#!/usr/bin/env python3 """ Fitting example: fit with masks """ from matplotlib import pyplot as plt import bornagain as ba from bornagain import deg, nm, ba_fitmonitor import lmfit import model1_cylinders as model def get_masked_simulation(P, add_masks=True): simulation = model.get_simulation(P) if add_masks: add_mask_to_simulation(simulation) return simulation def add_mask_to_simulation(simulation): """ Here we demonstrate how to add masks to the simulation. Only unmasked areas will be simulated and then used during the fit. Masks can have different geometrical shapes (ba.Rectangle, ba.Ellipse, Line) with the mask value either "True" (detector bin is excluded from the simulation) or False (will be simulated). Every subsequent mask overrides previously defined mask in this area. In the code below we put masks in such way that simulated image will look like a Pac-Man from ancient arcade game. """ # mask all detector (put mask=True to all detector channels) simulation.detector().maskAll() w2 = 0.7*deg # half width of detector frame in ?? # set mask to simulate pacman's head simulation.detector().addMask(ba.Ellipse(1*w2, 1*w2, 0.5*w2, 0.5*w2), False) # set mask for pacman's eye simulation.detector().addMask(ba.Ellipse(1.11*w2, 1.25*w2, 0.05*w2, 0.05*w2), True) # set mask for pacman's mouth points = [[1*w2, 1*w2], [1.5*w2, 1.2*w2], [1.5*w2, 0.8*w2], [1*w2, 1*w2]] simulation.detector().addMask(ba.Polygon(points), True) # giving pacman something to eat simulation.detector().addMask( ba.Rectangle(1.45*w2, 0.95*w2, 1.55*w2, 1.05*w2), False) simulation.detector().addMask( ba.Rectangle(1.61*w2, 0.95*w2, 1.71*w2, 1.05*w2), False) simulation.detector().addMask( ba.Rectangle(1.75*w2, 0.95*w2, 1.85*w2, 1.05*w2), False) if __name__ == '__main__': data = model.fake_data().noisy(0.1, 0.1) fit_objective = ba.FitObjective() fit_objective.addFitPair(get_masked_simulation, data, 1) fit_objective.initPrint(10) observer = ba_fitmonitor.PlotterGISAS() fit_objective.initPlot(10, observer) P = lmfit.Parameters() P.add("radius", 6.*nm, min=4, max=8) P.add("height", 9.*nm, min=8, max=12) result = lmfit.minimize(fit_objective.evaluate_residuals, P) print(lmfit.fit_report(result)) finalP = result.params.valuesdict() ba.showSample3D(model.get_sample(finalP), sample_size=120*nm, seed=0) plt.show()