profile.PSF : brentq root-finder should start to 0 in this case

[ghost]

rmax finding should start from zero, peaked PSF function such as a Gaussian fade too quickly. the attached example psffunc returns an error because its rmax is less than 1 arcmin

def gaussian(x): FWHM=5./60. # Full Width Half Maximum in arcmin sigma=FWHM/2.35 # standard deviation as a function of the FWHM return np.exp(-x**2/(2*sigma**2))

[botteon]

I am also experiencing issues with a gaussian PSF. The error (brentq related) is ValueError: f(a) and f(b) must have different signs. However the solution proposed does not solve the problem

[domeckert]

Thanks @botteon , I'll have a look