Skip to content

Commit

Permalink
add batch poly zono and batch mat poly zono doc
Browse files Browse the repository at this point in the history
  • Loading branch information
@BuildingAtom
BuildingAtom committed Feb 13, 2024
1 parent 1a7d86d commit 4d88f2b
Show file tree
Hide file tree
Showing 2 changed files with 63 additions and 47 deletions.
55 changes: 33 additions & 22 deletions zonopy/contset/polynomial_zonotope/batch_mat_poly_zono.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -12,32 +12,43 @@
_matmul_genmpz_impl,
)

class batchMatPolyZonotope():
'''
<matPolyZonotope>
class batchMatPolyZonotope:
r''' Batched 2D Matrix Polynomial Zonotopes
c: <torch.Tensor> center maxtrix of the matrix polyonmial zonotope
, shape: [nx,ny]
G: <torch.tensor> generator tensor containing the dependent generators
, shape: [N, nx, ny]
Grest: <torch.Tensor> generator tensor containing the independent generators
, shape: [M, nx, ny]
expMat: <troch.Tensor> matrix containing the exponents for the dependent generators
, shape: [N, p]
id: <torch.Tensor> vector containing the integer identifiers for the dependent factors
, shape: [p]
Batched form of the :class:`matPolyZonotope` class.
This class is designed to represent a batch of 2D matrix polynomial zonotopes over the same domain
with arbitrary batch dimensions, extending the concept of polynomial zonotopes to matrix forms and allowing for batch processing.
It adheres to a formulation similar to the :class:`polyZonotope` and :class:`batchPolyZonotope` classes,
enabling efficient representation and manipulation of sets of matrix polynomial zonotopes.
Eq. (coeff. a1,a2,...,aN; b1,b2,...,bp \in [0,1])
G = [[Gd1],[Gd2],...,[GdN]]
Grest = [[Gi1],[Gi2],...,[GiM]]
(Gd1,Gd2,...,GdN,Gi1,Gi2,...,GiM \in R^(nx,ny), matrix)
expMat = [[i11,i12,...,i1p],[i21,i22,...,i2p],...,[iN1,iN2,...,iNp]]
id = [0,1,2,...,p-1]
This results in a :math:`\mathbf{Z} \in \mathbb{R}^{B_1 \times B_2 \times \cdots \times B_b \times (N+M+1) \times dx \times dy}` tensor,
where :math:`dx` and :math:`dy` are the dimensions of the matrices.
pZ = c + a1*Gi1 + a2*Gi2 + ... + aN*GiN + b1^i11*b2^i21*...*bp^ip1*Gd1 + b1^i12*b2^i22*...*bp^ip2*Gd2 + ...
+ b1^i1M*b2^i2M*...*bp^ipM*GdM
Refer to the :class:`matPolyZonotope` class for more detailed information on matrix polynomial zonotopes.
'''
def __init__(self,Z,n_dep_gens=0,expMat=None,id=None,copy_Z=True, device=None, dtype=None):
def __init__(self, Z, n_dep_gens=0, expMat=None, id=None, copy_Z=True, dtype=None, device=None):
r''' Constructor for the batchMatPolyZonotope class
Initializes a batch of 2D matrix polynomial zonotopes with specified characteristics.
Args:
Z (torch.Tensor): The center and generator tensor for the batch of matrix polynomial zonotopes.
The shape of Z should be :math:`(B_1, B_2, \cdots, B_b, N+M+1, dx, dy)`, where :math:`B_1, B_2, \cdots, B_b` are the batch dimensions,
:math:`N` is the number of dependent generators, :math:`M` is the number of independent generators,
and :math:`dx, dy` are the dimensions of the matrices.
n_dep_gens (int, optional): The number of dependent generators in the matrix polynomial zonotope. Default: 0.
expMat (torch.Tensor, optional): The exponent matrix for the dependent generators. If `None`, it will be assumed to be the identity matrix. Default: None.
id (torch.Tensor, optional): The integer identifiers for the dependent generators. If `None`, it will be the range of the number of dependent generators. Default: None.
copy_Z (bool, optional): If `True`, a copy of the input `Z` tensor will be made. Default: `True`.
dtype (torch.dtype, optional): The data type for the elements of the matrix polynomial zonotope. If `None`, it will be inferred from `Z`. Default: `None`.
device (torch.device, optional): The device on which the matrix polynomial zonotope is to be stored. If `None`, it will be inferred from `Z`. Default: `None`.
Raises:
AssertionError: If the dimension of the `Z` input is not 5 or more, indicating incorrect batch dimensions or matrix dimensions.
AssertionError: If the exponent matrix does not seem to be valid for the given number of dependent generators or ids.
AssertionError: If the number of dependent generators does not match the number of ids provided.
AssertionError: If the exponent matrix is not a non-negative integer matrix.
'''
# If compress=2, it will always copy.

# Make sure Z is a tensor and shaped right
Expand Down
55 changes: 30 additions & 25 deletions zonopy/contset/polynomial_zonotope/batch_poly_zono.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -19,34 +19,39 @@
import zonopy.internal as zpi

class batchPolyZonotope:
'''
pZ: <polyZonotope>
c: <torch.Tensor> center of the polyonmial zonotope
, shape: [B1, B2, .. , Bb, nx]
G: <torch.Tensor> generator matrix containing the dependent generators
, shape: [B1, B2, .. , Bb, N, nx]
Grest: <torch.Tensor> generator matrix containing the independent generators
, shape: [B1, B2, .. , Bb, M, nx]
expMat: <troch.Tensor> matrix containing the exponents for the dependent generators
, shape: [N, p]
id: <torch.Tensor> vector containing the integer identifiers for the dependent factors
, shape: [p]
compress: <int> level for compressing dependent generators with expodent
0: no compress, 1: compress zero dependent generators, 2: compress zero dependent generators and remove redundant expodent
Eq. (coeff. a1,a2,...,aN; b1,b2,...,bp \in [0,1])
G = [gd1,gd2,...,gdN]
Grest = [gi1,gi2,...,giM]
expMat = [[i11,i12,...,i1N],[i21,i22,...,i2N],...,[ip1,ip2,...,ipN]]
id = [0,1,2,...,p-1]
pZ = c + a1*gi1 + a2*gi2 + ... + aN*giN + b1^i11*b2^i21*...*bp^ip1*gd1 + b1^i12*b2^i22*...*bp^ip2*gd2 + ...
+ b1^i1M*b2^i2M*...*bp^ipM*gdM
r''' Batched 1D polynomial zonotope
Batched form of the :class:`polyZonotope` class.
This class is used to represent a batch of polynomial zonotopes over the same domain
with arbitrary batch dimensions.
It follows a similar formulation from the :class:`polyZonotope` class as the
:class:`batchZonotope` class did from :class:`zonotope`.
This results in a :math:`\mathbf{Z} \in \mathbb{R}^{B_1 \times B_2 \times \cdots \times B_b \times (N+M+1) \times d}` tensor
Refer to the :class:`polyZonotope` class for more information polynomial zonotops.
'''
# NOTE: property for mat pz
def __init__(self,Z,n_dep_gens=0,expMat=None,id=None,copy_Z=True, dtype=None, device=None):
r''' Constructor for the batchPolyZonotope class
Args:
Z (torch.Tensor): The center and generator matrix of the polynomial zonotope.
The shape of Z should be :math:`(B_1, B_2, \cdots, B_b, N+M+1, d)` where :math:`B_1, B_2, \cdots, B_b` are the batch dimensions,
:math:`N` is the number of dependent generators, :math:`M` is the number of independent generators, and :math:`d` is the dimension of the zonotope.
n_dep_gens (int, optional): The number of dependent generators in the polynomial zonotope. Default is 0.
expMat (torch.Tensor, optional): The exponent matrix of the dependent generators. If ``None``, it will be the identity matrix. Default: None
id (torch.Tensor, optional): The integer identifiers for the dependent generators. If ``None``, it will be the range of the number of dependent generators. Default: None
copy_Z (bool, optional): If ``True``, it will copy the input ``Z`` value. Default: ``True``
dtype (torch.dtype, optional): The data type of the polynomial zonotope. If ``None``, it will be inferred. Default: ``None``
device (torch.device, optional): The device of the polynomial zonotope. If ``None``, it will be inferred. Default: ``None``
Raises:
AssertionError: If the dimension of Z input is not 3 or more.
AssertionError: If the exponent matrix does not seem to be valid for the given dependent generators or ids.
AssertionError: If the number of dependent generators does not match the number of ids.
AssertionError: If the exponent matrix is not a non-negative integer matrix.
'''
# If compress=2, it will always copy.

# Make sure Z is a tensor and shaped right
Expand Down

0 comments on commit 4d88f2b

Please sign in to comment.