AttributeError
outis a plain array NumPy's reduction accepted.
argpartition(int, int?, string, string)
The indices that partition the matrix around the kth element along an
axis, as a matrix (NumPy's m.argpartition(kth, axis)).
public NDMatrix argpartition(int kth, int? axis = -1, string kind = "introselect", string order = null)
Parameters
kthintThe element position that ends up in its sorted place.
axisint?The axis (default: the last), or
nullfor the flattened matrix.kindstringOnly
"introselect".orderstringMust be
null(no structured dtypes).
Returns
- NDMatrix
A new int64 matrix.
Exceptions
- ValueError
kthis out of bounds, an unknownkind, or anorder.- AxisError
axisis out of bounds.
argsort(int?)
The indices that sort the matrix along an axis, as a matrix (NumPy's m.argsort(axis));
with null the indices of the flattened matrix as a row.
public NDMatrix argsort(int? axis = -1)
Parameters
axisint?The axis (default: the last), or
null.
Returns
- NDMatrix
A new int64 matrix.
Exceptions
- AxisError
axisis out of bounds.
astype(DType, char, string, bool)
Copy of the matrix cast to dtype with a memory order and a casting
rule, as a matrix (NumPy's m.astype(dtype, order='K', casting='unsafe', copy=True)).
public NDMatrix astype(DType dtype, char order = 'K', string casting = "unsafe", bool copy = true)
Parameters
dtypeDTypeThe target dtype.
orderchar'C','F','A'or'K'.castingstringNumPy's casting rule name (
"unsafe","same_kind","safe","equiv","no").copyboolfalsereturns this matrix itself when nothing has to change.
Returns
- NDMatrix
A new matrix that owns its data, or this matrix.
Remarks
One signature in NumPy's keyword order (API-parity batch 3, 2026-11-11): it replaces the former
astype(dtype, copy) / astype(dtype, copy, order, casting) pair, so a.astype(t, 'F') reads as
in NumPy and a copy is written copy: false (a bool in the order slot does not compile).
subok stays out (the parity plan's exclusion).
Exceptions
- InvalidCastException
The cast is not allowed under
casting.- ValueError
An unknown order or casting name.
byteswap(bool)
The matrix with its bytes swapped, as a matrix.
public NDMatrix byteswap(bool inplace = false)
Parameters
inplacebooltrueswaps this matrix's own bytes and returns this matrix.
Returns
- NDMatrix
A new matrix, or this matrix when
inplaceistrue.
Exceptions
- ValueError
inplaceon a read-only matrix.
choose(NDArray[], string)
Builds a matrix by picking, per element, from choices the array this
matrix's integer value names (NumPy's m.choose(choices, mode)).
public NDMatrix choose(NDArray[] choices, string mode = "raise")
Parameters
choicesNDArray[]The arrays to choose from; they broadcast against this matrix.
modestring"raise","wrap"or"clip"for a value out of range.
Returns
- NDMatrix
A new matrix.
Exceptions
- ValueError
A value is out of range under
"raise", or the shapes do not broadcast.- ArgumentException
An unknown mode.
clip(NDArray, NDArray)
The elements limited to an interval, as a matrix (NumPy's m.clip(min, max)).
public NDMatrix clip(NDArray min = null, NDArray max = null)
Parameters
minNDArrayThe lower bound, or
nullfor none. It broadcasts against the matrix.maxNDArrayThe upper bound, or
nullfor none. With no bound at all the result is a copy.
Returns
- NDMatrix
A matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
A bound does not broadcast against the matrix.
compress(NDArray, int?)
The rows or columns where condition is true, as a matrix (NumPy's
m.compress(condition, axis)); without an axis the flattened elements as a row.
public NDMatrix compress(NDArray condition, int? axis = null)
Parameters
Returns
- NDMatrix
A new matrix that owns its data.
Exceptions
- ValueError
conditionis not 1-d.- IndexError
conditionis longer than the axis.- AxisError
axisis out of bounds.
conj()
The complex conjugate, as a matrix. A matrix that is not complex is returned ITSELF (as NumPy does), not a copy.
public NDMatrix conj()
Returns
- NDMatrix
The conjugated matrix, or this matrix.
conjugate()
The complex conjugate, as a matrix. A matrix that is not complex is returned ITSELF (as NumPy does), not a copy.
public NDMatrix conjugate()
Returns
- NDMatrix
The conjugated matrix, or this matrix.
copy(char)
A copy of the matrix, as a matrix (NumPy's m.copy(order)).
public NDMatrix copy(char order = 'C')
Parameters
orderchar'C','F','A'or'K'.
Returns
- NDMatrix
A new matrix that owns its data.
Exceptions
- ArgumentException
An unknown order.
cumprod(int?, DType)
The running product along an axis, as a matrix (NumPy's m.cumprod(axis, dtype)).
public NDMatrix cumprod(int? axis = null, DType dtype = null)
Parameters
axisint?The axis, or
nullfor the flattened matrix (a row).dtypeDTypeThe dtype of the products, or
nullfor NumPy's default.
Returns
- NDMatrix
A matrix (a view of the computed array).
Exceptions
- AxisError
axisis out of bounds.
cumsum(int?, DType)
The running sum along an axis, as a matrix (NumPy's m.cumsum(axis, dtype)); without an
axis the running sum of the flattened matrix as a row.
public NDMatrix cumsum(int? axis = null, DType dtype = null)
Parameters
Returns
- NDMatrix
A matrix (a view of the computed array).
Exceptions
- AxisError
axisis out of bounds.
diagonal(int, int, int)
The diagonal as a ROW (1, K) (NumPy's m.diagonal()).
public NDMatrix diagonal(int offset = 0, int axis1 = 0, int axis2 = 1)
Parameters
offsetint0 for the main diagonal, positive above it, negative below.
axis1intThe first axis of the plane.
axis2intThe second axis of the plane.
Returns
- NDMatrix
A READ-ONLY view of the diagonal.
Exceptions
- AxisError
An axis is out of bounds.
- ArgumentException
The two axes are the same.
dot(NDArray)
NumPy's m.dot(b): the dot product through the ARRAY method, folded onto two axes. With a 1-d
b of length N the product of an (M, N) matrix is the 1-d vector of length M,
which comes back as a ROW (1, M) — not the column a matrix product would give.
public NDMatrix dot(NDArray b)
Parameters
bNDArrayThe right operand, as it is (it is NOT seen as a matrix first, unlike
m * b).
Returns
- NDMatrix
A new matrix.
Exceptions
- ArgumentException
The shapes are not aligned.
- ValueError
The product has more than two axes longer than 1.
flatten(char)
NumPy's matrix.flatten(order='C'): a COPY of the elements as one row
(1, size).
public NDMatrix flatten(char order = 'C')
Parameters
orderchar'C'row-major,'F'column-major,'A'column-major when the matrix is F-contiguous,'K'memory order.
Returns
- NDMatrix
A new
(1, size)matrix that owns its data.
Exceptions
- ArgumentException
orderis not one of the four codes.
getA()
NumPy's matrix.getA(): the same as A.
public NDArray getA()
Returns
- NDArray
A plain array view of this matrix.
getA1()
NumPy's matrix.getA1(): the same as A1.
public NDArray getA1()
Returns
- NDArray
The elements as a 1-d array.
getH()
NumPy's matrix.getH(): the same as H.
public NDMatrix getH()
Returns
- NDMatrix
The conjugate transpose.
getI()
NumPy's matrix.getI(): the same as I.
public NDMatrix getI()
Returns
- NDMatrix
The inverse or pseudo-inverse.
Exceptions
- LinAlgError
A square matrix is singular.
- TypeError
The dtype is not supported by
np.linalg.- NotSupportedException
A non-square matrix without a LAPACK backend.
- ValueError
The matrix object is not two-dimensional.
getT()
NumPy's matrix.getT(): the same as T.
public NDMatrix getT()
Returns
- NDMatrix
The transposed matrix (a view).
getfield(DType, int)
A field of the matrix's elements read as dtype at a byte offset, as a
matrix (a view).
public NDMatrix getfield(DType dtype, int offset = 0)
Parameters
Returns
- NDMatrix
A view of the field.
Exceptions
- ValueError
The field does not fit inside an element.
max(int, NDMatrix)
NumPy's matrix.max(axis, out) for a given axis: the maxima as a matrix that keeps the
reduced axis with length 1.
public NDMatrix max(int axis, NDMatrix @out = null)
Parameters
axisintThe axis to reduce along.
outNDMatrixA matrix of the result's shape to store into, or
null.
Returns
- NDMatrix
The maxima;
outitself when it is given.
Exceptions
- AxisError
axisis out of bounds.- ValueError
The reduced axis is empty (
zero-size array to reduction operation maximum which has no identity), oroutis read-only or misshapen.
max(NDAxis?, NDArray)
NumPy's matrix.max(axis=None, out=None); the largest element as a scalar without an
axis.
public NDArray max(NDAxis? axis = null, NDArray @out = null)
Parameters
axisNDAxis?The axis, or
nullfor the maximum of every element. A tuple reduces the axes it names together and keeps them with length 1, as every axis but None does (the(1, 1)matrix for(0, 1), the whole matrix for()).outNDArrayA matrix of the kept shape, or
null.
Returns
- NDArray
A 0-d array when
axisisnull; otherwise the matrix.
Exceptions
- AxisError
axisis out of bounds.- ValueError
An empty reduction, or
outis read-only or misshapen.- AttributeError
outis a plain (non-matrix) array: NumPy's reduction stores into it, then_collapsedoes not exist on it ('numpy.ndarray' object has no attribute '_collapse').
mean(int, DType, NDMatrix)
NumPy's matrix.mean(axis, dtype, out) for a given axis: the means as a matrix that
keeps the reduced axis with length 1.
public NDMatrix mean(int axis, DType dtype = null, NDMatrix @out = null)
Parameters
axisintThe axis to average along.
dtypeDTypeThe dtype the mean is computed in, or
null: float64 for an integer or bool matrix, the matrix's own dtype for a float one.outNDMatrixA matrix of the result's shape to store into, or
null.
Returns
- NDMatrix
The means;
outitself when it is given.
Exceptions
- AxisError
axisis out of bounds.- ValueError
outis read-only or does not have the result's shape.
mean(NDAxis?, DType, NDArray)
NumPy's matrix.mean(axis=None, dtype=None, out=None); a scalar without an axis.
public NDArray mean(NDAxis? axis = null, DType dtype = null, NDArray @out = null)
Parameters
axisNDAxis?The axis, or
nullfor the mean of every element. A tuple reduces the axes it names together and keeps them with length 1, as every axis but None does (the(1, 1)matrix for(0, 1), the whole matrix for()).dtypeDTypeThe dtype the mean is computed in, or
nullfor NumPy's default.outNDArrayA matrix of the kept shape, or
null. NumPy stores the SUM into it and then divides it in place, both with an unsafe cast, so an integeroutholds truncated values.
Returns
- NDArray
A 0-d array when
axisisnull; otherwise the matrix.
Exceptions
- AxisError
axisis out of bounds.- ValueError
outis read-only or does not have the result's shape.- AttributeError
outis a plain (non-matrix) array: NumPy's reduction stores into it, then_collapsedoes not exist on it ('numpy.ndarray' object has no attribute '_collapse').
min(int, NDMatrix)
NumPy's matrix.min(axis, out) for a given axis: the minima as a matrix that keeps the
reduced axis with length 1.
public NDMatrix min(int axis, NDMatrix @out = null)
Parameters
axisintThe axis to reduce along.
outNDMatrixA matrix of the result's shape to store into, or
null.
Returns
- NDMatrix
The minima;
outitself when it is given.
Exceptions
- AxisError
axisis out of bounds.- ValueError
An empty reduction, or
outis read-only or misshapen.
min(NDAxis?, NDArray)
NumPy's matrix.min(axis=None, out=None); the smallest element as a scalar without an
axis.
public NDArray min(NDAxis? axis = null, NDArray @out = null)
Parameters
axisNDAxis?The axis, or
nullfor the minimum of every element. A tuple reduces the axes it names together and keeps them with length 1, as every axis but None does (the(1, 1)matrix for(0, 1), the whole matrix for()).outNDArrayA matrix of the kept shape, or
null.
Returns
- NDArray
A 0-d array when
axisisnull; otherwise the matrix.
Exceptions
- AxisError
axisis out of bounds.- ValueError
An empty reduction, or
outis read-only or misshapen.- AttributeError
outis a plain (non-matrix) array: NumPy's reduction stores into it, then_collapsedoes not exist on it ('numpy.ndarray' object has no attribute '_collapse').
prod(int, DType, NDMatrix)
NumPy's matrix.prod(axis, dtype, out) for a given axis: the products as a matrix that
keeps the reduced axis with length 1.
public NDMatrix prod(int axis, DType dtype = null, NDMatrix @out = null)
Parameters
axisintThe axis to multiply along.
dtypeDTypeThe dtype the product is computed in, or
nullfor NumPy's default.outNDMatrixA matrix of the result's shape to store into, or
null.
Returns
- NDMatrix
The products;
outitself when it is given.
Exceptions
- AxisError
axisis out of bounds.- ValueError
outis read-only or does not have the result's shape.
prod(NDAxis?, DType, NDArray)
NumPy's matrix.prod(axis=None, dtype=None, out=None); a scalar without an axis.
public NDArray prod(NDAxis? axis = null, DType dtype = null, NDArray @out = null)
Parameters
axisNDAxis?The axis, or
nullto multiply every element. A tuple reduces the axes it names together and keeps them with length 1, as every axis but None does (the(1, 1)matrix for(0, 1), the whole matrix for()).dtypeDTypeThe dtype the product is computed in, or
nullfor NumPy's default.outNDArrayA matrix of the kept shape, or
null.
Returns
- NDArray
A 0-d array when
axisisnull; otherwise the matrix.
Exceptions
- AxisError
axisis out of bounds.- ValueError
outis read-only or does not have the result's shape.- AttributeError
outis a plain (non-matrix) array: NumPy's reduction stores into it, then_collapsedoes not exist on it ('numpy.ndarray' object has no attribute '_collapse').
ptp(int, NDArray)
NumPy's matrix.ptp(axis, out) for a given axis: maximum minus minimum along the axis, as a
row for axis 0 and a column for axis 1.
public NDMatrix ptp(int axis, NDArray @out = null)
Parameters
axisint0 or 1. NumPy's matrix rejects every other value, also -1 and -2.
outNDArrayMust be
nullin practice: NumPy's method fails for every out.
Returns
- NDMatrix
The ranges as a matrix.
Exceptions
- AxisError
axisis out of bounds.- ValueError
axisis a valid negative axis (unsupported axis), the reduced axis is empty, oroutis given (NumPy's reduction or broadcast text; a plain out of the right shape holds the MAXIMA afterwards, as in NumPy).- TypeError
The matrix is bool (NumPy's boolean subtract is not defined).
ptp(int?, NDArray)
NumPy's matrix.ptp(axis=None, out=None); the range of every element as a scalar without
an axis.
public NDArray ptp(int? axis = null, NDArray @out = null)
Parameters
Returns
- NDArray
A 0-d array when
axisisnull; otherwise the matrix.
Exceptions
- AxisError
axisis out of bounds.- ValueError
An unsupported axis, an empty reduction, or an
out.- TypeError
The matrix is bool.
ravel(char)
NumPy's matrix.ravel(order='C'): the elements as one row (1, size), a VIEW
when the memory layout allows it and a copy otherwise. One signature: m.ravel() binds here with the
default 'C'.
public NDMatrix ravel(char order = 'C')
Parameters
orderchar'C','F','A'or'K'.
Returns
- NDMatrix
A
(1, size)matrix; writes reach this matrix when it is a view.
Exceptions
- ArgumentException
orderis not one of the four codes.
repeat(NDArray, int?)
Each element repeated, as a matrix (NumPy's m.repeat(repeats, axis)); without an axis
the flattened elements as a row.
public NDMatrix repeat(NDArray repeats, int? axis = null)
Parameters
repeatsNDArrayThe repeat count per element (one value, or one per element of the axis).
axisint?The axis to repeat along, or
null.
Returns
- NDMatrix
A new matrix that owns its data.
Remarks
One signature (API-parity batch 3): m.repeat(2) binds here, 2 converting to a 0-d array.
Exceptions
- ValueError
A negative count, or counts that do not match the axis.
- AxisError
axisis out of bounds.
reshape(Shape, char, bool?)
The elements under another shape, folded back onto two axes — NumPy's m.reshape(shape, /, *, order='C',
copy=None): m.reshape(6) is the row (1, 6), m.reshape(1, 2, 3) is (2, 3).
public NDMatrix reshape(Shape shape, char order = 'C', bool? copy = null)
Parameters
shapeShapeThe requested dimensions; one may be -1 (inferred). An int, a tuple, an
int[]/long[], a C# collection expression or a Shape. NumPy's name (it wasnewShapeuntil API-parity batch 9).orderchar'C','F'or'A': the index order the elements are read and placed in.copybool?NumPy's
copykeyword (API-parity batch 7): null copies only when needed, true always copies, false raisesUnable to avoid creating a copy while reshaping.when no view can express the reshape — reshape(Shape, char, bool?)'s modes, the result wrapped as a matrix.
Returns
- NDMatrix
A view when the layout allows it, otherwise a copy.
Remarks
Since API-parity batch 9 this and the loose integers (reshape(params long[]),
reshape(params ReadOnlySpan<long>)) are the matrix method's spellings, as on NDArray: the
(Shape) and (int[]) forms went (an int[] converts to Shape).
Exceptions
- IncorrectShapeException
The requested size differs from the matrix's size.
- ValueError
More than two requested dimensions are longer than 1, 'K', or the refusal of
copyfalse.
reshape(params long[])
The matrix under NumPy's loose integers (m.reshape(2, 3)) when they arrive as a C#
long[]; see reshape(Shape, char, bool?).
public NDMatrix reshape(params long[] shape)
Parameters
shapelong[]The requested dimensions; one may be -1.
Returns
- NDMatrix
A view when the layout allows it, otherwise a copy, wrapped as a matrix.
Exceptions
- IncorrectShapeException
The requested size differs from the matrix's size.
- ValueError
More than two requested dimensions are longer than 1.
reshape(params ReadOnlySpan<long>)
The matrix under NumPy's loose integers (m.reshape(2, 3)) as C# 13 passes them (a span on the
caller's stack); added in API-parity batch 9 so the matrix has NDArray's spellings.
public NDMatrix reshape(params ReadOnlySpan<long> shape)
Parameters
shapeReadOnlySpan<long>The requested dimensions; one may be -1.
Returns
- NDMatrix
A view when the layout allows it, otherwise a copy, wrapped as a matrix.
Exceptions
- IncorrectShapeException
The requested size differs from the matrix's size.
- ValueError
More than two requested dimensions are longer than 1.
round()
The elements rounded to the nearest integer (half to even), as a matrix (NumPy's
m.round()).
public NDMatrix round()
Returns
- NDMatrix
A matrix: a copy that owns its data for an integer or complex matrix, a view of the computed array for a float one. A bool matrix gives float16, as NumPy's
rintdoes.
round(int, NDArray)
The elements rounded to decimals places (NumPy's m.round(decimals, out)).
The CLASS of the result depends on the route NumPy's PyArray_Round takes, so the static type
is NDArray.
public NDArray round(int decimals, NDArray @out = null)
Parameters
decimalsintThe number of decimal places; negative rounds left of the point.
outNDArrayAn array to store into, or
null.
Returns
- NDArray
outwhen given. Otherwise: a MATRIX that owns its data for a complex matrix (NumPy copies the matrix and rounds the two parts in place) and for an integer matrix withdecimals>= 0 (a copy); a MATRIX viewing the computed array fordecimals0 of a float or bool matrix (therintufunc); a plain ARRAY for every other case (NumPy computes those into an array it allocates itself).
Exceptions
- ValueError
outis read-only or misshapen.- InvalidCastException
The result cannot be cast to
out's dtype.
squeeze(NDAxis?)
NumPy's matrix.squeeze(axis=None). A matrix cannot lose an axis, so a squeezed row stays a row
and a squeezed COLUMN becomes a row (1, N).
public NDMatrix squeeze(NDAxis? axis = null)
Parameters
axisNDAxis?The axis to remove, a tuple of axes (each of length 1; checked in order, NumPy's
PyArray_ConvertMultiAxis), ornullfor every axis of length 1.
Returns
- NDMatrix
This matrix ITSELF when it has no axis of length 1 and
axisisnull(as NumPy); otherwise a view.
Remarks
The axis is an NDAxis since API-parity batch 4 (2026-11-12), as
squeeze(NDAxis?)'s; an int? argument converts.
Exceptions
- AxisError
An axis is out of bounds.
- ValueError
An axis longer than 1 is named (
cannot select an axis to squeeze out which has size not equal to one), or a tuple names an axis twice (duplicate value in 'axis').
std(int, DType, NDMatrix, int)
NumPy's matrix.std(axis, dtype, out, ddof) for a given axis: the standard deviations
as a matrix that keeps the reduced axis with length 1.
public NDMatrix std(int axis, DType dtype = null, NDMatrix @out = null, int ddof = 0)
Parameters
axisintThe axis to compute along.
dtypeDTypeThe dtype of the computation, or
nullfor NumPy's default.outNDMatrixA FLOAT or complex matrix of the result's shape to store into, or
null.ddofintDelta degrees of freedom.
Returns
- NDMatrix
The standard deviations;
outitself when it is given.
Exceptions
- AxisError
axisis out of bounds.- ValueError
outis read-only or does not have the result's shape.- TypeError
outordtypeis an integer or bool dtype: NumPy'sCannot cast ufunc 'sqrt' output from dtype('float64') to dtype('int64') with casting rule 'same_kind'.
std(NDAxis?, DType, NDArray, int)
NumPy's matrix.std(axis=None, dtype=None, out=None, ddof=0); a scalar without an axis.
public NDArray std(NDAxis? axis = null, DType dtype = null, NDArray @out = null, int ddof = 0)
Parameters
axisNDAxis?The axis, or
nullfor the standard deviation of every element. A tuple reduces the axes it names together and keeps them with length 1, as every axis but None does (the(1, 1)matrix for(0, 1), the whole matrix for()).dtypeDTypeThe dtype of the computation, or
nullfor NumPy's default.outNDArrayA float or complex matrix of the kept shape, or
null.ddofintDelta degrees of freedom.
Returns
- NDArray
A 0-d array when
axisisnull; otherwise the matrix.
Exceptions
- AxisError
axisis out of bounds.- ValueError
outis read-only or does not have the result's shape.- TypeError
An integer or bool
outordtype: NumPy takes the square root in place under thesame_kindrule, which refuses the float root for an integer array (Cannot cast ufunc 'sqrt' output from dtype('float64') to dtype('int64') with casting rule 'same_kind'). Withoutthis is raised AFTER the variance was stored, sooutholds the variance then. It is raised foraxis: nulltoo: a matrix reduction is always the array form (file header), wherenp.stdof a plain array returns a truncated scalar.- AttributeError
outis a plain (non-matrix) array: NumPy's reduction stores into it, then_collapsedoes not exist on it ('numpy.ndarray' object has no attribute '_collapse').
sum(int, DType, NDMatrix)
NumPy's matrix.sum(axis, dtype, out) for a given axis: the sums as a matrix that keeps the
reduced axis with length 1 ((1, N) for axis 0, (M, 1) for axis 1 or -1).
public NDMatrix sum(int axis, DType dtype = null, NDMatrix @out = null)
Parameters
axisintThe axis to sum along (0, 1, -1 or -2).
dtypeDTypeThe dtype the sum is computed in, or
nullfor NumPy's default (integers narrower than 64 bits and bool accumulate in int64 / uint64).outNDMatrixA matrix of the result's shape to store into, or
null. Withoutdtypethe sum is computed inpromote_types(out.dtype, this.dtype)and stored with an unsafe cast: an int8 matrix into an int8 out wraps, into a bool out it stores whether the int8 sum is non-zero.
Returns
- NDMatrix
The sums;
outitself when it is given.
Exceptions
- AxisError
axisis out of bounds.- ValueError
outis read-only or does not have the result's shape (NumPy's reduction texts).
sum(NDAxis?, DType, NDArray)
NumPy's matrix.sum(axis=None, dtype=None, out=None). Without an axis the result is the sum of
all elements as a scalar.
public NDArray sum(NDAxis? axis = null, DType dtype = null, NDArray @out = null)
Parameters
axisNDAxis?The axis, or
nullto sum every element. A tuple reduces the axes it names together and keeps them with length 1, as every axis but None does (the(1, 1)matrix for(0, 1), the whole matrix for()).dtypeDTypeThe dtype the sum is computed in, or
nullfor NumPy's default.outNDArrayA matrix of the kept shape (
(1, 1)without an axis), ornull.
Returns
- NDArray
A 0-d array (the scalar) when
axisisnull; otherwise the matrix of sum(int, DType, NDMatrix) as an NDArray reference.
Exceptions
- AxisError
axisis out of bounds.- ValueError
outis read-only or does not have the result's shape.- AttributeError
outis a plain (non-matrix) array: NumPy's reduction stores into it, then_collapsedoes not exist on it ('numpy.ndarray' object has no attribute '_collapse').
swapaxes(int, int)
The matrix with two axes exchanged (NumPy's m.swapaxes(a, b)).
public NDMatrix swapaxes(int axis1, int axis2)
Parameters
Returns
- NDMatrix
A view.
Exceptions
- AxisError
An axis is out of bounds.
take(NDArray, int?, string)
Elements selected by index along an axis, as a matrix (NumPy's m.take(indices, axis, mode)).
Without an axis the matrix is read flat and the result is a row.
public NDMatrix take(NDArray indices, int? axis = null, string mode = "raise")
Parameters
indicesNDArrayThe indices; a C# integer selects one element (a
(1, 1)result).axisint?The axis, or
nullto index the flattened matrix.modestring"raise","wrap"or"clip"for an index out of bounds.
Returns
- NDMatrix
A new matrix that owns its data.
Exceptions
- IndexError
An index is out of bounds under
"raise". Without an axis the text names axis 1 (index 12 is out of bounds for axis 1 with size 12): NumPy ravels the matrix to its(1, size)row first and takes along that row.- AxisError
axisis out of bounds.- ArgumentException
An index array of a non-integer dtype, or an unknown mode.
to_device(string, object)
Array-API device transfer: only "cpu" exists, and the matrix itself is returned.
public NDMatrix to_device(string device, object stream = null)
Parameters
Returns
- NDMatrix
This matrix.
Exceptions
- ArgumentNullException
deviceisnull.- ValueError
A stream, then another device: to_device(string, object)'s NumPy texts.
trace(int, int, int, DType)
The sum of the diagonal as a (1, 1) matrix (NumPy's m.trace(); np.trace(m)
is a scalar).
public NDMatrix trace(int offset = 0, int axis1 = 0, int axis2 = 1, DType dtype = null)
Parameters
offsetintThe diagonal's offset.
axis1intThe first axis of the plane.
axis2intThe second axis of the plane.
dtypeDTypeThe dtype of the sum, or
nullfor NumPy's default.
Returns
- NDMatrix
A new
(1, 1)matrix.
Exceptions
- AxisError
An axis is out of bounds.
- ArgumentException
The two axes are the same.
transpose(params int[])
The matrix with its axes permuted (NumPy's m.transpose(*axes)); without an argument the
transpose.
public NDMatrix transpose(params int[] axes)
Parameters
axesint[]NumPy's
*axes: the permutation as separate integers (m.transpose(1, 0)) or one array ({1, 0}or{0, 1}); no argument, an empty array ornullreverses the axes.
Returns
- NDMatrix
A view.
Remarks
One member, NumPy's call shapes: params makes m.transpose(1, 0) compile (the former
transpose(int[] premute = null) refused it). It hides transpose(params int[]), whose
parameter is the same array; the empty array reads as the no-argument call, as there.
Exceptions
- ValueError
The permutation does not have one entry per axis (
axes don't match array) or repeats an axis.- AxisError
An axis is out of bounds.
var(int, DType, NDMatrix, int)
NumPy's matrix.var(axis, dtype, out, ddof) for a given axis: the variances as a matrix
that keeps the reduced axis with length 1.
public NDMatrix var(int axis, DType dtype = null, NDMatrix @out = null, int ddof = 0)
Parameters
axisintThe axis to compute along.
dtypeDTypeThe dtype of the computation, or
null: float64 for an integer or bool matrix, the matrix's dtype for a float one.outNDMatrixA matrix of the result's shape to store into, or
null.ddofintDelta degrees of freedom: the divisor is
N - ddof.
Returns
- NDMatrix
The variances;
outitself when it is given.
Exceptions
- AxisError
axisis out of bounds.- ValueError
outis read-only or does not have the result's shape.
var(NDAxis?, DType, NDArray, int)
NumPy's matrix.var(axis=None, dtype=None, out=None, ddof=0); a scalar without an axis.
public NDArray var(NDAxis? axis = null, DType dtype = null, NDArray @out = null, int ddof = 0)
Parameters
axisNDAxis?The axis, or
nullfor the variance of every element. A tuple reduces the axes it names together and keeps them with length 1, as every axis but None does (the(1, 1)matrix for(0, 1), the whole matrix for()).dtypeDTypeThe dtype of the computation, or
nullfor NumPy's default.outNDArrayA matrix of the kept shape, or
null.ddofintDelta degrees of freedom.
Returns
- NDArray
A 0-d array when
axisisnull; otherwise the matrix.
Remarks
With out NumPy hands it to the LAST sum only (the sum of the squared deviations):
the mean and the deviations are computed as without it, that sum runs in
promote_types(out.dtype, deviation dtype) unless dtype is given, it is
stored into out with an unsafe cast, and out is then divided
in place by max(N - ddof, 0).
Exceptions
- AxisError
axisis out of bounds.- ValueError
outis read-only or does not have the result's shape.- AttributeError
outis a plain (non-matrix) array: NumPy's reduction stores into it, then_collapsedoes not exist on it ('numpy.ndarray' object has no attribute '_collapse').
view(DType)
A new matrix object over the same memory (NumPy's m.view()), optionally reading the
bytes as another dtype.
public NDMatrix view(DType dtype = null)
Parameters
dtypeDTypeThe dtype to read the bytes as, or
nullto keep it. A dtype of another size changes the number of columns.
Returns
- NDMatrix
A view; writes reach this matrix.
Exceptions
- ArgumentException
The new item size does not divide the last axis.
- InvalidOperationException
A dtype of another size over a non-contiguous last axis.
view(DType, Type)
NumPy's m.view(dtype, type): a view of this matrix as the array class type, its
bytes read as dtype.
public NDArray view(DType dtype, Type type)
Parameters
dtypeDTypeThe dtype to read the bytes as, or
nullto keep it.typeTypeThe array class, or
nullfor this matrix's own class.
Returns
- NDArray
A view of the class asked for.
Exceptions
- ValueError
typeis not an array class, the class's finalize rule refuses the dtype, or an itemsize error.- NotSupportedException
An array class with no NumPy counterpart.
view(Type)
NumPy's m.view(type): a view of this matrix as the array class type
(typeof(NDArray) is the plain 2-D array), or — when type is not an array class — read
as the dtype that CLR type names.
public NDArray view(Type type)
Parameters
typeTypeThe array class, or a CLR element type.
Returns
Remarks
Declared here because the matrix's own view(DType) would otherwise take a Type argument through its implicit conversion to DType.
Exceptions
- ValueError
The class's finalize rule refuses this matrix's dtype, or an itemsize error.
- NotSupportedException
An array class with no NumPy counterpart.
Operators
operator +(NDArray, NDMatrix)
Elementwise addition with broadcasting, returned as a matrix (NumPy's matrix + x). The dtype
follows the array operator of the same name.
public static NDMatrix operator +(NDArray left, NDMatrix right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator +(NDMatrix, NDArray)
Elementwise addition with broadcasting, returned as a matrix (NumPy's matrix + x). The dtype
follows the array operator of the same name.
public static NDMatrix operator +(NDMatrix left, NDArray right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator +(NDMatrix, NDMatrix)
Elementwise addition with broadcasting, returned as a matrix (NumPy's matrix + x). The dtype
follows the array operator of the same name.
public static NDMatrix operator +(NDMatrix left, NDMatrix right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator +(NDMatrix, object)
Elementwise addition with broadcasting, returned as a matrix (NumPy's matrix + x). The dtype
follows the array operator of the same name.
public static NDMatrix operator +(NDMatrix left, object right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator +(object, NDMatrix)
Elementwise addition with broadcasting, returned as a matrix (NumPy's matrix + x). The dtype
follows the array operator of the same name.
public static NDMatrix operator +(object left, NDMatrix right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator &(NDArray, NDMatrix)
Elementwise AND with broadcasting, returned as a matrix (NumPy's matrix & x;
logical for bool, bitwise for integers).
public static NDMatrix operator &(NDArray left, NDMatrix right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator &(NDMatrix, NDArray)
Elementwise AND with broadcasting, returned as a matrix (NumPy's matrix & x;
logical for bool, bitwise for integers).
public static NDMatrix operator &(NDMatrix left, NDArray right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator &(NDMatrix, NDMatrix)
Elementwise AND with broadcasting, returned as a matrix (NumPy's matrix & x;
logical for bool, bitwise for integers).
public static NDMatrix operator &(NDMatrix left, NDMatrix right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator &(NDMatrix, object)
Elementwise AND with broadcasting, returned as a matrix (NumPy's matrix & x;
logical for bool, bitwise for integers).
public static NDMatrix operator &(NDMatrix left, object right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator &(object, NDMatrix)
Elementwise AND with broadcasting, returned as a matrix (NumPy's matrix & x;
logical for bool, bitwise for integers).
public static NDMatrix operator &(object left, NDMatrix right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator |(NDArray, NDMatrix)
Elementwise OR with broadcasting, returned as a matrix (NumPy's matrix | x).
public static NDMatrix operator |(NDArray left, NDMatrix right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator |(NDMatrix, NDArray)
Elementwise OR with broadcasting, returned as a matrix (NumPy's matrix | x).
public static NDMatrix operator |(NDMatrix left, NDArray right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator |(NDMatrix, NDMatrix)
Elementwise OR with broadcasting, returned as a matrix (NumPy's matrix | x).
public static NDMatrix operator |(NDMatrix left, NDMatrix right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator |(NDMatrix, object)
Elementwise OR with broadcasting, returned as a matrix (NumPy's matrix | x).
public static NDMatrix operator |(NDMatrix left, object right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator |(object, NDMatrix)
Elementwise OR with broadcasting, returned as a matrix (NumPy's matrix | x).
public static NDMatrix operator |(object left, NDMatrix right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator /(NDArray, NDMatrix)
Elementwise TRUE division with broadcasting, returned as a matrix (NumPy's matrix / x).
Integer operands give float64. It is not a multiplication by an inverse: use m * x.I for that.
public static NDMatrix operator /(NDArray left, NDMatrix right)
Parameters
Returns
- NDMatrix
The quotient as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator /(NDMatrix, NDArray)
Elementwise TRUE division with broadcasting, returned as a matrix (NumPy's matrix / x).
Integer operands give float64. It is not a multiplication by an inverse: use m * x.I for that.
public static NDMatrix operator /(NDMatrix left, NDArray right)
Parameters
Returns
- NDMatrix
The quotient as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator /(NDMatrix, NDMatrix)
Elementwise TRUE division with broadcasting, returned as a matrix (NumPy's matrix / x).
Integer operands give float64. It is not a multiplication by an inverse: use m * x.I for that.
public static NDMatrix operator /(NDMatrix left, NDMatrix right)
Parameters
Returns
- NDMatrix
The quotient as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator /(NDMatrix, object)
Elementwise TRUE division with broadcasting, returned as a matrix (NumPy's matrix / x).
Integer operands give float64. It is not a multiplication by an inverse: use m * x.I for that.
public static NDMatrix operator /(NDMatrix left, object right)
Parameters
Returns
- NDMatrix
The quotient as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator /(object, NDMatrix)
Elementwise TRUE division with broadcasting, returned as a matrix (NumPy's matrix / x).
Integer operands give float64. It is not a multiplication by an inverse: use m * x.I for that.
public static NDMatrix operator /(object left, NDMatrix right)
Parameters
Returns
- NDMatrix
The quotient as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator ==(NDArray, NDMatrix)
Elementwise equality with broadcasting as a BOOL MATRIX (NumPy's matrix == x). It is not a
reference or a whole-matrix test: use is null for a null check, Equals(object)
or np.array_equal for "all elements equal".
public static NDMatrix operator ==(NDArray left, NDMatrix right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array). A comparison with
nullis a 0-d bool.
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator ==(NDMatrix, NDArray)
Elementwise equality with broadcasting as a BOOL MATRIX (NumPy's matrix == x). It is not a
reference or a whole-matrix test: use is null for a null check, Equals(object)
or np.array_equal for "all elements equal".
public static NDMatrix operator ==(NDMatrix left, NDArray right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array). A comparison with
nullis a 0-d bool.
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator ==(NDMatrix, NDMatrix)
Elementwise equality with broadcasting as a BOOL MATRIX (NumPy's matrix == x). It is not a
reference or a whole-matrix test: use is null for a null check, Equals(object)
or np.array_equal for "all elements equal".
public static NDMatrix operator ==(NDMatrix left, NDMatrix right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array). A comparison with
nullis a 0-d bool.
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator ==(NDMatrix, object)
Elementwise equality with broadcasting as a BOOL MATRIX (NumPy's matrix == x). It is not a
reference or a whole-matrix test: use is null for a null check, Equals(object)
or np.array_equal for "all elements equal".
public static NDMatrix operator ==(NDMatrix left, object right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array). A comparison with
nullis a 0-d bool.
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator ==(object, NDMatrix)
Elementwise equality with broadcasting as a BOOL MATRIX (NumPy's matrix == x). It is not a
reference or a whole-matrix test: use is null for a null check, Equals(object)
or np.array_equal for "all elements equal".
public static NDMatrix operator ==(object left, NDMatrix right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array). A comparison with
nullis a 0-d bool.
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator ^(NDArray, NDMatrix)
Elementwise XOR with broadcasting, returned as a matrix (NumPy's matrix ^ x). It is NOT
a power: Python's ** is __pow__(int).
public static NDMatrix operator ^(NDArray left, NDMatrix right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator ^(NDMatrix, NDArray)
Elementwise XOR with broadcasting, returned as a matrix (NumPy's matrix ^ x). It is NOT
a power: Python's ** is __pow__(int).
public static NDMatrix operator ^(NDMatrix left, NDArray right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator ^(NDMatrix, NDMatrix)
Elementwise XOR with broadcasting, returned as a matrix (NumPy's matrix ^ x). It is NOT
a power: Python's ** is __pow__(int).
public static NDMatrix operator ^(NDMatrix left, NDMatrix right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator ^(NDMatrix, object)
Elementwise XOR with broadcasting, returned as a matrix (NumPy's matrix ^ x). It is NOT
a power: Python's ** is __pow__(int).
public static NDMatrix operator ^(NDMatrix left, object right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator ^(object, NDMatrix)
Elementwise XOR with broadcasting, returned as a matrix (NumPy's matrix ^ x). It is NOT
a power: Python's ** is __pow__(int).
public static NDMatrix operator ^(object left, NDMatrix right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator >(NDArray, NDMatrix)
Elementwise > with broadcasting as a bool matrix (NumPy's matrix > x). The
result indexes a matrix directly: m[m > 5].
public static NDMatrix operator >(NDArray left, NDMatrix right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator >(NDMatrix, NDArray)
Elementwise > with broadcasting as a bool matrix (NumPy's matrix > x). The
result indexes a matrix directly: m[m > 5].
public static NDMatrix operator >(NDMatrix left, NDArray right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator >(NDMatrix, NDMatrix)
Elementwise > with broadcasting as a bool matrix (NumPy's matrix > x). The
result indexes a matrix directly: m[m > 5].
public static NDMatrix operator >(NDMatrix left, NDMatrix right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator >(NDMatrix, object)
Elementwise > with broadcasting as a bool matrix (NumPy's matrix > x). The
result indexes a matrix directly: m[m > 5].
public static NDMatrix operator >(NDMatrix left, object right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator >(object, NDMatrix)
Elementwise > with broadcasting as a bool matrix (NumPy's matrix > x). The
result indexes a matrix directly: m[m > 5].
public static NDMatrix operator >(object left, NDMatrix right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator >=(NDArray, NDMatrix)
Elementwise >= with broadcasting as a bool matrix (NumPy's matrix >= x).
public static NDMatrix operator >=(NDArray left, NDMatrix right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator >=(NDMatrix, NDArray)
Elementwise >= with broadcasting as a bool matrix (NumPy's matrix >= x).
public static NDMatrix operator >=(NDMatrix left, NDArray right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator >=(NDMatrix, NDMatrix)
Elementwise >= with broadcasting as a bool matrix (NumPy's matrix >= x).
public static NDMatrix operator >=(NDMatrix left, NDMatrix right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator >=(NDMatrix, object)
Elementwise >= with broadcasting as a bool matrix (NumPy's matrix >= x).
public static NDMatrix operator >=(NDMatrix left, object right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator >=(object, NDMatrix)
Elementwise >= with broadcasting as a bool matrix (NumPy's matrix >= x).
public static NDMatrix operator >=(object left, NDMatrix right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
implicit operator NDMatrix(bool)
A C# bool as a 0-d matrix scalar (NumPy's Python bool operand of a matrix
expression). It is NOT np.matrix(value), which is (1, 1).
public static implicit operator NDMatrix(bool value)
Parameters
valueboolThe value.
Returns
implicit operator NDMatrix(byte)
A C# bool as a 0-d matrix scalar (NumPy's Python bool operand of a matrix
expression). It is NOT np.matrix(value), which is (1, 1).
public static implicit operator NDMatrix(byte value)
Parameters
valuebyteThe value.
Returns
implicit operator NDMatrix(char)
A C# bool as a 0-d matrix scalar (NumPy's Python bool operand of a matrix
expression). It is NOT np.matrix(value), which is (1, 1).
public static implicit operator NDMatrix(char value)
Parameters
valuecharThe value.
Returns
implicit operator NDMatrix(decimal)
A C# bool as a 0-d matrix scalar (NumPy's Python bool operand of a matrix
expression). It is NOT np.matrix(value), which is (1, 1).
public static implicit operator NDMatrix(decimal value)
Parameters
valuedecimalThe value.
Returns
implicit operator NDMatrix(double)
A C# bool as a 0-d matrix scalar (NumPy's Python bool operand of a matrix
expression). It is NOT np.matrix(value), which is (1, 1).
public static implicit operator NDMatrix(double value)
Parameters
valuedoubleThe value.
Returns
implicit operator NDMatrix(Half)
A C# bool as a 0-d matrix scalar (NumPy's Python bool operand of a matrix
expression). It is NOT np.matrix(value), which is (1, 1).
public static implicit operator NDMatrix(Half value)
Parameters
valueHalfThe value.
Returns
implicit operator NDMatrix(short)
A C# bool as a 0-d matrix scalar (NumPy's Python bool operand of a matrix
expression). It is NOT np.matrix(value), which is (1, 1).
public static implicit operator NDMatrix(short value)
Parameters
valueshortThe value.
Returns
implicit operator NDMatrix(int)
A C# bool as a 0-d matrix scalar (NumPy's Python bool operand of a matrix
expression). It is NOT np.matrix(value), which is (1, 1).
public static implicit operator NDMatrix(int value)
Parameters
valueintThe value.
Returns
implicit operator NDMatrix(long)
A C# bool as a 0-d matrix scalar (NumPy's Python bool operand of a matrix
expression). It is NOT np.matrix(value), which is (1, 1).
public static implicit operator NDMatrix(long value)
Parameters
valuelongThe value.
Returns
implicit operator NDMatrix(Complex)
A C# bool as a 0-d matrix scalar (NumPy's Python bool operand of a matrix
expression). It is NOT np.matrix(value), which is (1, 1).
public static implicit operator NDMatrix(Complex value)
Parameters
valueComplexThe value.
Returns
implicit operator NDMatrix(sbyte)
A C# bool as a 0-d matrix scalar (NumPy's Python bool operand of a matrix
expression). It is NOT np.matrix(value), which is (1, 1).
public static implicit operator NDMatrix(sbyte value)
Parameters
valuesbyteThe value.
Returns
implicit operator NDMatrix(float)
A C# bool as a 0-d matrix scalar (NumPy's Python bool operand of a matrix
expression). It is NOT np.matrix(value), which is (1, 1).
public static implicit operator NDMatrix(float value)
Parameters
valuefloatThe value.
Returns
implicit operator NDMatrix(ushort)
A C# bool as a 0-d matrix scalar (NumPy's Python bool operand of a matrix
expression). It is NOT np.matrix(value), which is (1, 1).
public static implicit operator NDMatrix(ushort value)
Parameters
valueushortThe value.
Returns
implicit operator NDMatrix(uint)
A C# bool as a 0-d matrix scalar (NumPy's Python bool operand of a matrix
expression). It is NOT np.matrix(value), which is (1, 1).
public static implicit operator NDMatrix(uint value)
Parameters
valueuintThe value.
Returns
implicit operator NDMatrix(ulong)
A C# bool as a 0-d matrix scalar (NumPy's Python bool operand of a matrix
expression). It is NOT np.matrix(value), which is (1, 1).
public static implicit operator NDMatrix(ulong value)
Parameters
valueulongThe value.
Returns
operator !=(NDArray, NDMatrix)
Elementwise inequality with broadcasting as a bool matrix (NumPy's matrix != x).
public static NDMatrix operator !=(NDArray left, NDMatrix right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator !=(NDMatrix, NDArray)
Elementwise inequality with broadcasting as a bool matrix (NumPy's matrix != x).
public static NDMatrix operator !=(NDMatrix left, NDArray right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator !=(NDMatrix, NDMatrix)
Elementwise inequality with broadcasting as a bool matrix (NumPy's matrix != x).
public static NDMatrix operator !=(NDMatrix left, NDMatrix right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator !=(NDMatrix, object)
Elementwise inequality with broadcasting as a bool matrix (NumPy's matrix != x).
public static NDMatrix operator !=(NDMatrix left, object right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator !=(object, NDMatrix)
Elementwise inequality with broadcasting as a bool matrix (NumPy's matrix != x).
public static NDMatrix operator !=(object left, NDMatrix right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator <<(NDMatrix, NDArray)
Elementwise left shift with broadcasting, returned as a matrix (NumPy's matrix << x).
C# requires the left operand of a shift operator to be the declaring type, so there is no
(array, matrix) form: write np.asmatrix(np.left_shift(a, m)).
public static NDMatrix operator <<(NDMatrix left, NDArray right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator <<(NDMatrix, NDMatrix)
Elementwise left shift with broadcasting, returned as a matrix (NumPy's matrix << x).
C# requires the left operand of a shift operator to be the declaring type, so there is no
(array, matrix) form: write np.asmatrix(np.left_shift(a, m)).
public static NDMatrix operator <<(NDMatrix left, NDMatrix right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator <<(NDMatrix, object)
Elementwise left shift with broadcasting, returned as a matrix (NumPy's matrix << x).
C# requires the left operand of a shift operator to be the declaring type, so there is no
(array, matrix) form: write np.asmatrix(np.left_shift(a, m)).
public static NDMatrix operator <<(NDMatrix left, object right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator <(NDArray, NDMatrix)
Elementwise < with broadcasting as a bool matrix (NumPy's matrix < x).
public static NDMatrix operator <(NDArray left, NDMatrix right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator <(NDMatrix, NDArray)
Elementwise < with broadcasting as a bool matrix (NumPy's matrix < x).
public static NDMatrix operator <(NDMatrix left, NDArray right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator <(NDMatrix, NDMatrix)
Elementwise < with broadcasting as a bool matrix (NumPy's matrix < x).
public static NDMatrix operator <(NDMatrix left, NDMatrix right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator <(NDMatrix, object)
Elementwise < with broadcasting as a bool matrix (NumPy's matrix < x).
public static NDMatrix operator <(NDMatrix left, object right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator <(object, NDMatrix)
Elementwise < with broadcasting as a bool matrix (NumPy's matrix < x).
public static NDMatrix operator <(object left, NDMatrix right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator <=(NDArray, NDMatrix)
Elementwise <= with broadcasting as a bool matrix (NumPy's matrix <= x).
public static NDMatrix operator <=(NDArray left, NDMatrix right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator <=(NDMatrix, NDArray)
Elementwise <= with broadcasting as a bool matrix (NumPy's matrix <= x).
public static NDMatrix operator <=(NDMatrix left, NDArray right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator <=(NDMatrix, NDMatrix)
Elementwise <= with broadcasting as a bool matrix (NumPy's matrix <= x).
public static NDMatrix operator <=(NDMatrix left, NDMatrix right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator <=(NDMatrix, object)
Elementwise <= with broadcasting as a bool matrix (NumPy's matrix <= x).
public static NDMatrix operator <=(NDMatrix left, object right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator <=(object, NDMatrix)
Elementwise <= with broadcasting as a bool matrix (NumPy's matrix <= x).
public static NDMatrix operator <=(object left, NDMatrix right)
Parameters
Returns
- NDMatrix
A bool matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator !(NDMatrix)
Elementwise logical NOT as a bool matrix (NumSharp's !array, NumPy's
np.logical_not).
public static NDMatrix operator !(NDMatrix operand)
Parameters
operandNDMatrixThe matrix.
Returns
- NDMatrix
A bool matrix (a view of the computed array).
operator %(NDArray, NDMatrix)
Elementwise remainder with broadcasting, returned as a matrix (NumPy's matrix % x; the
result takes the divisor's sign).
public static NDMatrix operator %(NDArray left, NDMatrix right)
Parameters
Returns
- NDMatrix
The remainder as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator %(NDMatrix, NDArray)
Elementwise remainder with broadcasting, returned as a matrix (NumPy's matrix % x; the
result takes the divisor's sign).
public static NDMatrix operator %(NDMatrix left, NDArray right)
Parameters
Returns
- NDMatrix
The remainder as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator %(NDMatrix, NDMatrix)
Elementwise remainder with broadcasting, returned as a matrix (NumPy's matrix % x; the
result takes the divisor's sign).
public static NDMatrix operator %(NDMatrix left, NDMatrix right)
Parameters
Returns
- NDMatrix
The remainder as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator %(NDMatrix, object)
Elementwise remainder with broadcasting, returned as a matrix (NumPy's matrix % x; the
result takes the divisor's sign).
public static NDMatrix operator %(NDMatrix left, object right)
Parameters
Returns
- NDMatrix
The remainder as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator %(object, NDMatrix)
Elementwise remainder with broadcasting, returned as a matrix (NumPy's matrix % x; the
result takes the divisor's sign).
public static NDMatrix operator %(object left, NDMatrix right)
Parameters
Returns
- NDMatrix
The remainder as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator *(NDArray, NDMatrix)
NumPy's ndarray * matrix (matrix.rmul): np.dot(left, right) with the array as
it is, so a 1-d left of length M times an (M, K) matrix is a (1, K)
row.
public static NDMatrix operator *(NDArray left, NDMatrix right)
Parameters
Returns
- NDMatrix
The product as a new matrix.
Exceptions
- ArgumentNullException
An operand is
null.- ArgumentException
The inner dimensions differ.
- ValueError
The product has more than two axes longer than 1.
operator *(NDMatrix, NDArray)
NumPy's matrix * ndarray: np.dot(left, asmatrix(right)). A 1-d right
becomes a ROW (1, N) first, so the product only exists when the matrix has one column.
public static NDMatrix operator *(NDMatrix left, NDArray right)
Parameters
Returns
- NDMatrix
The product as a new matrix.
Exceptions
- ArgumentNullException
An operand is
null.- ArgumentException
The inner dimensions differ.
- ValueError
righthas more than two axes longer than 1.
operator *(NDMatrix, NDMatrix)
NumPy's matrix * matrix: the MATRIX product (np.dot), not the elementwise one. A 0-d
operand (a C# number) scales. For the elementwise product use np.multiply(a, b).
public static NDMatrix operator *(NDMatrix left, NDMatrix right)
Parameters
Returns
- NDMatrix
A new matrix of shape
(left rows, right columns)in the promoted dtype.
Exceptions
- ArgumentNullException
An operand is
null.- ArgumentException
The inner dimensions differ.
- ValueError
An array operand cannot be seen as a matrix.
operator *(NDMatrix, object)
NumPy's matrix * other for a list, a tuple, a C# array or a boxed number; the operand is
converted as matrix(object, DType, bool) converts its data.
public static NDMatrix operator *(NDMatrix left, object right)
Parameters
Returns
- NDMatrix
The product as a new matrix.
Exceptions
- ArgumentNullException
An operand is
null.- ArgumentException
The inner dimensions differ.
- ValueError
The operand is ragged or cannot be seen as a matrix.
- NotSupportedException
The operand holds strings or other objects.
operator *(object, NDMatrix)
NumPy's other * matrix for a list, a tuple, a C# array or a boxed number.
public static NDMatrix operator *(object left, NDMatrix right)
Parameters
Returns
- NDMatrix
The product as a new matrix.
Exceptions
- ArgumentNullException
An operand is
null.- ArgumentException
The inner dimensions differ.
- ValueError
The operand is ragged, or the product cannot be a matrix.
- NotSupportedException
The operand holds strings or other objects.
operator ~(NDMatrix)
Elementwise bitwise NOT, returned as a matrix (NumPy's ~matrix; logical NOT for bool).
public static NDMatrix operator ~(NDMatrix operand)
Parameters
operandNDMatrixThe matrix.
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
operator >>(NDMatrix, NDArray)
Elementwise right shift with broadcasting, returned as a matrix (NumPy's matrix >> x;
arithmetic for signed integers).
public static NDMatrix operator >>(NDMatrix left, NDArray right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator >>(NDMatrix, NDMatrix)
Elementwise right shift with broadcasting, returned as a matrix (NumPy's matrix >> x;
arithmetic for signed integers).
public static NDMatrix operator >>(NDMatrix left, NDMatrix right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator >>(NDMatrix, object)
Elementwise right shift with broadcasting, returned as a matrix (NumPy's matrix >> x;
arithmetic for signed integers).
public static NDMatrix operator >>(NDMatrix left, object right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator -(NDArray, NDMatrix)
Elementwise subtraction with broadcasting, returned as a matrix (NumPy's matrix - x).
public static NDMatrix operator -(NDArray left, NDMatrix right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator -(NDMatrix, NDArray)
Elementwise subtraction with broadcasting, returned as a matrix (NumPy's matrix - x).
public static NDMatrix operator -(NDMatrix left, NDArray right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator -(NDMatrix, NDMatrix)
Elementwise subtraction with broadcasting, returned as a matrix (NumPy's matrix - x).
public static NDMatrix operator -(NDMatrix left, NDMatrix right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator -(NDMatrix, object)
Elementwise subtraction with broadcasting, returned as a matrix (NumPy's matrix - x).
public static NDMatrix operator -(NDMatrix left, object right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator -(object, NDMatrix)
Elementwise subtraction with broadcasting, returned as a matrix (NumPy's matrix - x).
public static NDMatrix operator -(object left, NDMatrix right)
Parameters
Returns
- NDMatrix
The result as a matrix (a view of the computed array).
Exceptions
- IncorrectShapeException
The operands do not broadcast.
- ValueError
The broadcast result has more than two axes longer than 1.
operator -(NDMatrix)
Elementwise negation, returned as a matrix (NumPy's -matrix).
public static NDMatrix operator -(NDMatrix operand)
Parameters
operandNDMatrixThe matrix.
Returns
- NDMatrix
The negated matrix (a view of the computed array).
operator +(NDMatrix)
NumPy's +matrix: the np.positive ufunc, a copy returned as a matrix.
public static NDMatrix operator +(NDMatrix operand)
Parameters
operandNDMatrixThe matrix.
Returns
- NDMatrix
A new matrix with the same elements (a view of the computed array).
Exceptions
- TypeError
The matrix is bool:
np.positivehas no bool loop, as in NumPy (ufunc 'positive' did not contain a loop with signature matching types ...).