Model solution

pathfinder/model/oscillating/__init__.py

@@ -1,4 +1,5 @@
 from pathfinder.model.oscillating.oscillating_dipole import OscillatingDipole, OscillatingDipoleArrangement
 from pathfinder.model.oscillating.dot import DotMeasurement
+from pathfinder.model.oscillating.model import DotOscillatingDipoleModel
 
-__all__ = ['DotMeasurement', 'OscillatingDipole', 'OscillatingDipoleArrangement']
+__all__ = ['DotMeasurement', 'OscillatingDipole', 'OscillatingDipoleArrangement', 'DotOscillatingDipoleModel']

pathfinder/model/oscillating/model.py

@@ -0,0 +1,44 @@
+from typing import Callable, Sequence
+import numpy
+import scipy.optimize
+from pathfinder.model.oscillating.dot import DotMeasurement
+
+
+class DotOscillatingDipoleModel():
+	'''
+	Model of n static oscillating with a collection of voltage measurements
+	at dots at different positions and frequencies.
+
+	Parameters
+	----------
+	dots : Sequence[DotMeasurement]
+		A collection of dots representing a series of measured noise spectrum values.
+	n: int
+		The number of dipoles to assume.
+	'''
+	def __init__(self, dots: Sequence[DotMeasurement], n: int) -> None:
+		self.dots = dots
+		self.m = len(dots)
+		self.n = n
+
+	def __repr__(self) -> str:
+		return f'DotOscillatingDipoleModel({repr(list(self.dots))}, {self.n})'
+
+	def costs(self) -> Callable[[numpy.ndarray], numpy.ndarray]:
+		def costs_to_return(pt: numpy.ndarray) -> numpy.ndarray:
+			return numpy.array([dot.cost(pt) for dot in self.dots])
+
+		return costs_to_return
+
+	def jac(self) -> Callable[[numpy.ndarray], numpy.ndarray]:
+		def jac_to_return(pts: numpy.ndarray) -> numpy.ndarray:
+			return numpy.array([dot.jac(pts) for dot in self.dots])
+
+		return jac_to_return
+
+	def sol(self, initial_dipole=(0.1, 0.1, 0.1), initial_position=(.1, .1, .1), initial_frequency=1, use_root=True):
+		initial = numpy.tile(numpy.concatenate((initial_dipole, initial_position, initial_frequency)), self.n)
+		if use_root and self.m == 7 * self.n:
+			# We are perfectly specified
+			return scipy.optimize.root(self.costs(), initial, jac=self.jac(), tol=1e-12)
+		return scipy.optimize.least_squares(self.costs(), initial, jac=self.jac(), ftol=1e-15, gtol=3e-16)

tests/model/oscillating/test_oscmodel.py

@@ -0,0 +1,35 @@
+import numpy
+import pathfinder.model.oscillating as model
+
+
+def test_dotdipolemodel_repr():
+	mod = model.DotOscillatingDipoleModel((), 1)
+	assert repr(mod) == "DotOscillatingDipoleModel([], 1)"
+
+
+def test_dotdipolemodel_m():
+	mod = model.DotOscillatingDipoleModel([model.DotMeasurement(1, (-1, -1, -1), 11), model.DotMeasurement(2, (0, 0, 0), 3)], 1)
+	assert mod.m == 2
+
+
+def test_dotdipolemodel_cost():
+	mod = model.DotOscillatingDipoleModel([model.DotMeasurement(.05, (-1, -1, -1), 11), model.DotMeasurement(.002, (0, 0, 0), 3)], 1)
+	costs = mod.costs()
+	jac = mod.jac()
+
+	pt_to_test = numpy.array((1, 2, 3, 4, 5, 6, 7))
+	expected_cost = [-0.04998578036712978, -0.00191383161468913]
+	expected_jacobian = [
+		[
+			3.742008650059147e-6, 4.4904103800709765e-6, 5.238812110082806e-6,
+			-3.12967996186765e-6, -3.1568945702317167e-6, -3.184109178595783e-6,
+			8.603475350051953e-7
+		], [
+			0.000021542096327717702, 0.000026927620409647127, 0.000032313144491576555,
+			-0.000021472154456523818, -0.0000228010500092077, -0.000024129945561891588,
+			-8.489496089740964e-6
+		]
+	]
+
+	numpy.testing.assert_allclose(costs(pt_to_test), expected_cost, err_msg="Costs aren't as expected.")
+	numpy.testing.assert_allclose(jac(pt_to_test), expected_jacobian, err_msg="Jacobian aren't as expected.")