3.6.4. horton.matrix.dense – Dense matrix implementations

3.6.4.1. Naming scheme for the contract, slice or expand methods

The name of contract, slice and expand methods is as follows:

[iadd_]{contract,slice,expand}[_X][_Y][_to_Z]

where each part between square brackets is optional. X, Y and Z can be any of one, two, three or four. The names contract, slice and expand have the following meaning:

contract
Some actual contraction takes place, i.e. summing of (products) of matrix elements.
slice
A subset of the elements is merely selected, without any addition of elements. This is never combined with iadd_.
expand
Products of elements are computed but these products are not added. Similar to an outer product but more general.

When iadd_ is used as a prefix, the result of the contraction is added in-place to the self object. In this case, the _to_Z part is never present. A contraction of all input arguments is made. The dimensionality of the input arguments is indicated by _X and _Y.

When _to_Z is present, the contraction involves self and possibly other arguments whose dimensionality is indicated by _X and _Y. In this case, iadd_ can not be present. The result is written to an output argument. If the output argument is not provided, fresh memory is allocated to compute the contraction and the result is returned. (This allocation is provided for convenience but should not be used in critical situations.)

Some remarks:

  • Similar conventions apply to an expand method.
  • All contract and expand methods are implemented with the driver method DenseLinalgFactory.einsum. However, other implementations than Dense are free to implement things differently.
  • All contract and expand methods never touch the internals of higher-index objects.

For more specific information, read the documentation of the individual classes and methods below.

3.6.4.2. Handling of index symmetry

The dense matrix classes do not exploit matrix symmetry to reduce memory needs. Instead they will happily store non-symmetric data if need be. There are however a few methods in the DenseTwoIndex and DenseFourIndex classes below that take a symmetry argument to check or enforce a certain index symmetry.

The symmetry argument is always an integer that corresponds to the redundancy of the off-diagonal matrix elements in the dense storage. In practice this means the following:

  • DenseTwoIndex
    • symmetry=1: Nothing is checked/enforced
    • symmetry=2: Hermitian index symmetry is checked/enforced (default), i.e. \(\langle i \vert A \vert j \rangle = ` :math:\)langle j vert A vert i rangle`
  • DenseFourIndex
    • symmetry=1: Nothing is checked/enforced
    • symmetry=2: Dummy index symmetry is checked/enforced, i.e. \(\langle ij \vert B \vert kl \rangle =\) \(\langle ji \vert B \vert lk \rangle\)
    • symmetry=4: Hermitian and real index symmetry are checked/enforced, i.e. \(\langle ij \vert B \vert kl \rangle =\) \(\langle kl \vert B \vert ij \rangle =\) \(\langle kj \vert B \vert il \rangle =\) \(\langle il \vert B \vert kj \rangle\). (This only makes sense because the basis functions are assumed to be real.)
    • symmetry=8: All possible symmetries are checked/enforced, i.e. \(\langle ij \vert B \vert kl \rangle =\) \(\langle kl \vert B \vert ij \rangle =\) \(\langle kj \vert B \vert il \rangle =\) \(\langle il \vert B \vert kj \rangle =\) \(\langle ji \vert B \vert lk \rangle =\) \(\langle lk \vert B \vert ji \rangle =\) \(\langle jk \vert B \vert li \rangle =\) \(\langle li \vert B \vert jk \rangle\). (This only makes sense because the basis functions are assumed to be real.)

3.6.4.3. Dense matrix classes

class horton.matrix.dense.DenseLinalgFactory(default_nbasis=None)

Bases: horton.matrix.base.LinalgFactory

Optional arguments:

default_nbasis
The default basis size when constructing new operators/expansions.
create_expansion(nbasis=None, nfn=None)

Create a DenseExpansion with defaults from the LinalgFactory

Optional arguments:

nbasis
The number of basis functions. When not given, the default_nbasis value of the DenseLinalgFactory instance will be used.
nfn
The number of orbitals. When not given, the default_nbasis value of the DenseLinalgFactory instance will be used.
create_four_index(nbasis=None, nbasis1=None, nbasis2=None, nbasis3=None)

Create a DenseFourIndex with defaults from the LinalgFactory

Optional arguments:

nbasis
The number of basis functions. When not given, the default_nbasis value of the DenseLinalgFactory instance will be used.
create_one_index(nbasis=None)

Create a DenseOneIndex with defaults from the LinalgFactory

Optional arguments:

nbasis
The number of basis functions. When not given, the default_nbasis value of the DenseLinalgFactory instance will be used.
create_three_index(nbasis=None, nbasis1=None, nbasis2=None)

Create a DenseThreeIndex with defaults from the LinalgFactory

Optional arguments:

nbasis
The number of basis functions. When not given, the default_nbasis value of the DenseLinalgFactory instance will be used.
create_two_index(nbasis=None, nbasis1=None)

Create a DenseTwoIndex with defaults from the LinalgFactory

Optional arguments:

nbasis
The number of basis functions. When not given, the default_nbasis value of the DenseLinalgFactory instance will be used.
nbasis1
The number of basis functions for the second axis if it differes from nbasis.
static einsum(subscripts, out, factor, clear, *operands)

einsum wrapper for DenseNBody objects

Arguments:

subscripts
The usual subscripts argument for the einsum function, except that the format is slightly limited. All indexes must be explicitly stated. If the output argument is a scalar, -> may not be present. In all other cases, -> must be present.
out
An output DenseNBody object. This argument is not allowed when the result is a scalar.
factor
Multiply the contraction by a scalar factor.
clear
When set to False, the output argument is not zeroed before adding the contraction.
operands
A list where each element is (i) a DenseNBody object or (ii) a two-tuple of DenseNBody object and ranges to select for slicing. The ranges must be a tuple of integers of the form (begin0, end0, begin1, end1, ...) where the number of (begin, end) pairs depends on the dimensionality of the corresponding tensor.

Returns: the out object. When the out argument is given, this is returned with in-place modifications.

from_hdf5(grp)

Construct an instance from data previously stored in an h5py.Group.

Arguments:

grp
An h5py.Group object.
set_default_nbasis(nbasis)
static tensordot(a, b, axes, out=None, factor=1.0, clear=True)

tensordot wrapper for dense n-index objects.

Arguments:

a, b, axes
See documentation of numpy.tensordot

Optional arguments:

out
The output argument, an instance of one of the Dense?Index classes.
to_hdf5(grp)

Write a LinalgFactory to an HDF5 group

Argument:

grp
A h5py.Group instance to write to.
class horton.matrix.dense.DenseOneIndex(nbasis)

Bases: horton.matrix.base.OneIndex

Arguments:

nbasis
The number of basis functions.
assign(other, begin0=0, end0=None)

Assign with the contents of another object

Arguments:

other
Another DenseOneIndex object or a scalar.

Optional arguments:

begin0, end0
If specified, only a sublock of self is assigned.
change_basis_signs(signs)

Correct for different sign conventions of the basis functions.

Arguments:

signs
A numpy array with sign changes indicated by +1 and -1.
clear()

Reset all elements to zero.

copy(begin=0, end=None)

Return a copy of (a part of) the object

Optional arguments:

begin, end
Can be used to select a subblock of the object. When not given, the full range is used.
divide(other, factor=1.0, out=None)

Divide two DenseOneIndex object, multiplied by factor, and return output

Arguments:

other
A DenseOneIndex instance to be divided

Optional arguments:

factor
A scalar factor
out
The output DenseOneIndex
dot(other, factor=1.0)

Dot product with other DenseOneIndex object, multiplied by factor

Arguments:

other
A DenseOneIndex instance to be added

Optional arguments:

factor
A scalar factor
classmethod from_hdf5(grp)

Construct an instance from data previously stored in an h5py.Group.

Arguments:

grp
An h5py.Group object.
get_element(i)

Return a matrix element

get_max(abs=True)

Return maximum (absolute) element

iadd(other, factor=1.0)

Add another DenseOneIndex object in-place, multiplied by factor

Arguments:

other
A DenseOneIndex instance to be added

Optional arguments:

factor
A scalar factor
iscale(factor)

In-place multiplication with a scalar

Arguments:

factor
A scalar factor.
mult(other, out=None, factor=1.0)

Muliply with other DenseOneIndex object, multiplied by factor

Arguments:

other
A DenseOneIndex instance to be added

Optional arguments:

out
The output argument (DenseOneIndex with proper size).
factor
A scalar factor
new()

Return a new one-index object with the same nbasis

norm()

Calculate L2 norm

permute_basis(permutation)

Reorder the coefficients for a given permutation of basis functions.

Arguments:

permutation
An integer numpy array that defines the new order of the basis functions.
randomize()

Fill with random normal data

set_element(i, value)

Set a matrix element

sort_indices(reverse=False)

Return indices of sorted arguments in decreasing order

Optional arguements

reverse
If True search order is reversed to increasing arguements
to_hdf5(grp)

Dump this object in an h5py.Group

Arguments:

grp
An h5py.Group object.
trace(begin0=0, end0=None)

Calculate trace

Optional arguments:

begin0, end0
Can be used to select a subblock of the object. When not given, the full range is used.
nbasis

The number of basis functions

ndim

The number of axes in the N-index object.

shape

The shape of the object

class horton.matrix.dense.DenseExpansion(nbasis, nfn=None)

Bases: horton.matrix.base.Expansion

Arguments:

nbasis
The number of basis functions.

Optional arguments:

nfn
The number of functions to store. Defaults to nbasis.
do_energies
Also allocate an array to store an energy corresponding to each function.
assign(other)

Assign with the contents of another object

Arguments:

other
Another DenseExpansion or DenseTwoIndex object. If DenseTwoIndex, energies and occupations are set to zero.
assign_dot(left, right)

Dot product of orbitals in a DenseExpansion and TwoIndex object

Arguments:

left, right:
An expansion and a two-index object, or a two-index and an expansion object.

The transformed orbitals are stored in self.

assign_occupations(occupation)

Assign orbital occupations

Arguments:

occupation
The orbital occupations to be updated. An OneIndex instance
change_basis_signs(signs)

Correct for different sign conventions of the basis functions.

Arguments:

signs
A numpy array with sign changes indicated by +1 and -1.
check_normalization(overlap, eps=0.0001)

Check that the occupied orbitals are normalized.

Arguments:

overlap
The overlap two-index operator

Optional arguments:

eps
The allowed deviation from unity, very loose by default.
check_orthonormality(overlap, eps=0.0001)

Check that the occupied orbitals are orthogonal and normalized.

Arguments:

overlap
The overlap two-index operator

Optional arguments:

eps
The allowed deviation from unity, very loose by default.
clear()

Reset all elements to zero.

copy()

Return a copy of the object

derive_naturals(dm, overlap)

Arguments:

dm
A DenseTwoIndex object with the density matrix
overlap
A DenseTwoIndex object with the overlap matrix

Optional arguments:

error_eigen(fock, overlap)

Compute the error of the orbitals with respect to the eigenproblem

Arguments:

fock
A DenseTwoIndex Hamiltonian (or Fock) operator.
overlap
A DenseTwoIndex overlap operator.

Returns: the RMSD error on the orbital energies

from_fock(fock, overlap)

Diagonalize a Fock matrix to obtain orbitals and energies

This method updated the attributes coeffs and energies in-place.

Arguments:

fock
The fock matrix, an instance of DenseTwoIndex.
overlap
The overlap matrix, an instance of DenseTwoIndex.
from_fock_and_dm(fock, dm, overlap, epstol=1e-08)

Combined Diagonalization of a Fock and a density matrix

This routine first diagonalizes the Fock matrix to obtain orbitals and orbital energies. Then, using first order (degenerate) perturbation theory, the occupation numbers are computed and, if needed, the the degeneracies of the Fock orbitals are lifted. It is assumed that the Fock and the density matrices commute. This method updated the attributes coeffs, energies and occupations in-place.

Arguments:

fock
The fock matrix, an instance of DenseTwoIndex.
dm
The density matrix, an instance of DenseTwoIndex.
overlap
The overlap matrix, an instance of DenseTwoIndex.

Optional arguments:

epstol
The threshold for recognizing degenerate energy levels. When two subsequent energy levels are separated by an orbital energy less than epstol, they are considered to be degenerate. When a series of energy levels have an orbital energy spacing between subsequent levels that is smaller than epstol, they are all considered to be part of the same degenerate group. For every degenerate set of orbitals, the density matrix is used to (try to) lift the degeneracy.
classmethod from_hdf5(grp)

Construct an instance from data previously stored in an h5py.Group.

Arguments:

grp
An h5py.Group object.
get_homo_energy(offset=0)

Return a homo energy

Optional arguments:

offset
By default, the (highest) homo energy is returned. When this index is above zero, the corresponding lower homo energy is returned.
get_homo_index(offset=0)

Return the index of a HOMO orbital.

Optional arguments:

offset
By default, the (highest) homo energy is returned. When this index is above zero, the corresponding lower homo energy is returned.
get_lumo_energy(offset=0)

Return a lumo energy

Optional arguments:

offset
By default, the (lowest) lumo energy is returned. When this index is above zero, the corresponding higher homo energy is returned.
get_lumo_index(offset=0)

Return the index of a LUMO orbital.

Optional arguments:

offset
By default, the (lowest) lumo energy is returned. When this index is above zero, the corresponding higher homo energy is returned.
imul(other)

Inplace multiplication with other DenseOneIndex.

The attributes energies and occupations are not altered.

Arguments:

other
A DenseOneIndex object.
itranspose()

In-place transpose

new()

Return a new expansion object with the same nbasis and nfn

permute_basis(permutation)

Reorder the coefficients for a given permutation of basis functions (rows).

Arguments:

permutation
An integer numpy array that defines the new order of the basis functions.
permute_orbitals(permutation)

Reorder the coefficients for a given permutation of orbitals (columns).

Arguments:

permutation
An integer numpy array that defines the new order of the orbitals.
randomize()

Fill with random normal data

rotate_2orbitals(angle=0.7853981633974483, index0=None, index1=None)

Rotate two orbitals

Optional arguments:

angle
The rotation angle, defaults to 45 deg.
index0, index1
The orbitals to rotate, defaults to HOMO and LUMO,

The attributes energies and occupations are not altered.

rotate_random()

Apply random unitary transformation distributed with Haar measure

The attributes energies and occupations are not altered.

swap_orbitals(swaps)

Change the order of the orbitals using pair-exchange

Arguments:

swaps
An integer numpy array with two columns where every row corresponds to one swap.

The attributes energies and occupations are also reordered.

to_dm(out=None, factor=1.0, clear=True, other=None)

Compute the density matrix

Optional arguments:

out
An output density matrix (DenseTwoIndex instance).
factor
The density matrix is multiplied by the given scalar.
clear
When set to False, the output density matrix is not zeroed first.
other
Another DenseExpansion object to construct a transfer-density matrix.
to_hdf5(grp)

Dump this object in an h5py.Group

Arguments:

grp
An h5py.Group object.
coeffs

The matrix with the expansion coefficients

energies

The orbital energies

homo_energy

Return a homo energy

Optional arguments:

offset
By default, the (highest) homo energy is returned. When this index is above zero, the corresponding lower homo energy is returned.
lumo_energy

Return a lumo energy

Optional arguments:

offset
By default, the (lowest) lumo energy is returned. When this index is above zero, the corresponding higher homo energy is returned.
nbasis

The number of basis functions

nfn

The number of orbitals (or functions in general)

occupations

The orbital occupations

class horton.matrix.dense.DenseTwoIndex(nbasis, nbasis1=None)

Bases: horton.matrix.base.TwoIndex

Arguments:

nbasis
The number of basis functions. (Number of rows. Also number of columns, unless nbasis1 is given.)

Optional arguments:

nbasis1
When given, this is the number of columns (second index).

Note that by default the two-index object is assumed to be Hermitian. Only when nbasis1 is given, this assumption is dropped.

assign(other, ind=None)

Assign a new contents to the two-index object

Arguments:

other
The new data, may be DenseTwoIndex, DenseOneIndex, a scalar value, or an ndarray.
ind
If provided, only these elements (of DenseOneIndex) are assigned
assign_diagonal(value)

Set diagonal elements to value

Arguments:

value
Either a scalar or a DenseOneIndex object
assign_dot(other, tf2)

Dot product of orbitals in a DenseExpansion and TwoIndex object

Arguments:

other
An expansion object with input orbitals
tf2
A two-index object

The transformed array is stored in self.

assign_two_index_transform(ao_integrals, exp0, exp1=None)

Perform two index transformation: exp0.T * ao_integrals * exp0

Arguments:

ao_integrals
Something computed in the atomic orbital basis
exp0
The molecular orbitals.

Optional arguments:

exp1
When given, the following is computed: exp0.T * ao_integrals * exp1
change_basis_signs(signs)

Correct for different sign conventions of the basis functions.

Arguments:

signs
A numpy array with sign changes indicated by +1 and -1.
clear()

Reset all elements to zero.

contract_to_one(subscripts, out=None, factor=1.0, clear=True, begin0=0, end0=None, begin1=0, end1=None)

Contract self to OneIndex.

Arguments:

subscripts
ab->b: contract first index, ab->a: contract second index.

Optional arguments

out, factor, clear
See DenseLinalgFactory.einsum()
begin0, end0, begin1, end1
Can be used to contract only a part of the two-index object

Returns: the contracted one-index object.

contract_two(subscripts, two, begin0=0, end0=None, begin1=0, end1=None)

Compute the trace using with other two-index objects

Arguments:

subscripts
ab,ba: trace of matrix product. ab,ab: trace after element-wise multiplication (expectation_value).
two
A DenseTwoIndex instance
begin0, end0, begin1, end1
Can be used to select a subblock of the (other) object. When not given, the full range is used.
contract_two_to_one(subscripts, other, out=None, factor=1.0, clear=True, begin0=0, end0=None, begin1=0, end1=None)

Contract two TwoIndex objects to a one OneIndex.

Arguments:

subscripts:
ab,ab->b: contract first index. ab,ab->a: contract second index.
other
The second DenseTwoIndex object.

Optional arguments:

out, factor, clear
See DenseLinalgFactory.einsum()
begin0, end0, begin1, end1
Can be used to contract only a part of the other two-index object. When not given, the full range is taken.

Returns: the contracted one-index object.

contract_two_to_two(subscripts, other, out=None, factor=1.0, clear=True, begin0=0, end0=None, begin1=0, end1=None)

Contract two TwoIndex objects to a one twoIndex.

Arguments:

subscripts:
ab,cb->ac, ab,ca->cb, ab,bc->ac, ab,ac->cb, ab,ac->bc, ab,cb->ca
other
The second DenseTwoIndex object.

Optional arguments:

out, factor, clear
See DenseLinalgFactory.einsum()
begin0, end0, begin1, end1
Can be used to select a subblock of the (other) object. When not given, the full range is used.

Returns: the contracted two-index object.

copy(begin0=0, end0=None, begin1=0, end1=None)

Return a copy of (a part of) the object

Optional arguments:

begin0, end0, begin1, end1
Can be used to select a subblock of the object. When not given, the full range is used. Only when the ranges are equal for both axes, an Hermitian two-index may be returned.
copy_diagonal(out=None, begin=0, end=None)

Copy (part of) the diagonal of the two-index object

Optional arguments:

out
The output argument (DenseOneIndex with proper size).
begin, end
Can be used to select a range of the diagonal. If not given, then the entire diagonal is copied.
copy_slice(ind, out=None)

Copy slice of the two-index object into one-index object

Arguments:

ind
2-Tuple of 1-dim arrays with row and column indices of TwoIndex object to be copied.

Optional arguments:

out
The output argument (DenseOneIndex with proper size).
diagonalize()

Return eigenvalues of two-index object

distance_inf(other)

The infinity norm distance between self and other

Arguments:

other
A DenseTwoIndex instance.
classmethod from_hdf5(grp)

Construct an instance from data previously stored in an h5py.Group.

Arguments:

grp
An h5py.Group object.
get_element(i, j)

Return a matrix element

iabs()

In-place absolute values

iadd(other, factor=1.0, begin0=0, end0=None, begin1=0, end1=None, transpose=False)

Add another DenseTwoIndex object in-place, multiplied by factor. If begin0, end0, begin1, end1 are specified, other is added to the selected range.

Arguments:

other
A DenseTwoIndex, DenseOneIndex instance or float to be added

Optional arguments:

factor
A scalar factor
begin0, end0, begin1, end1
When given, specify the ranges where the contribution will be added. When not given, the full range is used.
iadd_contract_two_one(subscripts, two, one, factor=1.0, begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None)

In-place addition of a contraction of a two- and one-index object

Arguments:

two, one
The two- and one-index objects, respectively. Instances of
subscripts
ab,b->ab: contract with the first index of the two-index object. ab,a->ab: contract with the second index of the two-index. ba,a->ab: contract with the second index and transpose. object.
factor
The term added is scaled by this factor.
begin0, end0, begin1, end1
Can be used to select a subblock of the two-index object (two). When not given, the full range is used.
begin2, end2
Can be used to select a subblock of the one-index object (one). When not given, the full range is used.
iadd_dot(other0, other1, factor=1.0, begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None, begin3=0, end3=None)

In-place addition of dot product: other0 * other1

Arguments:

other0, other1
Two-index objects that go into the kronecker product.

Optional arguments:

factor
The term added is scaled by this factor.
begin0, end0, begin1, end1
Can be used to select a subblock of the other0 object. When not given, the full range is used.
begin2, end2, begin3, end3
Can be used to select a subblock of the other1 object. When not given, the full range is used.
iadd_dott(other0, other1, factor=1.0)

In-place addition of dot product: other0 * other1.T

Arguments:

other0, other1
Two-index objects that go into the kronecker product.

Optional arguments:

factor
The term added is scaled by this factor.
iadd_kron(other0, other1, factor=1.0)

In-place addition of kronecker product of two other DenseTwoIndex

Arguments:

other0, other1
Two-index objects that go into the kronecker product.

Optional arguments:

factor
The term added is scaled by this factor.
iadd_mult(other0, other1, factor=1.0, begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None, begin3=0, end3=None, transpose0=False, transpose1=False)

In-place addition of multiplication: other0 * other1

Arguments:

other0, other1
Two-index objects that go into the product.

Optional arguments:

factor
The term added is scaled by this factor.
begin0, end0, begin1, end1
Can be used to select a subblock of the other1 object. When not given, the full range is used.
begin2, end2, begin3, end3
Can be used to select a subblock of the other0 object. When not given, the full range is used.
transpose0, transpose1
Can be used to select transpose of other0, other1 objects
iadd_one_mult(other0, other1, factor=1.0, transpose0=False, transpose1=False)

In-place addition of multiplication: other0 * other1

Arguments:

other0, other1
One-index objects that go into the product.

Optional arguments:

factor
The term added is scaled by this factor.
transpose0, transpose1
Can be used to select transpose of one-index objects.
iadd_outer(other0, other1, factor=1.0)

In-place addition of outer product of two other DenseTwoIndex

Arguments:

other0, other1
Two-index objects that go into the outer product. They are raveled before taking the outer product.

Optional arguments:

factor
The term added is scaled by this factor.
iadd_shift(lshift)

Add positive shift to elements. If negative subtract shift

Arguments:

lshift
A scalar used to augment the matrix elements.
iadd_slice(other, factor=1.0, begin0=0, end0=None, begin1=0, end1=None)

Add slice of another DenseTwoIndex object in-place, multiplied by factor. The slice of other is added to the full range of the array.

Arguments:

other
A DenseTwoIndex instance to be added

Optional arguments:

factor
A scalar factor
begin0, end0, begin1, end1
When given, specify the ranges of contribution to be added. When not given, the full range is used.
iadd_t(other, factor=1.0, begin0=0, end0=None, begin1=0, end1=None)

See DenseTwoIndex.iadd(), transpose=True

iadd_tdot(other0, other1, factor=1.0)

In-place addition of dot product: other0.T * other1

Arguments:

other0, other1
Two-index objects that go into the kronecker product.

Optional arguments:

factor
The term added is scaled by this factor.
idot(other)

In-place dot product: self = self * other

Arguments:

other
The other array
imul(other, factor=1.0, begin0=0, end0=None, begin1=0, end1=None)

In-place element-wise multiplication: self *= other * factor

Arguments:

other
A DenseTwoIndex instance.

Optional arguments:

factor
The two-index object is scaled by this factor.
begin0, end0, begin1, end1
Can be used to select a subblock of the (other) object. When not given, the full range is used.
imul_t(other, factor=1.0, begin0=0, end0=None, begin1=0, end1=None)

In-place element-wise multiplication: self *= other.T * factor

Arguments:

other
A DenseTwoIndex instance.

Optional arguments:

factor
The two-index object is scaled by this factor.
begin0, end0, begin1, end1
Can be used to select a subblock of the (other) object. When not given, the full range is used.
inner(vec0, vec1)

Compute an inner product of two vectors using the two-index as a metric

Arguments:

vec0, vec1
The vectors, either DenseOneIndex or numpy arrays.
inverse()

Return the inverse of two-index object

iortho()

In-place orthogonalization

Arguments:

is_shape_symmetric(symmetry)

Check whether the symmetry argument matches the shape

is_symmetric(symmetry=2, rtol=1e-05, atol=1e-08)

Check the symmetry of the array.

Optional arguments:

symmetry
The symmetry to check. See Handling of index symmetry for more details.
rtol and atol
relative and absolute tolerance. See to np.allclose.
iscale(factor)

In-place multiplication with a scalar

Arguments:

factor
A scalar factor.
itranspose()

In-place transpose

new()

Return a new two-index object with the same nbasis (and nbasis1)

permute_basis(permutation)

Reorder the coefficients for a given permutation of basis functions.

The same permutation is applied to all indexes.

Arguments:

permutation
An integer numpy array that defines the new order of the basis functions.
randomize()

Fill with random normal data

set_element(i, j, value, symmetry=2)

Set a matrix element

Arguments:

i, j
The matrix indexes to be set
value
The value to be assigned to the matrix element.

Optional arguments:

symmetry
When 2 (the default), the element (j,i) is set to the same value. When set to 1 the opposite off-diagonal is not set. See Handling of index symmetry for more details.
sqrt()

Return the real part of the square root of two-index object

sum(begin0=0, end0=None, begin1=0, end1=None)

Return the sum of all elements (in the selected range)

Optional arguments:

begin0, end0, begin1, end1
Can be used to select a subblock of the object to be contracted.
symmetrize(symmetry=2)

Symmetrize in-place

Optional arguments:

symmetry
The symmetry to impose. See Handling of index symmetry for more details.
to_hdf5(grp)

Dump this object in an h5py.Group

Arguments:

grp
An h5py.Group object.
trace(begin0=0, end0=None, begin1=0, end1=None)

Return the trace of the two-index object.

Optional arguments:

begin0, end0, begin1, end1
Can be used to select a subblock of the object to be contracted.
nbasis

The number of basis functions

nbasis1

The other size of the two-index object

ndim

The number of axes in the N-index object.

shape

The shape of the object

class horton.matrix.dense.DenseThreeIndex(nbasis, nbasis1=None, nbasis2=None)

Bases: horton.matrix.base.ThreeIndex

Arguments:

nbasis
The number of basis functions.
assign(other)

Assign with the contents of another object

Arguments:

other
Another DenseThreeIndex object or a scalar.
change_basis_signs(signs)

Correct for different sign conventions of the basis functions.

Arguments:

signs
A numpy array with sign changes indicated by +1 and -1.
clear()

Reset all elements to zero.

contract_to_two(subscripts, out=None, factor=1.0, clear=True, begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None)

Contract self to TwoIndex.

Arguments:

subscripts
abc->ac: contract first index.

Optional arguments

out, factor, clear
See DenseLinalgFactory.einsum()
begin0, end0, begin1, end1, begin2, end2
Can be used to contract only a part of the three-index object. When not given, the full range is used.

Returns: the contracted two-index object.

contract_two_to_three(subscripts, two, out=None, factor=1.0, clear=True, begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None)

Contracts with two-index to obtain three-index.

Arguments:

subscripts
One of abc,ad->cdb, abc,ad->cbd, abc,da->dbc, abc,da->bdc
inp
A DenseTwoIndex input object.

Optional arguments:

out, factor, clear
See DenseLinalgFactory.einsum()
begin0, end0, begin1, end1, begin2, end2
Can be used to contract only a part of the three-index object
contract_two_to_two(subscripts, inp, out=None, factor=1.0, clear=True)

Contracts self with two-index to obtain two-index.

Arguments:

subscripts
One of abc,ab->ac, abc,ab->ca, abc,bc->ba, abc,bc->ab, abc,cb->ac.
inp
A DenseTwoIndex input object.

Optional arguments:

out, factor, clear
See DenseLinalgFactory.einsum()
copy(begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None)

Return a copy of (a part of) the object

Optional arguments:

begin0, end0, begin1, end1, begin2, end2
Can be used to select a subblock of the object. When not given, the full range is used.
classmethod from_hdf5(grp)

Construct an instance from data previously stored in an h5py.Group.

Arguments:

grp
An h5py.Group object.
get_element(i, j, k)

Return a matrix element

iadd(other, factor=1.0)

Add another DenseThreeIndex object in-place, multiplied by factor

Arguments:

other
A DenseThreeIndex instance to be added

Optional arguments:

factor
The added term is scaled by this factor.
iadd_expand_two_one(subscripts, two, one, factor=1.0)

In-place addition of expanded two-index and one-index to three-index.

Arguments:

subscripts
Contraction type: ab,c->cab or ab,c->acb.
two
A DenseTwoIndex object.
one
A DenseOneIndex object.

Optional arguments:

factor
The added term is scaled by this factor.
iadd_expand_two_two(subscripts, two0, two1, factor=1.0, begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None, begin3=0, end3=None)

In-place addition of expanded two-index and two-index to three-index.

Arguments:

subscripts
Contraction type: ac,bc->abc, ab,bc->abc, ab,ac->acb, cb,ac->acb, ac,ab->abc, ab,ac->abc
two0, two1
A DenseTwoIndex object.

Optional arguments:

factor
The added term is scaled by this factor.
begin0, end0, begin1, end1
Can be used to contract only a part of the two1 object. When not given, the full range is used.
begin2, end2, begin3, end3
Can be used to contract only a part of the two0 object. When not given, the full range is used.
iadd_slice(other, factor=1.0, begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None)

Add slice of another DenseThreeIndex object in-place, multiplied by factor

Arguments:

other
A DenseThreeIndex instance to be added

Optional arguments:

factor
The added term is scaled by this factor.
begin0, end0, begin1, end1, begin2, end2
Can be used to add only a part of the three-index object
iscale(factor)

In-place multiplication with a scalar

Arguments:

factor
A scalar factor.
new()

Return a new three-index object with the same nbasis

permute_basis(permutation)

Reorder the coefficients for a given permutation of basis functions.

Arguments:

permutation
An integer numpy array that defines the new order of the basis functions.
randomize()

Fill with random normal data

set_element(i, j, k, value)

Set a matrix element

to_hdf5(grp)

Dump this object in an h5py.Group

Arguments:

grp
An h5py.Group object.
nbasis

The number of basis functions

nbasis1

The number of basis functions

nbasis2

The number of basis functions

ndim

The number of axes in the N-index object.

shape

The shape of the object

class horton.matrix.dense.DenseFourIndex(nbasis, nbasis1=None, nbasis2=None, nbasis3=None)

Bases: horton.matrix.base.FourIndex

Arguments:

nbasis
The number of basis functions.

Optional arguments:

nbasis1, nbasis2, nbasis3

assign(other)

Assign with the contents of another object

Arguments:

other
Another DenseFourIndex object or a np ndarrray.
assign_four_index_transform(ao_integrals, exp0, exp1=None, exp2=None, exp3=None, method='tensordot')

Perform four index transformation.

Arguments:

oa_integrals
A DenseFourIndex with integrals in atomic orbitals.
exp0
A DenseExpansion object with molecular orbitals

Optional arguments:

exp1, exp2, exp3
Can be provided to transform each index differently.
method
Either einsum or tensordot (default).
change_basis_signs(signs)

Correct for different sign conventions of the basis functions.

Arguments:

signs
A numpy array with sign changes indicated by +1 and -1.
clear()

Reset all elements to zero.

contract_four(subscripts, other, factor=1.0, begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None, begin3=0, end3=None)

Compute the trace with other four-index object

Arguments:

subscripts
abcd,abcd, abcd,adcb, abcd,acbd, abcd,acdb
other
A DenseFourIndex instance

Optional arguments:

factor
A scalar factor
begin0, end0, begin1, end1, begin2, end2, begin3, end3
Can be used to select a subblock of the other object. When not given, the full range is used.
contract_four_to_four(subscripts, four, out=None, factor=1.0, clear=True, begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None, begin3=0, end3=None)

Contracts with a four-index object to obtain a four-index object.

Arguments:

subscripts
Any of abcd,cedf->abfe, abcd,cefd->abfe, abcd,cedf->feab, abcd,cefd->feab, abcd,eadf->fbce, abcd,eadf->cefb, abcd,aedf->cbfe, abcd,aedf->fecb, abcd,acef->ebfd, abcd,bdef->aecf, abcd,cedf->abef, abcd,cefd->abef, abcd,cedf->efab, abcd,cefd->efab, abcd,eadf->ebcf, abcd,eadf->cfeb, abcd,eadf->cbef, abcd,eadf->efcb, abcd,eafd->cbef, abcd,eafd->efcb
four
An instance of DenseFourIndex.

Optional arguments:

out, factor, clear
See DenseLinalgFactory.einsum()
begin0, end0, begin1, end1, begin2, end2, begin3, end3
Can be used to select a subblock of the other four-index object (four). When not given, the full range is used.
contract_four_to_two(subscripts, four, out=None, factor=1.0, clear=True, begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None, begin3=0, end3=None)

Contracts with a four-index object to obtain a two-index object.

Arguments:

subscripts
Any of abcd,aced->be, abcd,acde->be, abcd,aced->eb, abcd,acde->eb, abcd,ecbd->ae, abcd,ecdb->ae, abcd,ecdb->ea, abcd,ecbd->ea, abcd,acbe->ed, abcd,aceb->ed, abcd,aebd->ce, abcd,aedb->ce
four
An instance of DenseFourIndex.

Optional arguments:

out, factor, clear
See DenseLinalgFactory.einsum()
begin0, end0, begin1, end1, begin2, end2, begin3, end3
Can be used to select a subblock of the other four-index object (four). When not given, the full range is used.
contract_three_to_three(subscripts, three, out=None, factor=1.0, clear=True, begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None)

Contracts with a three-index object to obtain a three-index object.

Arguments:

subscripts
Any of abcd,ace->ebd, abcd,ebd->ace
three
An instance of DenseThreeIndex.

Optional arguments:

out, factor, clear
See DenseLinalgFactory.einsum()
begin0, end0, begin1, end1, begin2, end2
Can be used to select a subblock of the three-index object (three). When not given, the full range is used.
contract_to_two(subscripts, out=None, factor=1.0, clear=True, begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None, begin3=0, end3=None)

Contract four-index object to two-index object

Arguments:

subscripts
abcb->ac, abbc->ac

Optional arguments:

out, factor, clear
See DenseLinalgFactory.einsum()
begin0, end0, begin1, end1, begin2, end2, begin3, end3
Can be used to select a subblock of the object. When not given, the full range is used.
contract_two(subscripts, other, factor=1.0, begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None, begin3=0, end3=None)

Compute diagonal trace with two-index objects

Arguments:

subscripts
aabb,ab.
other
A DenseTwoIndex instance

Optional arguments:

factor
A scalar factor. See DenseLinalgFactory.einsum()
begin0, end0, begin1, end1, begin2, end2, begin3, end3
Can be used to select a subblock of the four-index object. When not given, the full range is used.
contract_two_to_four(subscripts, two, out=None, factor=1.0, clear=True, begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None, begin3=0, end3=None, begin4=0, end4=None, begin5=0, end5=None)

Contracts with a two-index object to obtain a four-index object.

Arguments:

subscripts
Any of abcd,cd->acbd, abcd,cd->acdb, abcd,cb->acdb, abcd,cb->acbd, abcd,ab->acbd, abcd,ab->acdb, abcd,ad->acbd, abcd,ad->acdb, abcd,ad->abcd, abcd,ad->abdc, abcd,bd->abcd, abcd,bd->abdc, abcd,bc->abdc, abcd,bc->abcd, abcd,ac->abcd, abcd,ac->abdc, abcd,bd->cabd, abcd,bc->dabc, abcd,ca->cabd, abcd,da->dabc, abcd,dc->dabc, abcd,ba->dabc, aabc,dc->adbc, aabc,db->adbc, abcc,ad->abcd, abcc,bd->abcd, abcd,bc->acbd, abcd,eb->aecd, abcd,eb->cdae, abcd,ed->ceab, abcd,ed->abce, abcd,ae->cdeb, abcd,ae->ebcd, abcd,ce->edab, abcd,ce->abed, abcd,ab->abcd, abcd,cb->abcd, abcd,ec->eadb, abcd,ec->dbea, abcd,ae->cedb, abcd,ae->dbce, abcd,ec->ebad, abcd,ed->ebac, abcd,ec->adeb, abcd,ed->aceb, abcd,ae->debc, abcd,ae->cebd, abcd,ae->bcde, abcd,ae->bdce
two
An instance of DenseTwoIndex.

Optional arguments:

out, factor, clear
See DenseLinalgFactory.einsum()
begin0, end0, begin1, end1, begin2, end2, begin3, end3
Can be used to select a subblock of the four-index object. When not given, the full range is used.
begin4, end4, begin5, end5
Can be used to select a subblock of the two-index object (two). When not given, the full range is used.
contract_two_to_three(subscripts, two, out=None, factor=1.0, clear=True, begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None, begin3=0, end3=None)

Contracts with a two-index object to obtain a three-index object.

Arguments:

subscripts
Any of aabc,ad->bcd, 'abcd,ac->bdc, abcd,ad->bcd, abcd,bc->abd, abcd,ac->abd, abcc,dc->dab, abcd,ac->bcd, abcc,cd->dab, abcc,dc->abd, abcb,db->adc, abcc,dc->adb, abcb,db->dac, abcc,cd->abd
two
An instance of DenseTwoIndex.

Optional arguments:

out, factor, clear
See DenseLinalgFactory.einsum()
begin0, end0, begin1, end1, begin2, end2, begin3, end3
Can be used to select a subblock of the four-index object. When not given, the full range is used.
contract_two_to_two(subscripts, two, out=None, factor=1.0, clear=True, begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None, begin3=0, end3=None, begin4=0, end4=None, begin5=0, end5=None)

Contract self with a two-index to obtain a two-index.

Arguments:

subscripts
Any of abcd,bd->ac (direct), abcd,cb->ad (exchange), aabb,cb->ac, abcc,bc->ab, aabc,ab->bc, aabc,ac->bc, abcc,ac->ab, abcb,cb->ac, abcb,ab->ac, abcc,bc->ba, abcc,bc->ab, abcd,ac->db, abcd,ad->cb, abcd,ac->bd, abcd,ad->bc, abcd,ab->cd
two
The input two-index object. (DenseTwoIndex)

Optional arguments:

out, factor, clear
See DenseLinalgFactory.einsum()
begin0, end0, begin1, end1, begin2, end2, begin3, end3
Can be used to select a subblock of the four-index object. When not given, the full range is used.
begin4, end4, begin5, end5
Can be used to select a subblock of the two-index object (two). When not given, the full range is used.
copy(begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None, begin3=0, end3=None)

Return a copy of (a part of) the object

Optional arguments:

begin0, end0, begin1, end1, begin2, end2, begin3, end3
Can be used to select a subblock of the object. When not given, the full range is used.
classmethod from_hdf5(grp)

Construct an instance from data previously stored in an h5py.Group.

Arguments:

grp
An h5py.Group object.
get_element(i, j, k, l)

Return a matrix element

iadd(other, factor=1.0)

Add another DenseFourIndex object in-place, multiplied by factor

Arguments:

other
A DenseFourIndex instance to be added

Optional arguments:

factor
The added term is scaled by this factor.
iadd_exchange()

In-place addition of its own exchange contribution

iadd_expand_three_to_four(axis, three, factor=1.0)

Expand three-index object along one axis and add it to four-index object.

Arguments:

axis
Any of 1-3-1-2 (abac += abc), 1-2-1-2 (abca += abc), 0-2-0-2 (abcb += abc), 0-3-0-2 (abbc += abc), 0-2-0-1 (abcb += acb). The expansion is performed for repeated indices in the four-index object
three
An instance of DenseThreeIndex.

Optional arguments:

factor
See DenseLinalgFactory.einsum()
iadd_expand_two_to_four(axis, two, factor=1.0, begin0=0, end0=None, begin1=0, end1=None)

Expand two-index object along two axes and add it to four-index object.

Arguments:

axis
Any of 1-3 (abac += bc), 0-2 (abcb += ac), 0-3 (abbc += ac), 1-2 (abca += bc), diag (abab += ab). The expansion is performed for the repeated indices in the four-index object
two
An instance of DenseTwoIndex.

Optional arguments:

factor
See DenseLinalgFactory.einsum()
begin0, end0, begin1, end1
Can be used to select a subblock of the two-index object (two). When not given, the full range is used.
imul(other, factor=1.0)

In-place element-wise multiplication: self *= other * factor

Arguments:

other
A DenseFourIndex instance.

Optional arguments:

factor
The four-index object is scaled by this factor.
is_shape_symmetric(symmetry)

Check whether the symmetry argument matches the shape

is_symmetric(symmetry=8, rtol=1e-05, atol=1e-08)

Check the symmetry of the array.

Optional arguments:

symmetry
The symmetry to check. See Handling of index symmetry for more details. In addition to 1, 2, 4, 8, also ‘cdab’ is supported.
rtol and atol
relative and absolute tolerance. See to np.allclose.
iscale(factor)

In-place multiplication with a scalar

Arguments:

factor
A scalar factor.
itranspose()

In-place transpose: 0,1,2,3 -> 1,0,3,2

new()

Return a new four-index object with the same nbasis

permute_basis(permutation)

Reorder the coefficients for a given permutation of basis functions.

Arguments:

permutation
An integer numpy array that defines the new order of the basis functions.
randomize()

Fill with random normal data

reshape(shape)

Reshape array

Optional arguments:

shape
List containing the new dimension of each axis.
set_element(i, j, k, l, value, symmetry=8)

Set a matrix element

Arguments:

i, j, k, l
The matrix indexes to be set
value
The value to be assigned to the matrix element.

Optional arguments:

symmetry
The level of symmetry to be enforced when setting the matrix element. See Handling of index symmetry for more details.
slice_to_four(subscripts, out=None, factor=1.0, clear=True, begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None, begin3=0, end3=None)

Returns a four-index contraction of the four-index object.

Arguments:

superscripts
Any of abcd->abcd

Optional arguments:

out, factor, clear
See DenseLinalgFactory.einsum()
begin0, end0, begin1, end1, begin2, end2, begin3, end3
Can be used to select a subblock of the object. When not given, the full range is used.
slice_to_three(subscripts, out=None, factor=1.0, clear=True)

Returns a three-index contraction of the four-index object.

Arguments:

superscripts
Any of abcc->bac, abcc->abc, abcb->abc, abbc->abc

Optional arguments:

out, factor, clear
See DenseLinalgFactory.einsum()
slice_to_two(subscripts, out=None, factor=1.0, clear=True, begin0=0, end0=None, begin1=0, end1=None, begin2=0, end2=None, begin3=0, end3=None)

Returns a two-index contraction of the four-index object.

Arguments:

superscripts
Any of aabb->ab, abab->ab, abba->ab

Optional arguments:

out, factor, clear
See DenseLinalgFactory.einsum()
begin0, end0, begin1, end1, begin2, end2, begin3, end3
Can be used to select a subblock of the object. When not given, the full range is used.
sum()

Return the sum of all elements

symmetrize(symmetry=8)

Symmetrize in-place

Optional arguments:

symmetry
The symmetry to impose. See Handling of index symmetry for more details.
to_hdf5(grp)

Dump this object in an h5py.Group

Arguments:

grp
An h5py.Group object.
nbasis

The number of basis functions

nbasis1

The number of basis functions

nbasis2

The number of basis functions

nbasis3

The number of basis functions

ndim

The number of axes in the N-index object.

shape

The shape of the object