import numpy as np import scipy.integrate as integrate import scipy.special import matplotlib.pyplot as plt import matplotlib as mpl def sphere(q, R): return 9*((np.sin(q*R)-(q*R)*np.cos(q*R))/(q*R)**3)**2 def thin_rod(q, R): return (2*si(q*R)/(q*R) - (sinc(q*R/2))**2) def thin_disc(q, R): return (2-2*scipy.special.jv(1, 2*q*R)/(q*R))/(q*R)**2 def sinc(x): return np.sinc(x/np.pi) def si(t): if not isinstance(t, float): int_res = np.empty_like(t) for j in range(len(t)): int_res[j] = integrate.quad(sinc, 0, t[j])[0] return int_res else: return integrate.quad(sinc, 0, t)[0] rg = 10 q = np.linspace(0.0001, np.sqrt(4)/rg, 1000) # q = np.linspace(0.0001, 3, 1000) r_sphere = np.sqrt(5/3.)*rg r_rod = np.sqrt(12.)*rg r_disc = np.sqrt(2.)*rg sphere_i = sphere(q, r_sphere) rod_i = thin_rod(q, r_rod) disc_i = thin_disc(q, r_disc) guinier_i = np.exp(-(q*rg)**2/3.) sphere_i = sphere_i/sphere_i[0] rod_i = rod_i/rod_i[0] disc_i = disc_i/disc_i[0] guinier_i = guinier_i/guinier_i[0] def tick_formatter(x, pos): return "{}".format(round(x,1)) plt.figure(figsize=(6,4)) plt.axvline(1**2, color='0.6', linestyle='--') plt.axvline(1.3**2, color='0.6', linestyle='--') plt.semilogy((q*rg)**2, guinier_i, label='Guinier Approx.', color='k') plt.semilogy((q*rg)**2, sphere_i, label='Sphere') plt.semilogy((q*rg)**2, rod_i, label='Thin Rod') plt.semilogy((q*rg)**2, disc_i, label='Thin Disc') plt.legend() plt.xlabel('$(qR_g)^2$') plt.ylabel('Intensity (log scale)') plt.title('Guinier approximation validity') plt.subplots_adjust(left=0.11, right=0.96, top=0.94, bottom=0.12) axes = plt.gca() axes.yaxis.set_major_formatter(mpl.ticker.FuncFormatter(tick_formatter)) axes.yaxis.set_minor_formatter(mpl.ticker.FuncFormatter(tick_formatter)) plt.savefig('guinier_shapes.png', dpi=300) plt.show() # plt.semilogy(q, sphere_i) # plt.semilogy(q, rod_i) # plt.semilogy(q, disc_i) # # plt.semilogy(q, guinier_i) # plt.show() # err = np.random.randn(1000) # err = err/100. # print(err+1) # np.savetxt('sphere.dat', np.column_stack((q, sphere_i*(err+1), err))) # np.savetxt('rod.dat', np.column_stack((q, rod_i*(err+1), err))) # np.savetxt('disc.dat', np.column_stack((q, disc_i*(err+1), err)))