energy_operators.py 12 KB
Newer Older
Martin Reinecke's avatar
Martin Reinecke committed
1 2 3 4 5 6 7 8 9 10 11 12 13
# 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/>.
#
14
# Copyright(C) 2013-2019 Max-Planck-Society
Martin Reinecke's avatar
Martin Reinecke committed
15
#
16
# NIFTy is being developed at the Max-Planck-Institut fuer Astrophysik.
Martin Reinecke's avatar
Martin Reinecke committed
17

Philipp Arras's avatar
Philipp Arras committed
18 19
import numpy as np

Philipp Arras's avatar
Philipp Arras committed
20
from .. import utilities
Martin Reinecke's avatar
Martin Reinecke committed
21
from ..domain_tuple import DomainTuple
Philipp Arras's avatar
Philipp Arras committed
22 23
from ..field import Field
from ..linearization import Linearization
Philipp Arras's avatar
Philipp Arras committed
24 25
from ..sugar import makeDomain, makeOp
from .linear_operator import LinearOperator
Martin Reinecke's avatar
Martin Reinecke committed
26
from .operator import Operator
Martin Reinecke's avatar
fix  
Martin Reinecke committed
27
from .sampling_enabler import SamplingEnabler
Philipp Arras's avatar
Philipp Arras committed
28
from .sandwich_operator import SandwichOperator
Martin Reinecke's avatar
Martin Reinecke committed
29
from .simple_linear_operators import VdotOperator
Martin Reinecke's avatar
Martin Reinecke committed
30 31 32


class EnergyOperator(Operator):
Philipp Arras's avatar
Philipp Arras committed
33
    """Operator which has a scalar domain as target domain.
34

Martin Reinecke's avatar
Martin Reinecke committed
35
    It is intended as an objective function for field inference.
36

Philipp Arras's avatar
Philipp Arras committed
37 38 39
    Examples
    --------
     - Information Hamiltonian, i.e. negative-log-probabilities.
Martin Reinecke's avatar
Martin Reinecke committed
40
     - Gibbs free energy, i.e. an averaged Hamiltonian, aka Kullback-Leibler
Philipp Arras's avatar
Philipp Arras committed
41
       divergence.
42
    """
Martin Reinecke's avatar
Martin Reinecke committed
43 44 45 46
    _target = DomainTuple.scalar_domain()


class SquaredNormOperator(EnergyOperator):
Philipp Arras's avatar
Philipp Arras committed
47
    """Computes the L2-norm of the output of an operator.
48

Philipp Arras's avatar
Philipp Arras committed
49 50 51
    Parameters
    ----------
    domain : Domain, DomainTuple or tuple of Domain
52
        Domain of the operator in which the L2-norm shall be computed.
Martin Reinecke's avatar
Martin Reinecke committed
53
    """
Philipp Arras's avatar
Philipp Arras committed
54

Martin Reinecke's avatar
Martin Reinecke committed
55 56 57 58
    def __init__(self, domain):
        self._domain = domain

    def apply(self, x):
59
        self._check_input(x)
Martin Reinecke's avatar
Martin Reinecke committed
60
        if isinstance(x, Linearization):
Martin Reinecke's avatar
Martin Reinecke committed
61
            val = Field.scalar(x.val.vdot(x.val))
Martin Reinecke's avatar
Martin Reinecke committed
62
            jac = VdotOperator(2*x.val)(x.jac)
63
            return x.new(val, jac)
Martin Reinecke's avatar
Martin Reinecke committed
64
        return Field.scalar(x.vdot(x))
Martin Reinecke's avatar
Martin Reinecke committed
65 66 67


class QuadraticFormOperator(EnergyOperator):
Philipp Arras's avatar
Philipp Arras committed
68
    """Computes the L2-norm of a Field or MultiField with respect to a
69
    specific kernel given by `endo`.
Philipp Arras's avatar
Philipp Arras committed
70 71 72

    .. math ::
        E(f) = \\frac12 f^\\dagger \\text{endo}(f)
73 74 75

    Parameters
    ----------
Philipp Arras's avatar
Philipp Arras committed
76
    endo : EndomorphicOperator
77
         Kernel of the quadratic form
Martin Reinecke's avatar
Martin Reinecke committed
78
    """
Philipp Arras's avatar
Philipp Arras committed
79 80

    def __init__(self, endo):
Martin Reinecke's avatar
Martin Reinecke committed
81
        from .endomorphic_operator import EndomorphicOperator
Philipp Arras's avatar
Philipp Arras committed
82
        if not isinstance(endo, EndomorphicOperator):
Martin Reinecke's avatar
Martin Reinecke committed
83
            raise TypeError("op must be an EndomorphicOperator")
Philipp Arras's avatar
Philipp Arras committed
84 85
        self._op = endo
        self._domain = endo.domain
Martin Reinecke's avatar
Martin Reinecke committed
86 87

    def apply(self, x):
88
        self._check_input(x)
Martin Reinecke's avatar
Martin Reinecke committed
89
        if isinstance(x, Linearization):
Martin Reinecke's avatar
Martin Reinecke committed
90 91
            t1 = self._op(x.val)
            jac = VdotOperator(t1)(x.jac)
Martin Reinecke's avatar
Martin Reinecke committed
92
            val = Field.scalar(0.5*x.val.vdot(t1))
93
            return x.new(val, jac)
Martin Reinecke's avatar
Martin Reinecke committed
94
        return Field.scalar(0.5*x.vdot(self._op(x)))
Martin Reinecke's avatar
Martin Reinecke committed
95 96 97


class GaussianEnergy(EnergyOperator):
Philipp Arras's avatar
Docs  
Philipp Arras committed
98
    """Computes a negative-log Gaussian.
99

Philipp Arras's avatar
Philipp Arras committed
100
    Represents up to constants in :math:`m`:
Martin Reinecke's avatar
Martin Reinecke committed
101

Philipp Arras's avatar
Philipp Arras committed
102 103
    .. math ::
        E(f) = - \\log G(f-m, D) = 0.5 (f-m)^\\dagger D^{-1} (f-m),
Martin Reinecke's avatar
cleanup  
Martin Reinecke committed
104

Philipp Arras's avatar
Philipp Arras committed
105 106
    an information energy for a Gaussian distribution with mean m and
    covariance D.
107

Philipp Arras's avatar
Philipp Arras committed
108 109 110 111 112 113
    Parameters
    ----------
    mean : Field
        Mean of the Gaussian. Default is 0.
    covariance : LinearOperator
        Covariance of the Gaussian. Default is the identity operator.
Philipp Arras's avatar
Fixup  
Philipp Arras committed
114
    domain : Domain, DomainTuple, tuple of Domain or MultiDomain
Philipp Arras's avatar
Philipp Arras committed
115 116 117 118 119 120
        Operator domain. By default it is inferred from `mean` or
        `covariance` if specified

    Note
    ----
    At least one of the arguments has to be provided.
Martin Reinecke's avatar
Martin Reinecke committed
121
    """
Martin Reinecke's avatar
Martin Reinecke committed
122

Martin Reinecke's avatar
Martin Reinecke committed
123
    def __init__(self, mean=None, covariance=None, domain=None):
Philipp Arras's avatar
Philipp Arras committed
124 125 126 127 128 129
        if mean is not None and not isinstance(mean, Field):
            raise TypeError
        if covariance is not None and not isinstance(covariance,
                                                     LinearOperator):
            raise TypeError

Martin Reinecke's avatar
Martin Reinecke committed
130 131 132 133 134 135 136 137 138 139
        self._domain = None
        if mean is not None:
            self._checkEquivalence(mean.domain)
        if covariance is not None:
            self._checkEquivalence(covariance.domain)
        if domain is not None:
            self._checkEquivalence(domain)
        if self._domain is None:
            raise ValueError("no domain given")
        self._mean = mean
Martin Reinecke's avatar
Martin Reinecke committed
140 141 142 143
        if covariance is None:
            self._op = SquaredNormOperator(self._domain).scale(0.5)
        else:
            self._op = QuadraticFormOperator(covariance.inverse)
Martin Reinecke's avatar
Martin Reinecke committed
144 145 146
        self._icov = None if covariance is None else covariance.inverse

    def _checkEquivalence(self, newdom):
Martin Reinecke's avatar
fix  
Martin Reinecke committed
147
        newdom = makeDomain(newdom)
Martin Reinecke's avatar
Martin Reinecke committed
148
        if self._domain is None:
Philipp Arras's avatar
Philipp Arras committed
149
            self._domain = newdom
Martin Reinecke's avatar
Martin Reinecke committed
150
        else:
Philipp Arras's avatar
Philipp Arras committed
151
            if self._domain != newdom:
Martin Reinecke's avatar
Martin Reinecke committed
152 153 154
                raise ValueError("domain mismatch")

    def apply(self, x):
155
        self._check_input(x)
Philipp Arras's avatar
Philipp Arras committed
156
        residual = x if self._mean is None else x - self._mean
Philipp Arras's avatar
Changes  
Philipp Arras committed
157
        res = self._op(residual).real
158
        if not isinstance(x, Linearization) or not x.want_metric:
Martin Reinecke's avatar
Martin Reinecke committed
159 160 161 162 163 164
            return res
        metric = SandwichOperator.make(x.jac, self._icov)
        return res.add_metric(metric)


class PoissonianEnergy(EnergyOperator):
Philipp Arras's avatar
Docs  
Philipp Arras committed
165 166
    """Computes likelihood Hamiltonians of expected count field constrained by
    Poissonian count data.
167

Philipp Arras's avatar
Philipp Arras committed
168
    Represents up to an f-independent term :math:`log(d!)`:
169

Philipp Arras's avatar
Philipp Arras committed
170 171
    .. math ::
        E(f) = -\\log \\text{Poisson}(d|f) = \\sum f - d^\\dagger \\log(f),
172

Philipp Arras's avatar
Philipp Arras committed
173
    where f is a :class:`Field` in data space with the expectation values for
Martin Reinecke's avatar
Martin Reinecke committed
174
    the counts.
Philipp Arras's avatar
Philipp Arras committed
175 176 177 178 179 180

    Parameters
    ----------
    d : Field
        Data field with counts. Needs to have integer dtype and all field
        values need to be non-negative.
Martin Reinecke's avatar
Martin Reinecke committed
181
    """
Philipp Arras's avatar
Philipp Arras committed
182

183
    def __init__(self, d):
Philipp Arras's avatar
Philipp Arras committed
184 185 186 187
        if not isinstance(d, Field) or not np.issubdtype(d.dtype, np.integer):
            raise TypeError
        if np.any(d.local_data < 0):
            raise ValueError
188 189
        self._d = d
        self._domain = DomainTuple.make(d.domain)
Martin Reinecke's avatar
Martin Reinecke committed
190 191

    def apply(self, x):
192
        self._check_input(x)
Martin Reinecke's avatar
Martin Reinecke committed
193 194
        res = x.sum() - x.log().vdot(self._d)
        if not isinstance(x, Linearization):
Martin Reinecke's avatar
Martin Reinecke committed
195
            return Field.scalar(res)
196 197
        if not x.want_metric:
            return res
Martin Reinecke's avatar
Martin Reinecke committed
198 199 200
        metric = SandwichOperator.make(x.jac, makeOp(1./x.val))
        return res.add_metric(metric)

201

202
class InverseGammaLikelihood(EnergyOperator):
Philipp Arras's avatar
Docs  
Philipp Arras committed
203
    """Computes the negative log-likelihood of the inverse gamma distribution.
204 205 206

    It negative log-pdf(x) is given by

Martin Reinecke's avatar
Martin Reinecke committed
207 208 209 210 211 212 213
    .. math ::

        \\sum_i (\\alpha_i+1)*\\ln(x_i) + \\beta_i/x_i

    This is the likelihood for the variance :math:`x=S_k` given data
    :math:`\\beta = 0.5 |s_k|^2` where the Field :math:`s` is known to have
    the covariance :math:`S_k`.
214 215 216 217 218 219 220

    Parameters
    ----------
    beta : Field
        beta parameter of the inverse gamma distribution
    alpha : Scalar, Field, optional
        alpha parameter of the inverse gamma distribution
221
    """
Philipp Arras's avatar
Philipp Arras committed
222

223 224
    def __init__(self, beta, alpha=-0.5):
        if not isinstance(beta, Field):
Philipp Arras's avatar
Philipp Arras committed
225
            raise TypeError
226 227
        self._beta = beta
        if np.isscalar(alpha):
Martin Reinecke's avatar
Martin Reinecke committed
228 229
            alpha = Field.from_local_data(
                beta.domain, np.full(beta.local_data.shape, alpha))
230 231 232 233
        elif not isinstance(alpha, Field):
            raise TypeError
        self._alphap1 = alpha+1
        self._domain = DomainTuple.make(beta.domain)
234 235

    def apply(self, x):
236
        self._check_input(x)
237
        res = x.log().vdot(self._alphap1) + (1./x).vdot(self._beta)
238 239
        if not isinstance(x, Linearization):
            return Field.scalar(res)
240 241
        if not x.want_metric:
            return res
242
        metric = SandwichOperator.make(x.jac, makeOp(self._alphap1/(x.val**2)))
243 244 245
        return res.add_metric(metric)


Martin Reinecke's avatar
Martin Reinecke committed
246
class BernoulliEnergy(EnergyOperator):
Philipp Arras's avatar
Philipp Arras committed
247
    """Computes likelihood energy of expected event frequency constrained by
248 249
    event data.

Philipp Arras's avatar
Philipp Arras committed
250 251 252 253 254 255 256
    .. math ::
        E(f) = -\\log \\text{Bernoulli}(d|f)
             = -d^\\dagger \\log f  - (1-d)^\\dagger \\log(1-f),

    where f is a field defined on `d.domain` with the expected
    frequencies of events.

257 258
    Parameters
    ----------
Martin Reinecke's avatar
Martin Reinecke committed
259
    d : Field
Philipp Arras's avatar
Philipp Arras committed
260
        Data field with events (1) or non-events (0).
Martin Reinecke's avatar
Martin Reinecke committed
261
    """
Philipp Arras's avatar
Philipp Arras committed
262

263
    def __init__(self, d):
Philipp Arras's avatar
Philipp Arras committed
264 265 266 267 268
        print(d.dtype)
        if not isinstance(d, Field) or not np.issubdtype(d.dtype, np.integer):
            raise TypeError
        if not np.all(np.logical_or(d.local_data == 0, d.local_data == 1)):
            raise ValueError
Martin Reinecke's avatar
Martin Reinecke committed
269
        self._d = d
270
        self._domain = DomainTuple.make(d.domain)
Martin Reinecke's avatar
Martin Reinecke committed
271 272

    def apply(self, x):
273
        self._check_input(x)
Philipp Arras's avatar
Philipp Arras committed
274
        v = -(x.log().vdot(self._d) + (1. - x).log().vdot(1. - self._d))
Martin Reinecke's avatar
Martin Reinecke committed
275
        if not isinstance(x, Linearization):
Martin Reinecke's avatar
Martin Reinecke committed
276
            return Field.scalar(v)
277 278
        if not x.want_metric:
            return v
Philipp Arras's avatar
Philipp Arras committed
279
        met = makeOp(1./(x.val*(1. - x.val)))
Martin Reinecke's avatar
Martin Reinecke committed
280 281 282 283
        met = SandwichOperator.make(x.jac, met)
        return v.add_metric(met)


284
class StandardHamiltonian(EnergyOperator):
Philipp Arras's avatar
Philipp Arras committed
285 286
    """Computes an information Hamiltonian in its standard form, i.e. with the
    prior being a Gaussian with unit covariance.
287

Philipp Arras's avatar
Philipp Arras committed
288
    Let the likelihood energy be :math:`E_{lh}`. Then this operator computes:
289

Philipp Arras's avatar
Philipp Arras committed
290 291
    .. math ::
         H(f) = 0.5 f^\\dagger f + E_{lh}(f):
292

Martin Reinecke's avatar
Martin Reinecke committed
293
    Other field priors can be represented via transformations of a white
294 295
    Gaussian field into a field with the desired prior probability structure.

Martin Reinecke's avatar
Martin Reinecke committed
296
    By implementing prior information this way, the field prior is represented
297 298 299
    by a generative model, from which NIFTy can draw samples and infer a field
    using the Maximum a Posteriori (MAP) or the Variational Bayes (VB) method.

Philipp Arras's avatar
Philipp Arras committed
300 301 302 303 304 305 306 307
    The metric of this operator can be used as covariance for drawing Gaussian
    samples.

    Parameters
    ----------
    lh : EnergyOperator
        The likelihood energy.
    ic_samp : IterationController
308
        Tells an internal :class:`SamplingEnabler` which convergence criterion
Philipp Arras's avatar
Philipp Arras committed
309 310 311 312 313 314 315
        to use to draw Gaussian samples.


    See also
    --------
    `Encoding prior knowledge in the structure of the likelihood`,
    Jakob Knollmüller, Torsten A. Ensslin,
Martin Reinecke's avatar
Martin Reinecke committed
316
    `<https://arxiv.org/abs/1812.04403>`_
Martin Reinecke's avatar
Martin Reinecke committed
317
    """
Philipp Arras's avatar
Philipp Arras committed
318

Martin Reinecke's avatar
Martin Reinecke committed
319 320 321 322
    def __init__(self, lh, ic_samp=None):
        self._lh = lh
        self._prior = GaussianEnergy(domain=lh.domain)
        self._ic_samp = ic_samp
Martin Reinecke's avatar
Martin Reinecke committed
323
        self._domain = lh.domain
Martin Reinecke's avatar
Martin Reinecke committed
324 325

    def apply(self, x):
326
        self._check_input(x)
Philipp Arras's avatar
Philipp Arras committed
327 328 329
        if (self._ic_samp is None or not isinstance(x, Linearization)
                or not x.want_metric):
            return self._lh(x) + self._prior(x)
Martin Reinecke's avatar
Martin Reinecke committed
330
        else:
331
            lhx, prx = self._lh(x), self._prior(x)
Martin Reinecke's avatar
Martin Reinecke committed
332 333
            mtr = SamplingEnabler(lhx.metric, prx.metric.inverse,
                                  self._ic_samp, prx.metric.inverse)
Philipp Arras's avatar
Philipp Arras committed
334
            return (lhx + prx).add_metric(mtr)
Martin Reinecke's avatar
Martin Reinecke committed
335

Philipp Arras's avatar
Philipp Arras committed
336 337 338
    def __repr__(self):
        subs = 'Likelihood:\n{}'.format(utilities.indent(self._lh.__repr__()))
        subs += '\nPrior: Quadratic{}'.format(self._lh.domain.keys())
Martin Reinecke's avatar
Martin Reinecke committed
339
        return 'StandardHamiltonian:\n' + utilities.indent(subs)
Philipp Arras's avatar
Philipp Arras committed
340

Martin Reinecke's avatar
Martin Reinecke committed
341

Martin Reinecke's avatar
Martin Reinecke committed
342
class AveragedEnergy(EnergyOperator):
Philipp Arras's avatar
Docs  
Philipp Arras committed
343
    """Averages an energy over samples.
Martin Reinecke's avatar
Martin Reinecke committed
344

345 346 347
    Parameters
    ----------
    h: Hamiltonian
Philipp Arras's avatar
Philipp Arras committed
348
       The energy to be averaged.
Martin Reinecke's avatar
Martin Reinecke committed
349
    res_samples : iterable of Fields
Torsten Ensslin's avatar
Torsten Ensslin committed
350 351
       Set of residual sample points to be added to mean field for
       approximate estimation of the KL.
352

Philipp Arras's avatar
Docs  
Philipp Arras committed
353 354 355 356 357
    Notes
    -----
    - Having symmetrized residual samples, with both :math:`v_i` and
      :math:`-v_i` being present, ensures that the distribution mean is
      exactly represented.
Torsten Ensslin's avatar
Fix te  
Torsten Ensslin committed
358

Philipp Arras's avatar
Docs  
Philipp Arras committed
359 360 361
    - :class:`AveragedEnergy(h)` approximates
      :math:`\\left< H(f) \\right>_{G(f-m,D)}` if the residuals :math:`f-m`
      are drawn from a Gaussian distribution with covariance :math:`D`.
Martin Reinecke's avatar
Martin Reinecke committed
362
    """
Martin Reinecke's avatar
Martin Reinecke committed
363 364 365

    def __init__(self, h, res_samples):
        self._h = h
Martin Reinecke's avatar
Martin Reinecke committed
366
        self._domain = h.domain
Martin Reinecke's avatar
Martin Reinecke committed
367 368 369
        self._res_samples = tuple(res_samples)

    def apply(self, x):
370
        self._check_input(x)
Philipp Arras's avatar
Philipp Arras committed
371 372
        mymap = map(lambda v: self._h(x + v), self._res_samples)
        return utilities.my_sum(mymap)*(1./len(self._res_samples))