Paranoia

Today I got paranoid with implicit conversions.

I was sure that iu would coerce iu * j to a complex type, divided by a integer would still be of a complex type.

But somehow I was getting a zeroed array in my output, so I got afraid I was having a integer division problem.

! d/dky F(k) = - i * ky * F(k)
do j = -Gy, Gy
    iky = iu * j / (2 * Gy + 1)
    dF(:,j,:) = - iky * F(:,j,:)
end do

I tried group the multiplication this way I would force the type coercion, right? Still didn’t worked.

! d/dky F(k) = - i * ky * F(k)
do j = -Gy, Gy
    iky = (iu * j) / (2 * Gy + 1)
    dF(:,j,:) = - iky * F(:,j,:)
end do

So I got paranoid and made a explicit conversion…

iky = iu * real(j,wp) / (2 * Gy + 1)

it didn’t worked either! So I got even MORE paranoid explicit conversions everywhere…

iky = iu * real(j,wp) / real(2 * Gy + 1,wp)

Started checking every other possibility… everything was a lie!

iky = iu * real(j,wp) / (2.0_wp * Gy + 1.0_wp)

Until I finally saw the type of iky:

real(wp) :: ikx, iky, ikz