Пример #1
0
class SersicEllipseKappa(LensProfileBase):
    """
    this class contains the function and the derivatives of an elliptical sersic profile
    with the ellipticity introduced in the convergence (not the potential).

    This requires the use of numerical integrals (Keeton 2004)
    """
    param_names = [
        'k_eff', 'R_sersic', 'n_sersic', 'e1', 'e2', 'center_x', 'center_y'
    ]
    lower_limit_default = {
        'k_eff': 0,
        'R_sersic': 0,
        'n_sersic': 0.5,
        'e1': -0.5,
        'e2': -0.5,
        'center_x': -100,
        'center_y': -100
    }
    upper_limit_default = {
        'k_eff': 10,
        'R_sersic': 100,
        'n_sersic': 8,
        'e1': 0.5,
        'e2': 0.5,
        'center_x': 100,
        'center_y': 100
    }

    def __init__(self):

        self._sersic = Sersic()
        super(SersicEllipseKappa, self).__init__()

    def function(self,
                 x,
                 y,
                 n_sersic,
                 R_sersic,
                 k_eff,
                 e1,
                 e2,
                 center_x=0,
                 center_y=0):

        raise Exception('not yet implemented')

        # phi_G, q = param_util.ellipticity2phi_q(e1, e2)
        #
        # if isinstance(x, float) and isinstance(y, float):
        #
        #     x_, y_ = self._coord_rotate(x, y, phi_G, center_x, center_y)
        #     integral = quad(self._integrand_I, 0, 1, args=(x_, y_, q, n_sersic, R_sersic, k_eff, center_x, center_y))[0]
        #
        # else:
        #
        #     assert isinstance(x, np.ndarray) or isinstance(x, list)
        #     assert isinstance(y, np.ndarray) or isinstance(y, list)
        #     x = np.array(x)
        #     y = np.array(y)
        #     shape0 = x.shape
        #     assert shape0 == y.shape
        #
        #     if isinstance(phi_G, float) or isinstance(phi_G, int):
        #         phiG = np.ones_like(x) * float(phi_G)
        #         q = np.ones_like(x) * float(q)
        #     integral = []
        #     for i, (x_i, y_i, phi_i, q_i) in \
        #             enumerate(zip(x.ravel(), y.ravel(), phiG.ravel(), q.ravel())):
        #
        #         integral.append(quad(self._integrand_I, 0, 1, args=(x_, y_, q, n_sersic,
        #                                                             R_sersic, k_eff, center_x, center_y))[0])
        #
        #
        # return 0.5 * q * integral

    def derivatives(self,
                    x,
                    y,
                    n_sersic,
                    R_sersic,
                    k_eff,
                    e1,
                    e2,
                    center_x=0,
                    center_y=0):

        phi_G, gam = param_util.shear_cartesian2polar(e1, e2)
        q = max(1 - gam, 0.00001)

        x, y = self._coord_rotate(x, y, phi_G, center_x, center_y)

        if isinstance(x, float) and isinstance(y, float):

            alpha_x, alpha_y = self._compute_derivative_atcoord(
                x,
                y,
                n_sersic,
                R_sersic,
                k_eff,
                phi_G,
                q,
                center_x=center_x,
                center_y=center_y)

        else:

            assert isinstance(x, np.ndarray) or isinstance(x, list)
            assert isinstance(y, np.ndarray) or isinstance(y, list)
            x = np.array(x)
            y = np.array(y)
            shape0 = x.shape
            assert shape0 == y.shape

            alpha_x, alpha_y = np.empty_like(x).ravel(), np.empty_like(
                y).ravel()

            if isinstance(phi_G, float) or isinstance(phi_G, int):
                phiG = np.ones_like(alpha_x) * float(phi_G)
                q = np.ones_like(alpha_x) * float(q)

            for i, (x_i, y_i, phi_i, q_i) in \
                    enumerate(zip(x.ravel(), y.ravel(), phiG.ravel(), q.ravel())):

                fxi, fyi = self._compute_derivative_atcoord(x_i,
                                                            y_i,
                                                            n_sersic,
                                                            R_sersic,
                                                            k_eff,
                                                            phi_i,
                                                            q_i,
                                                            center_x=center_x,
                                                            center_y=center_y)

                alpha_x[i], alpha_y[i] = fxi, fyi

            alpha_x = alpha_x.reshape(shape0)
            alpha_y = alpha_y.reshape(shape0)

        alpha_x, alpha_y = self._coord_rotate(alpha_x, alpha_y, -phi_G, 0, 0)

        return alpha_x, alpha_y

    def hessian(self,
                x,
                y,
                n_sersic,
                R_sersic,
                k_eff,
                e1,
                e2,
                center_x=0,
                center_y=0):
        """
        returns Hessian matrix of function d^2f/dx^2, d^2/dxdy, d^2/dydx, d^f/dy^2
        """
        alpha_ra, alpha_dec = self.derivatives(x, y, n_sersic, R_sersic, k_eff,
                                               e1, e2, center_x, center_y)
        diff = 0.000001
        alpha_ra_dx, alpha_dec_dx = self.derivatives(x + diff, y, n_sersic,
                                                     R_sersic, k_eff, e1, e2,
                                                     center_x, center_y)
        alpha_ra_dy, alpha_dec_dy = self.derivatives(x, y + diff, n_sersic,
                                                     R_sersic, k_eff, e1, e2,
                                                     center_x, center_y)

        f_xx = (alpha_ra_dx - alpha_ra) / diff
        f_xy = (alpha_ra_dy - alpha_ra) / diff
        f_yx = (alpha_dec_dx - alpha_dec) / diff
        f_yy = (alpha_dec_dy - alpha_dec) / diff

        return f_xx, f_xy, f_yx, f_yy

    def projected_mass(self, x, y, q, n_sersic, R_sersic, k_eff, u=1, power=1):

        b_n = self._sersic.b_n(n_sersic)

        elliptical_coord = self._elliptical_coord_u(x, y, u, q)**power
        elliptical_coord *= R_sersic**-power

        exponent = -b_n * (elliptical_coord**(1. / n_sersic) - 1)

        return k_eff * np.exp(exponent)

    def _integrand_J(self, u, x, y, n_sersic, q, R_sersic, k_eff, n_integral):

        kappa = self.projected_mass(x,
                                    y,
                                    q,
                                    n_sersic,
                                    R_sersic,
                                    k_eff,
                                    u=u,
                                    power=1)

        power = -(n_integral + 0.5)

        return kappa * (1 - (1 - q**2) * u)**power

    def _integrand_I(self, u, x, y, q, n_sersic, R_sersic, keff, centerx,
                     centery):

        ellip_coord = self._elliptical_coord_u(x, y, u, q)

        def_angle_circular = self._sersic.alpha_abs(ellip_coord, 0, n_sersic,
                                                    R_sersic, keff, centerx,
                                                    centery)

        return ellip_coord * def_angle_circular * (
            1 - (1 - q**2) * u)**-0.5 * u**-1

    def _compute_derivative_atcoord(self,
                                    x,
                                    y,
                                    n_sersic,
                                    R_sersic,
                                    k_eff,
                                    phi_G,
                                    q,
                                    center_x=0,
                                    center_y=0):

        alpha_x = x * q * quad(self._integrand_J,
                               0,
                               1,
                               args=(x, y, n_sersic, q, R_sersic, k_eff, 0))[0]
        alpha_y = y * q * quad(self._integrand_J,
                               0,
                               1,
                               args=(x, y, n_sersic, q, R_sersic, k_eff, 1))[0]

        return alpha_x, alpha_y

    @staticmethod
    def _elliptical_coord_u(x, y, u, q):

        fac = 1 - (1 - q**2) * u

        return (u * (x**2 + y**2 * fac**-1))**0.5

    @staticmethod
    def _coord_rotate(x, y, phi_G, center_x, center_y):

        x_shift = x - center_x
        y_shift = y - center_y
        cos_phi = np.cos(phi_G)
        sin_phi = np.sin(phi_G)

        x_ = cos_phi * x_shift + sin_phi * y_shift
        y_ = -sin_phi * x_shift + cos_phi * y_shift

        return x_, y_
Пример #2
0
class TestSersic(object):
    """
    tests the Gaussian methods
    """
    def setup(self):

        self.sersic_2 = SersicEllipseKappa()
        self.sersic = Sersic()
        self.sersic_light = Sersic_light()

    def test_function(self):

        x = 1
        y = 2
        n_sersic = 2.
        R_sersic = 1.
        k_eff = 0.2
        values = self.sersic.function(x, y, n_sersic, R_sersic, k_eff)
        npt.assert_almost_equal(values, 1.0272982586319199, decimal=10)

        x = np.array([0])
        y = np.array([0])
        values = self.sersic.function(x, y, n_sersic, R_sersic, k_eff)
        npt.assert_almost_equal(values[0], 0., decimal=9)

        x = np.array([2, 3, 4])
        y = np.array([1, 1, 1])
        values = self.sersic.function(x, y, n_sersic, R_sersic, k_eff)

        npt.assert_almost_equal(values[0], 1.0272982586319199, decimal=10)
        npt.assert_almost_equal(values[1], 1.3318743892966658, decimal=10)
        npt.assert_almost_equal(values[2], 1.584299393114988, decimal=10)

    def test_derivatives(self):
        x = np.array([1])
        y = np.array([2])
        n_sersic = 2.
        R_sersic = 1.
        k_eff = 0.2
        f_x, f_y = self.sersic.derivatives(x, y, n_sersic, R_sersic, k_eff)
        f_x2, f_y2 = self.sersic_2.derivatives(x, y, n_sersic, R_sersic, k_eff,
                                               0, 0.00000001)

        assert f_x[0] == 0.16556078301997193
        assert f_y[0] == 0.33112156603994386
        npt.assert_almost_equal(f_x2[0], f_x[0])
        npt.assert_almost_equal(f_y2[0], f_y[0])

        x = np.array([0])
        y = np.array([0])
        f_x, f_y = self.sersic.derivatives(x, y, n_sersic, R_sersic, k_eff)
        f_x2, f_y2 = self.sersic_2.derivatives(x, y, n_sersic, R_sersic, k_eff,
                                               0, 0.00000001)
        assert f_x[0] == 0
        assert f_y[0] == 0
        npt.assert_almost_equal(f_x2[0], f_x[0])
        npt.assert_almost_equal(f_y2[0], f_y[0])

        x = np.array([1, 3, 4])
        y = np.array([2, 1, 1])
        values = self.sersic.derivatives(x, y, n_sersic, R_sersic, k_eff)
        values2 = self.sersic_2.derivatives(x, y, n_sersic, R_sersic, k_eff, 0,
                                            0.00000001)
        assert values[0][0] == 0.16556078301997193
        assert values[1][0] == 0.33112156603994386
        assert values[0][1] == 0.2772992378623737
        assert values[1][1] == 0.092433079287457892
        npt.assert_almost_equal(values2[0][0], values[0][0])
        npt.assert_almost_equal(values2[1][0], values[1][0])
        npt.assert_almost_equal(values2[0][1], values[0][1])
        npt.assert_almost_equal(values2[1][1], values[1][1])

        values2 = self.sersic_2.derivatives(0.3, -0.2, n_sersic, R_sersic,
                                            k_eff, 0, 0.00000001)
        values = self.sersic.derivatives(0.3, -0.2, n_sersic, R_sersic, k_eff,
                                         0, 0.00000001)
        npt.assert_almost_equal(values2[0], values[0])
        npt.assert_almost_equal(values2[1], values[1])

    def test_differentails(self):
        x_, y_ = 1., 1
        n_sersic = 2.
        R_sersic = 1.
        k_eff = 0.2
        r = np.sqrt(x_**2 + y_**2)

        d_alpha_dr = self.sersic.d_alpha_dr(x_, y_, n_sersic, R_sersic, k_eff)
        alpha = self.sersic.alpha_abs(x_, y_, n_sersic, R_sersic, k_eff)

        f_xx_ = d_alpha_dr * calc_util.d_r_dx(
            x_, y_) * x_ / r + alpha * calc_util.d_x_diffr_dx(x_, y_)
        f_yy_ = d_alpha_dr * calc_util.d_r_dy(
            x_, y_) * y_ / r + alpha * calc_util.d_y_diffr_dy(x_, y_)
        f_xy_ = d_alpha_dr * calc_util.d_r_dy(
            x_, y_) * x_ / r + alpha * calc_util.d_x_diffr_dy(x_, y_)

        f_xx = (d_alpha_dr / r - alpha / r**2) * y_**2 / r + alpha / r
        f_yy = (d_alpha_dr / r - alpha / r**2) * x_**2 / r + alpha / r
        f_xy = (d_alpha_dr / r - alpha / r**2) * x_ * y_ / r
        npt.assert_almost_equal(f_xx, f_xx_, decimal=10)
        npt.assert_almost_equal(f_yy, f_yy_, decimal=10)
        npt.assert_almost_equal(f_xy, f_xy_, decimal=10)

    def test_hessian(self):
        x = np.array([1])
        y = np.array([2])
        n_sersic = 2.
        R_sersic = 1.
        k_eff = 0.2
        f_xx, f_xy, f_yx, f_yy = self.sersic.hessian(x, y, n_sersic, R_sersic,
                                                     k_eff)
        assert f_xx[0] == 0.1123170666045793
        npt.assert_almost_equal(f_yy[0], -0.047414082641598576, decimal=10)
        npt.assert_almost_equal(f_xy[0], -0.10648743283078525, decimal=10)
        npt.assert_almost_equal(f_xy, f_yx, decimal=5)
        x = np.array([1, 3, 4])
        y = np.array([2, 1, 1])
        values = self.sersic.hessian(x, y, n_sersic, R_sersic, k_eff)
        assert values[0][0] == 0.1123170666045793
        npt.assert_almost_equal(values[3][0],
                                -0.047414082641598576,
                                decimal=10)
        npt.assert_almost_equal(values[1][0], -0.10648743283078525, decimal=10)
        npt.assert_almost_equal(values[0][1],
                                -0.053273787681591328,
                                decimal=10)
        npt.assert_almost_equal(values[3][1], 0.076243427402007985, decimal=10)
        npt.assert_almost_equal(values[1][1],
                                -0.048568955656349749,
                                decimal=10)

        f_xx2, f_xy2, f_yx2, f_yy2 = self.sersic_2.hessian(
            x, y, n_sersic, R_sersic, k_eff, 0.0000001, 0)
        npt.assert_almost_equal(f_xx2, values[0])
        npt.assert_almost_equal(f_yy2, values[3], decimal=6)
        npt.assert_almost_equal(f_xy2, values[1], decimal=6)
        npt.assert_almost_equal(f_yx2, values[2], decimal=6)

    def test_alpha_abs(self):
        x = 1.
        dr = 0.0000001
        n_sersic = 2.5
        R_sersic = .5
        k_eff = 0.2
        alpha_abs = self.sersic.alpha_abs(x, 0, n_sersic, R_sersic, k_eff)
        f_dr = self.sersic.function(x + dr, 0, n_sersic, R_sersic, k_eff)
        f_ = self.sersic.function(x, 0, n_sersic, R_sersic, k_eff)
        alpha_abs_num = -(f_dr - f_) / dr
        npt.assert_almost_equal(alpha_abs_num, alpha_abs, decimal=3)

    def test_dalpha_dr(self):
        x = 1.
        dr = 0.0000001
        n_sersic = 1.
        R_sersic = .5
        k_eff = 0.2
        d_alpha_dr = self.sersic.d_alpha_dr(x, 0, n_sersic, R_sersic, k_eff)
        alpha_dr = self.sersic.alpha_abs(x + dr, 0, n_sersic, R_sersic, k_eff)
        alpha = self.sersic.alpha_abs(x, 0, n_sersic, R_sersic, k_eff)
        d_alpha_dr_num = (alpha_dr - alpha) / dr
        npt.assert_almost_equal(d_alpha_dr, d_alpha_dr_num, decimal=3)

    def test_mag_sym(self):
        """

        :return:
        """
        r = 2.
        angle1 = 0.
        angle2 = 1.5
        x1 = r * np.cos(angle1)
        y1 = r * np.sin(angle1)

        x2 = r * np.cos(angle2)
        y2 = r * np.sin(angle2)
        n_sersic = 4.5
        R_sersic = 2.5
        k_eff = 0.8
        f_xx1, f_xy1, f_yx1, f_yy1 = self.sersic.hessian(
            x1, y1, n_sersic, R_sersic, k_eff)
        f_xx2, f_xy2, f_yx2, f_yy2 = self.sersic.hessian(
            x2, y2, n_sersic, R_sersic, k_eff)
        kappa_1 = (f_xx1 + f_yy1) / 2
        kappa_2 = (f_xx2 + f_yy2) / 2
        npt.assert_almost_equal(kappa_1, kappa_2, decimal=10)
        A_1 = (1 - f_xx1) * (1 - f_yy1) - f_xy1 * f_yx1
        A_2 = (1 - f_xx2) * (1 - f_yy2) - f_xy2 * f_yx2
        npt.assert_almost_equal(A_1, A_2, decimal=10)

    def test_convergernce(self):
        """
        test the convergence and compares it with the original Sersic profile
        :return:
        """
        x = np.array([0, 0, 0, 0, 0])
        y = np.array([0.5, 1, 1.5, 2, 2.5])
        n_sersic = 4.5
        R_sersic = 2.5
        k_eff = 0.2
        f_xx, f_xy, f_yx, f_yy = self.sersic.hessian(x, y, n_sersic, R_sersic,
                                                     k_eff)
        kappa = (f_xx + f_yy) / 2.
        assert kappa[0] > 0
        flux = self.sersic_light.function(x,
                                          y,
                                          amp=1.,
                                          R_sersic=R_sersic,
                                          n_sersic=n_sersic)
        flux /= flux[0]
        kappa /= kappa[0]
        npt.assert_almost_equal(flux[1], kappa[1], decimal=5)

        xvalues = np.linspace(0.5, 3., 100)

        e1, e2 = 0.4, 0.
        q = ellipticity2phi_q(e1, e2)[1]
        kappa_ellipse = self.sersic_2.projected_mass(xvalues, 0, q, n_sersic,
                                                     R_sersic, k_eff)
        fxx, _, _, fyy = self.sersic_2.hessian(xvalues, 0, n_sersic, R_sersic,
                                               k_eff, e1, e2)

        npt.assert_almost_equal(kappa_ellipse, 0.5 * (fxx + fyy), decimal=5)

    def test_sersic_util(self):
        n = 1.
        Re = 2.
        k, bn = self.sersic.k_bn(n, Re)
        Re_new = self.sersic.k_Re(n, k)
        assert Re == Re_new