HMatrix¶

class
HMatrix
(*args)¶ Hierarchical matrices.
Hierarchical matrices (or HMatrix) are a compressed representation of dense matrices. In many applications, matrix coefficients represent an interaction between two degrees of freedom; when these interactions are smooth, it is possible to approximate subblocks by a local lowrank approximation B =~ UV^T where B has dimension (m,n), U (m,k), and V (n,k). Of course, this is interesting only if k is much lower than m and n.
In order to obtain this compressed representation, several different steps must be performed:
 Clustering: creation of rows and columns cluster trees
Vertices where interactions are computed are reordered to improve locality.
A binary space partition algorithm is used to recursively divide vertex set.
Root cell contains all vertices. At each recursion step, a cell is divided
into two new cells until it contains less than a given number of vertices.
Space partition is performed orthogonally to original axis, by cutting its
longest dimension.
 The ‘median’ clustering algorithm divides a cell into two cells containing the same number of degrees of freedom.
 The ‘geometric’ clustering algorithm divides a cell into two cells of the same geometric size
 The ‘hybrid’ clustering algorithm is a mix. It first performs a ‘median’ bisection; if volumes of these new cells are very different, a ‘geometric’ clustering is performed instead.
 Admissibility: creation of an empty HMatrix structure The first step created a full binary tree for rows and columns degrees of freedom. We will now create a hierarchical representation of our matrix by checking whether some blocks can be replaced by lowrank approximations. The whole matrix represents the interactions of all rows degrees of freedom against all columns degrees of freedom. It can not be approximated by a lowrank approximation, and thus it is replaced by 4 blocks obtained by considering interactions between rows and columns children nodes. This operation is performed recursively. At each step, we compute axis aligned bounding boxes rows_bbox and cols_bbox: if min(diameter(rows_bbox), diameter(cols_bbox)) <= eta*distance(rows_bbox, cols_bbox) then we consider that interaction between rows and columns degrees of freedom can have a local lowrank approximation, and recursion is stopped. Otherwise, we recurse until bottom cluster tree is reached. The whole matrix is thus represented by a 4tree where leaves will contain either lowrank approximation or full blocks. The eta parameter is called the admissibility factor, and it can be modified.
 Assembly: coefficients computations The hierarchical structure of the matrix has been computed during step 2. To compute coefficients, we call the assembleReal method and provide a callable to compute interaction between two nodes. Full blocks are computed by calling this callable for the whole block. If compression method is ‘SVD’, lowrank approximation is computed by first computing the whole block, then finding its singular value decomposition, and rank is truncated so that error does not exceed assemblyEpsilon. This method is precise, but very costly. If compression method is a variant of ACA, only few rows and columns are computed. This is much more efficient, but error may be larger than expected on some problems.
 Matrix computations Once an HMatrix is computed, usual linear algebra operations can be performed. Matrix can be factorized inplace, in order to solve systems. Or we can compute its product by a matrix or vector. But keep in mind that rows and columns are reordered internally, and thus results may differ sensibly from standard dense representation (for instance when computing a Cholesky or LU decomposition).
See also
Methods
assembleReal
(callable, symmetry)Assemble matrix. assembleTensor
(callable, outputDimension, …)Assemble matrix by block. compressionRatio
()Compression ratio accessor. dump
(name)Save matrix to a file. factorize
(method)Factorize matrix. fullrkRatio
()Block ratio accessor. gemm
(transA, transB, alpha, a, b, beta)Multiply matrix inplace self=alpha*op(A)*op(B)+beta*self. gemv
(trans, alpha, x, beta, y)Multiply vector inplace y=alpha*op(A)*x+beta*y. getClassName
()Accessor to the object’s name. getDiagonal
()Diagonal values accessor. getId
()Accessor to the object’s id. getImplementation
(*args)Accessor to the underlying implementation. getName
()Accessor to the object’s name. getNbColumns
()Accessor to the number of columns. getNbRows
()Accessor to the number of rows. norm
()Compute norm value. scale
(alpha)Scale matrix inplace A=alpha*A. setName
(name)Accessor to the object’s name. solve
(*args)Solve linear system op(A)*x=b, after A has been factorized. solveLower
(*args)Solve lower linear system op(L)*x=b, after A has been factorized. transpose
()Transpose matrix inplace. 
__init__
(*args)¶

assembleReal
(callable, symmetry)¶ Assemble matrix.
Parameters: f : assembly function
Callable that takes i,j int parameters and returns a float
symmetry : str
Symmetry flag, either N or L

assembleTensor
(callable, outputDimension, symmetry)¶ Assemble matrix by block.
Parameters: f : assembly function
Callable that takes i,j int parameters and returns a Matrix
outputDimension : int
Block dimension
symmetry : str
Symmetry flag, either N or L

compressionRatio
()¶ Compression ratio accessor.
Returns: ratio : 2tuple of int
Numbers of elements in the compressed and uncompressed forms.

dump
(name)¶ Save matrix to a file.
Parameters: fileName : str
File name to save to.

factorize
(method)¶ Factorize matrix.
Parameters: method : str
Factorization method, either one of: LDLt, LLt or LU

fullrkRatio
()¶ Block ratio accessor.
Returns: ratio : 2tuple of int
Numbers of elements in full blocks and low rank blocks.

gemm
(transA, transB, alpha, a, b, beta)¶ Multiply matrix inplace self=alpha*op(A)*op(B)+beta*self.
Parameters: transA : str
Whether to use A or A^t: either N or T.
transB : str
Whether to use B or B^t: either N or T.
alpha : float
Coefficient
a :
HMatrix
Multiplied matrix A.
b :
HMatrix
Multiplied matrix B.
beta : float
Coefficient.

gemv
(trans, alpha, x, beta, y)¶ Multiply vector inplace y=alpha*op(A)*x+beta*y.
Parameters: trans : str
Whether to use A or A^t: either N or T.
alpha : float
Coefficient
x : sequence of float
Vector to multiply.
beta : float
Coefficient.
y :
Point
Vector multiplied inplace.

getClassName
()¶ Accessor to the object’s name.
Returns: class_name : str
The object class name (object.__class__.__name__).

getId
()¶ Accessor to the object’s id.
Returns: id : int
Internal unique identifier.

getImplementation
(*args)¶ Accessor to the underlying implementation.
Returns: impl : Implementation
The implementation class.

getName
()¶ Accessor to the object’s name.
Returns: name : str
The name of the object.

getNbColumns
()¶ Accessor to the number of columns.
Returns: nbColumns : int
Number of columns.

getNbRows
()¶ Accessor to the number of rows.
Returns: nbRows : int
Number of rows.

norm
()¶ Compute norm value.
Returns: norm : float
Frobenius norm.

scale
(alpha)¶ Scale matrix inplace A=alpha*A.
Parameters: alpha : float
Coefficient.

setName
(name)¶ Accessor to the object’s name.
Parameters: name : str
The name of the object.

solve
(*args)¶ Solve linear system op(A)*x=b, after A has been factorized.
Parameters: b : sequence of float or
Matrix
Second term of the equation, vector or matrix.
trans : bool
Whether to solve the equation with A (False) or A^t (True). Defaults to False.
Returns: Equation solution, vector or matrix.

solveLower
(*args)¶ Solve lower linear system op(L)*x=b, after A has been factorized.
Parameters: b : sequence of float or
Matrix
Second term of the equation, vector or matrix.
trans : bool
Whether to solve the equation with L (False) or L^t (True). Defaults to False.
Returns: Equation solution, vector or matrix.

transpose
()¶ Transpose matrix inplace.
 Clustering: creation of rows and columns cluster trees
Vertices where interactions are computed are reordered to improve locality.
A binary space partition algorithm is used to recursively divide vertex set.
Root cell contains all vertices. At each recursion step, a cell is divided
into two new cells until it contains less than a given number of vertices.
Space partition is performed orthogonally to original axis, by cutting its
longest dimension.