Download this notebook:
Numerov.nb
The Numerov approach is a reasonably robust proceedure for solving the
time independent Schrodinger equation (in 1 D). The proceedure works by
specifying an initial value of ψ and its derivative at a
point and propagating on a spatial grid to some final
point. The basic equations of "motion" are
derived by simply taking a Taylor series expansion about the point
. The estimated value of the function at
some other nearby point,
or
is given by
![]()
![]()
Adding Eq. 1 and Eq. 2 and solving for f[n+1] yields to
![]()
Let f be a solution of the Schrodinger equation: (H - E) f = 0 . Thus, we can obtain the second derivative of f using the operator equation
![]()
where
![]()
expanding f'', yields
![]()
where we have truncted the series at
. Now, solving for
and substituting f'' = Gf, yields
![[Graphics:Images/index_gr_13.gif]](Images/index_gr_13.gif)
which is our working equation. Below are two examples of the Numerov method in action.
Here we use the Numerov proceedure to try to find the eigenstates of a Harmonic well. We know where the states should be a priori, so we can use this as a check of the proceedure. Our proceedure is to pick an energy and propagate a solution. A genuine eigenstate solution of this well will go to zero as x->±∞. So, if we see the wavefunction solution blowing up, we know we do not have a solution at our choice of energy. As it turns out, getting the exact solution is almost impossible. So, we will try to zero in on the solutions by counting the number of nodes in the final wavefunction. If as we increase the energy by one step we gain one additional node in the wave, we know that there must be a stable solution between the two choices of energy.
The Schrodinger equation this case is
![]()
and we will use atomic units throughout (ℏ = 1,
= 1).
![[Graphics:Images/index_gr_16.gif]](Images/index_gr_16.gif)
![[Graphics:Images/index_gr_18.gif]](Images/index_gr_18.gif)
![[Graphics:Images/index_gr_20.gif]](Images/index_gr_20.gif)
![[Graphics:Images/index_gr_22.gif]](Images/index_gr_22.gif)
![[Graphics:Images/index_gr_24.gif]](Images/index_gr_24.gif)
Same proceedure as before, except in spherical coordinates. The
system here is H + Kr scattering with potential and parameters as given in
Thijssen's book. For scattering, the Numerov proceedure is a
more preferable way to go about business since we only need to satisfy one
boundary condition at r = 0. As r -> ∞, we still need to
satisfy scattering boundary conditions, however, this we can do by simply
matching the log derivative of the Numerov solution to an asymptotic form
of
. This is how we will compute the phase
shifts and cross-sections.
The Schrodinger equation we solve in this case is in radial coordinates:
![]()
![[Graphics:Images/index_gr_27.gif]](Images/index_gr_27.gif)
![[Graphics:Images/index_gr_28.gif]](Images/index_gr_28.gif)
![[Graphics:Images/index_gr_29.gif]](Images/index_gr_29.gif)
![[Graphics:Images/index_gr_30.gif]](Images/index_gr_30.gif)
![[Graphics:Images/index_gr_33.gif]](Images/index_gr_33.gif)
![[Graphics:Images/index_gr_34.gif]](Images/index_gr_34.gif)
![[Graphics:Images/index_gr_37.gif]](Images/index_gr_37.gif)
![[Graphics:Images/index_gr_38.gif]](Images/index_gr_38.gif)
![[Graphics:Images/index_gr_41.gif]](Images/index_gr_41.gif)
![[Graphics:Images/index_gr_42.gif]](Images/index_gr_42.gif)
![[Graphics:Images/index_gr_58.gif]](Images/index_gr_58.gif)
![[Graphics:Images/index_gr_74.gif]](Images/index_gr_74.gif)
![[Graphics:Images/index_gr_90.gif]](Images/index_gr_90.gif)
![[Graphics:Images/index_gr_106.gif]](Images/index_gr_106.gif)
![[Graphics:Images/index_gr_122.gif]](Images/index_gr_122.gif)
Text and code copyright @ 2000 by Eric R. Bittner and University of Houston, all rights reserved. No part of this document may be used without the prior permission of the author.