Non-Overlapping Wave Functions/Charge Densities

This is a follow up of the recent post on The Secret of Covalent Bonding with further computations comparing the non-overlapping wave functions/charge densities of RealQM meeting with (i) continuity and zero derivative, with a hypothetical case of instead (ii) zero density. 

To pin-point the essential aspect, we consider the following 1d model problem of an atom with N electrons: 

Find the function $\Psi (x)$ on the interval $[0,1]$ which minimises the energy $E=E_k + E_p$ with

• $E_p=\frac{1}{2}\int_0^1D\psi^2dx$             (kinetic energy)
• $E_k =-\int_0^1P(x)\psi (x)$           (potential energy)

over wave functions $\psi (x)$ with $D\psi =\frac{d\psi}{dx}$ the derivative, with total charge 

• $\int_0^1\psi^2dx = N$,

where $-P(x)$ is a given potential.

The Ansatz of RealQM is to seek $\Psi (x)$ on the form 

•  $\Psi (x) = \Psi_1(x) + \Psi_2(x) + .... + \Psi_N(x)$,

where the $\Psi_i(x)$ are one-electron wave functions with disjoint supports which meet on a Bernoulli free boundary with continuity and zero derivative. Running this code in a case with $N=4$ and $P(x)$ the potential from two kernels, we get the following result for case (i): 

We see in red 4 non-overlapping wave functions meeting at a free boundary marked by green with continuity and zero derivative. The total energy is -247 with kinetic energy 30. 

We run the same code but require wave-functions to vanish at meeting points and get for case (ii):

We see total energy larger total energy -231 with much larger kinetic energy 308. 

We understand that requiring wave functions to vanish as in (ii) when meeting, gives much larger kinetic energy than asking only for continuity and zero derivate as in (i), and that the net effect is larger total energy.

The purpose of this exercise is to exhibit the possibility opened by RealQM of electron concentration between kernels decreasing potential energy without balancing increase of kinetic energy, which appears as the holy grail of covalent bonding. 

In StdQM wave functions have global support, which may be overlapping, and so like the functions in (ii) above have to pay a price of added kinetic energy from charge concentration which is less favourable for binding than RealQM.

The finding of this post agrees with the real case considered in previous posts.