Problem with a little big ray parameter about rf_inv

[SuwenJunliu]

Hi Yangjudiao,

I checked the program according to your parameter, and I am pretty sure that it was caused by the e_inverse function in forward.f90

    e_inv(:,:) = (0.d0, 0.d0)
    eta = sqrt(1.d0/(beta*beta) - p*p)
    xi  = sqrt(1.d0/(alpha*alpha) - p*p)
    bp = 1.d0 - 2.d0*beta*beta*p*p

If the p is too large, the 1.d0/(beta*beta) - p*p will become a negative number.
You could check the Aki & Richards, pp. 158, Eq. (5.65), or just the hint by Akuhara Aki & Richards, pp. 161, Eq. (5.71), maybe just change all those eta and xi to complex number

    real(8) :: bp
    !real(8) :: eta, xi, bp
    complex(8) :: eta, xi
    
    e_inv(:,:) = (0.d0, 0.d0)
    eta = sqrt(cmplx(1.d0/(beta*beta) - p*p, 0))
    xi  = sqrt(cmplx(1.d0/(alpha*alpha) - p*p, 0))
    bp = 1.d0 - 2.d0*beta*beta*p*p

But do we really need to deal with those much larger rayp?