Source code for nifty8.operators.simple_linear_operators

# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program.  If not, see <http://www.gnu.org/licenses/>.
#
# Copyright(C) 2013-2020 Max-Planck-Society
#
# NIFTy is being developed at the Max-Planck-Institut fuer Astrophysik.

from functools import partial

import numpy as np

from ..domain_tuple import DomainTuple
from ..domains.unstructured_domain import UnstructuredDomain
from ..field import Field
from ..multi_domain import MultiDomain
from ..multi_field import MultiField
from ..utilities import check_object_identity
from .endomorphic_operator import EndomorphicOperator
from .linear_operator import LinearOperator


[docs] class VdotOperator(LinearOperator): """Operator computing the scalar product of its input with a given Field. Parameters ---------- field : :class:`nifty8.field.Field` or :class:`nifty8.multi_field.MultiField` The field used to build the scalar product with the operator input """
[docs] def __init__(self, field): self._field = field self._domain = field.domain self._target = DomainTuple.scalar_domain() self._capability = self.TIMES | self.ADJOINT_TIMES try: from ..re import vdot self._jax_expr = partial(vdot, field.val) except ImportError: self._jax_expr = None
[docs] def apply(self, x, mode): self._check_mode(mode) if mode == self.TIMES: return self._field.vdot(x) return self._field*x.val[()]
[docs] class ConjugationOperator(EndomorphicOperator): """Operator computing the complex conjugate of its input. Parameters ---------- domain: Domain, tuple of domains or DomainTuple domain of the input field """
[docs] def __init__(self, domain): self._domain = DomainTuple.make(domain) self._capability = self._all_ops try: from jax import numpy as jnp from jax.tree_util import tree_map self._jax_expr = partial(tree_map, jnp.conjugate) except ImportError: self._jax_expr = None
[docs] def apply(self, x, mode): self._check_input(x, mode) return x.conjugate()
[docs] class WeightApplier(EndomorphicOperator): """Operator multiplying its input by a given power of dvol. Parameters ---------- domain: Domain, tuple of domains or DomainTuple domain of the input field spaces: list or tuple of int indices of subdomains for which the weights shall be applied power: int the power of to be used for the volume factors """
[docs] def __init__(self, domain, spaces, power): from .. import utilities self._domain = DomainTuple.make(domain) if spaces is None: self._spaces = None else: self._spaces = utilities.parse_spaces(spaces, len(self._domain)) self._power = int(power) self._capability = self._all_ops
[docs] def apply(self, x, mode): self._check_input(x, mode) power = self._power if (mode & 3) else -self._power return x.weight(power, spaces=self._spaces)
[docs] class Realizer(EndomorphicOperator): """Operator returning the real component of its input. Parameters ---------- domain: Domain, tuple of domains, DomainTuple or MultiDomain domain of the input field """
[docs] def __init__(self, domain): from ..sugar import makeDomain self._domain = makeDomain(domain) self._capability = self.TIMES | self.ADJOINT_TIMES try: from jax import numpy as jnp from jax.tree_util import tree_map self._jax_expr = partial(tree_map, jnp.real) except ImportError: self._jax_expr = None
[docs] def apply(self, x, mode): self._check_input(x, mode) return x.real
[docs] class Imaginizer(EndomorphicOperator): """Operator returning the imaginary component of its input. Parameters ---------- domain: Domain, tuple of domains, DomainTuple or MultiDomain domain of the input field """
[docs] def __init__(self, domain): from ..sugar import makeDomain self._domain = makeDomain(domain) self._capability = self.TIMES | self.ADJOINT_TIMES try: from jax import numpy as jnp from jax.tree_util import tree_map self._jax_expr = partial(tree_map, jnp.imag) except ImportError: self._jax_expr = None
[docs] def apply(self, x, mode): self._check_input(x, mode) if mode == self.TIMES: if not np.issubdtype(x.dtype, np.complexfloating): raise ValueError return x.imag if x.dtype not in (np.float64, np.float32): raise ValueError return 1j*x
[docs] class FieldAdapter(LinearOperator): """Operator for conversion between Fields and MultiFields. Parameters ---------- tgt : Domain, tuple of domains, DomainTuple, dict or MultiDomain: If this is a Domain, tuple of Domain or DomainTuple, this will be the operator's target, and its domain will be a MultiDomain consisting of its domain with the supplied `name` If this is a dict or MultiDomain, everything except for `name` will be stripped out of it, and the result will be the operator's target. Its domain will then be the DomainTuple corresponding to the single entry in the operator's domain. name : String The relevant key of the MultiDomain. """
[docs] def __init__(self, target, name): from ..sugar import makeDomain tmp = makeDomain(target) if isinstance(tmp, DomainTuple): self._target = tmp self._domain = MultiDomain.make({name: tmp}) else: self._domain = tmp[name] self._target = MultiDomain.make({name: tmp[name]}) self._capability = self.TIMES | self.ADJOINT_TIMES try: from .. import re as jft def wrap(x): return jft.Vector({name: x}) def unwrap(x): return x[name] if isinstance(tmp, DomainTuple): self._jax_expr = unwrap else: self._jax_expr = wrap except ImportError: self._jax_expr = None
[docs] def apply(self, x, mode): self._check_input(x, mode) if isinstance(x, MultiField): return x.values()[0] else: return MultiField(self._tgt(mode), (x,))
[docs] def __repr__(self): dom = self.domain.keys() if isinstance(self.domain, MultiDomain) else '()' tgt = self.target.keys() if isinstance(self.target, MultiDomain) else '()' return f'{tgt} <- {dom}'
class _SlowFieldAdapter(LinearOperator): """Operator for conversion between Fields and MultiFields. The operator is built so that the MultiDomain is always the target. Its domain is `tgt[name]` Parameters ---------- dom : dict or MultiDomain: the operator's dom name : String The relevant key of the MultiDomain. """ def __init__(self, domain, name): from ..sugar import makeDomain tmp = makeDomain(domain) if not isinstance(tmp, MultiDomain): raise TypeError("MultiDomain expected") self._name = str(name) self._domain = tmp self._target = tmp[name] self._capability = self.TIMES | self.ADJOINT_TIMES def apply(self, x, mode): self._check_input(x, mode) if isinstance(x, MultiField): return x[self._name] return MultiField.from_dict({self._name: x}, domain=self._tgt(mode)) def __repr__(self): return '_SlowFieldAdapter'
[docs] def ducktape(left, right, name): """Convenience function creating an operator that converts between a DomainTuple and a MultiDomain. Parameters ---------- left : None, Operator, or Domainoid Something describing the new operator's target domain. If `left` is an `Operator`, its domain is used as `left`. right : None, Operator, or Domainoid Something describing the new operator's input domain. If `right` is an `Operator`, its target is used as `right`. name : string The component of the `MultiDomain` that will be extracted/inserted Notes ----- - one of the involved domains must be a `DomainTuple`, the other a `MultiDomain`. - `left` and `right` must not be both `None`, but one of them can (and probably should) be `None`. In this case, the missing information is inferred. Returns ------- FieldAdapter or _SlowFieldAdapter an adapter operator converting between the two (possibly partially inferred) domains. """ from ..sugar import makeDomain from .operator import Operator if isinstance(right, Operator): right = right.target elif right is not None: right = makeDomain(right) if isinstance(left, Operator): left = left.domain elif left is not None: left = makeDomain(left) if left is None: # need to infer left from right if isinstance(right, MultiDomain): left = right[name] else: left = MultiDomain.make({name: right}) elif right is None: # need to infer right from left if isinstance(left, MultiDomain): right = left[name] else: right = MultiDomain.make({name: left}) lmulti = isinstance(left, MultiDomain) rmulti = isinstance(right, MultiDomain) if lmulti + rmulti != 1: raise ValueError("need exactly one MultiDomain") if lmulti: if len(left) == 1: return FieldAdapter(left, name) else: return _SlowFieldAdapter(left, name).adjoint if rmulti: if len(right) == 1: return FieldAdapter(left, name) else: return _SlowFieldAdapter(right, name) raise ValueError("must not arrive here")
[docs] class GeometryRemover(LinearOperator): """Operator which transforms between a structured and an unstructured domain. Parameters ---------- domain: Domain, tuple of Domain, or DomainTuple: the full input domain of the operator. space: int, optional The index of the subdomain on which the operator should act. If None, it acts on all spaces. Notes ----- The operator will convert every sub-domain of its input domain to an UnstructuredDomain with the same shape. No weighting by volume factors is carried out. """
[docs] def __init__(self, domain, space=None): self._domain = DomainTuple.make(domain) if space is not None: tgt = [dom for dom in self._domain] tgt[space] = UnstructuredDomain(self._domain[space].shape) else: tgt = [UnstructuredDomain(dom.shape) for dom in self._domain] self._target = DomainTuple.make(tgt) self._capability = self.TIMES | self.ADJOINT_TIMES def identity(x): return x self._jax_expr = identity
[docs] def apply(self, x, mode): self._check_input(x, mode) return x.cast_domain(self._tgt(mode))
[docs] class NullOperator(LinearOperator): """Operator corresponding to a matrix of all zeros. Parameters ---------- domain : DomainTuple or MultiDomain input domain target : DomainTuple or MultiDomain output domain """
[docs] def __init__(self, domain, target): from ..sugar import makeDomain self._domain = makeDomain(domain) self._target = makeDomain(target) self._capability = self.TIMES | self.ADJOINT_TIMES
@staticmethod def _nullfield(dom): if isinstance(dom, DomainTuple): return Field(dom, 0.) else: return MultiField.full(dom, 0.)
[docs] def apply(self, x, mode): self._check_input(x, mode) return self._nullfield(self._tgt(mode))
[docs] def __repr__(self): dom = self.domain.keys() if isinstance(self.domain, MultiDomain) else '()' tgt = self.target.keys() if isinstance(self.target, MultiDomain) else '()' return f'{tgt} <- NullOperator <- {dom}'
[docs] class PartialExtractor(LinearOperator):
[docs] def __init__(self, domain, target): if not isinstance(domain, MultiDomain): raise TypeError("MultiDomain expected") if not isinstance(target, MultiDomain): raise TypeError("MultiDomain expected") self._domain = domain self._target = target for key in self._target.keys(): check_object_identity(self._domain[key], self._target[key]) self._capability = self.TIMES | self.ADJOINT_TIMES self._compldomain = MultiDomain.make({kk: self._domain[kk] for kk in self._domain.keys() if kk not in self._target.keys()})
[docs] def apply(self, x, mode): self._check_input(x, mode) if mode == self.TIMES: return x.extract(self._target) res0 = MultiField.from_dict({key: x[key] for key in x.domain.keys()}) res1 = MultiField.full(self._compldomain, 0.) return res0.unite(res1)
[docs] def __repr__(self): return f'{self.target.keys()} <- {self.domain.keys()}'
[docs] class PrependKey(LinearOperator): """Prepend a string to all keys of a MultiDomain. Parameters ---------- domain : MultiDomain pre : str """
[docs] def __init__(self, domain, pre): if not isinstance(domain, MultiDomain): raise ValueError from ..sugar import makeDomain self._domain = makeDomain(domain) self._pre = str(pre) target = {self._pre+k: domain[k] for k in domain.keys()} self._target = makeDomain(MultiDomain.make(target)) self._capability = self.TIMES | self.ADJOINT_TIMES
[docs] def apply(self, x, mode): self._check_input(x, mode) if mode == self.TIMES: res = {self._pre+k:x[k] for k in self._domain.keys()} else: res = {k:x[self._pre+k] for k in self._domain.keys()} return MultiField.from_dict(res, domain=self._tgt(mode))
[docs] class DomainChangerAndReshaper(LinearOperator): """Convert nifty domains into each other and reshape field. This is only possible if `domain` and `target` have the same number of pixels. Parameters ---------- domain : DomainTuple Domain of the operator target : DomainTuple Target of the operator """
[docs] def __init__(self, domain, target): from ..sugar import makeDomain self._domain = makeDomain(domain) self._target = makeDomain(target) if self._domain.size != self._target.size: s = ["Domain and target do not have the same number of pixels", f"Domain: {self._domain.shape}", f"Target: {self._target.shape}"] raise ValueError("\n".join(s)) self._capability = self.TIMES | self.ADJOINT_TIMES if isinstance(self._domain, MultiDomain) or isinstance(self._target, MultiDomain): raise NotImplementedError("MultiDomains are not supported yet")
[docs] def apply(self, x, mode): from ..sugar import makeField self._check_input(x, mode) x = x.val tgt = self._tgt(mode) return makeField(tgt, x.reshape(tgt.shape))
[docs] def __repr__(self): return f"Reshape {self._shapes(self.target)} <- {self._shapes(self.domain)}"
@staticmethod def _shapes(dom): return " ".join([str(dd.shape) for dd in dom])
[docs] class ExtractAtIndices(LinearOperator): """Extract Field values at given indices along a subspace of a DomainTuple, and puts them in a field with an unstructured domain for the given subspace. Note: This operator supports having the same index several times. If this is not the case also the numerically faster `GeometryRemover` and `MaskOperator` might be useful. Parameters ---------- domain : DomainTuple Domain of the operator indices: tuple Tuple of indices to extract from the field along a given space. The length of the tuple needs to equal the number of axes of the space. Each entry of the tuple should be a tuple of indices to take along the respective axis of the space. Example for a 2d space: indices=((0,1,1,0), (3,4,1,5)) will extract the pixels (0,3), (1,4), (1,1) and (0,5). space: int The index of the space in the domain along which to extract values """
[docs] def __init__(self, domain, indices, space=0): from ..sugar import makeDomain self._domain = makeDomain(domain) if not isinstance(indices, tuple): raise TypeError("indices need to be a tuple") if not len(self._domain[space].shape) == len(indices): raise ValueError("Shape of indices don't match dimension of space") target = [ UnstructuredDomain(len(indices[0])) if i == space else sp for i, sp in enumerate(self._domain)] self._target = makeDomain(target) inds = [slice(None)] * len(domain.shape) dims_of_sps = [len(sp.shape) for sp in self._domain] start_ax_sp = int(np.sum(dims_of_sps[:space])) stop_ax_sp = start_ax_sp + dims_of_sps[space] for i, ax in enumerate(range(start_ax_sp, stop_ax_sp)): inds[ax] = indices[i] self._inds = tuple(inds) self._capability = self.TIMES | self.ADJOINT_TIMES
[docs] def apply(self, x, mode): self._check_input(x, mode) if mode == self.TIMES: res = x.val[self._inds] else: res = np.zeros(self._domain.shape, dtype=x.dtype) np.add.at(res, self._inds, x.val) return Field.from_raw(self._tgt(mode), res)