I read in volume III of Landau's and Lifshitz's Theoretical Physics that the action of classical mechanics can be seen as the complex phase of the wave function of quantum mechanics. I also read a reference somewhere else, that it was in a 1948 paper1 that Richard Feynman demonstrated this relationship. Landau and Lifshitz were not very informative, and I had no access to Feynman's old paper (I've since obtained a copy, but it's a Good Thing that I didn't have it earlier, as it gave me the motivation to do these calculations on my own), so I decided to work this out myself, at least in the simple case of a point particle in a potential field.
The Schrödinger equation of a point particle is well known:
iħ ∂φ= – ħ² ∂²φ+ Vφ ∂t 2m ∂q²
What if φ is expressed as ρeiS/ħ, where S is an arbitrary function of time, the coordinates, and perhaps momenta? Let's do the substitution, dropping the eiS/ħ part as it appears as a factor on both sides of the equation:
iħ ∂ρ– ρ ∂S= – ħ²┌
│
│
│
└ ∂²ρ+ 2i ∂ρ ∂S+ iρ ∂²S– ρ┌
│
│
└ ∂S┐
│
│
┘2 ┐
│
│
│
┘+ Vρ ∂t ∂t 2m ∂q² ħ ∂q ∂q ħ ∂q² ħ² ∂q
Separating the real and imaginary parts yields two equations. The equation for the real part looks like this:
– ρ ∂S= – ħ²┌
│
│
│
└ ∂²ρ– ρ┌
│
│
└ ∂S┐
│
│
┘2 ┐
│
│
│
┘+ Vρ ∂t 2m ∂q² ħ² ∂q
which, when divided by ρ, can be simplified as
– ∂S= 1┌
│
│
└ ∂S┐2
│
│
┘– ħ² ∂²ρ+ V ∂t 2m ∂q 2mρ ∂q²
The smallness of ħ suggests that, at least for macroscopic systems, the second term on the right hand side can be eliminated, leaving us with the equation
– ∂S= 1┌
│
│
└ ∂S┐2
│
│
┘+ V ∂t 2m ∂q
which is the equation of motion for a point particle governed by the action S.
As a footnote of sorts, I now found that this relationship is also demonstrated in §§31-32 of Dirac's book2.
1R. P. Feynman: "Space-time Approach to Non-relativistic Quantum
Mechanics", Rev. Mod. Physics, Vol 20, pp. 367-387, April 1948
2P. A. M. Dirac: "The Principles of Quantum Mechanics", Oxford
University Press, 1958