How to recover non-relativistic limit of the Dirac equation $$\left( i\gamma^{\mu}\mathcal{D}_{\mu} - m \right)\Psi(x) = 0$$ where $\mathcal{D}_{\mu} = \partial_{\mu} + iqA_{\mu}$. I do not assume that the vector potential $A_{\mu}$ is zero, but I have a particle in central potential and magnetic field. During derivation please highlight where does the non-relativistic assumptions is used and how. I have found some notes here, but I don't understand why can we Taylor expand this expression $$\frac{1}{E + m -qA^0}\vec{\sigma}\cdot(\vec{p} - q\vec{A}) \approx \frac{1}{E+m}\vec{\sigma}\cdot(\vec{p} - q\vec{A}) + \frac{1}{(E+m)^2}qA^0\vec{\sigma}\cdot\vec{p}$$ around $qA^0 = 0$ and neglect $\vec{A}$ in the second term.

  • 1
    $\begingroup$ The reason for neglecting $\vec A$ is that the magnetic field $\nabla \times \vec A$ is a factor of $v/c$ smaller than the electric field $-\nabla A^0$ (for a static vector potential). See also the answer to Why are magnetic fields so much weaker than electric? $\endgroup$
    – Praan
    Dec 1, 2015 at 23:12

1 Answer 1


The essential reason is that the "$E$" in your equation has no non-relativistic limit. This can be seen more clearly by separating out the rest-mass-energy $mc^2$ and writing $E = mc^2 + H$, with $H$ comprising the total of the kinetic and internal energy. The "$mc^2$" part blows up to infinity as $c$ goes to infinity.

On the other hand $H$ has a non-relativistic limit. At the same time, the mass in non-relativistic theory can be treated as the non-relativistic limit of the "moving mass" $M = E/c^2$. So, to get a non-relativistic version of Dirac requires actually extending the Poincaré group from 10 dimensions to 11, by replacing $E$ by $M$ and $H$, treating them as independent.

So, the confusion arises from confusing $E$ and $H$. The shifting from $E$ to $H$ is precisely what is done with the Foldy-Wouthuysen transformation.

In place of the mass shell identity $$\frac{E^2}{c^2} - |𝐩|^2 = (mc)^2,$$ you then have two identities: $$|𝐩|^2 - 2MH + \frac{1}{c^2}H^2 = 0, \quad M - \frac{1}{c^2}H = m.$$ If you solve for $H$, you will see more clearly the connection to non-relativistic theory ... and you will also see more clearly what that extra 11th dimension in the symmetry group reflects: $$H = \frac{|𝐩|^2}{m + M}.$$ Compare to the non-relativistic case: $$H = \frac{|𝐩|^2}{2m} + U$$ where $U$ is the internal energy - i.e. the "kinetic energy" arising from the kinetics of the internal constituency of a body ... except now, it's being generalized and treated as a thing in its own right that potentially applies even to elementary bodies. So, it's the rest frame value of $H$, just as $m$ (the rest-mass) is the rest frame value of $M$. Substituting into the equations for the two invariants leads to the following modified form: $$|𝐩|^2 - 2MH + \frac{1}{c^2}H^2 = -2mU + \frac{1}{c^2}U^2, \quad μ ≡ M - \frac{1}{c^2}H = m - \frac{1}{c^2}U.$$ The drop-down from extended Poincaré-11 to ordinary Poincaré-10 then corresponds to setting $U = 0$.

On the other hand, the non-relativistic limit corresponds to setting $1/c^2$ to $0$. The non-relativistic limit of $M$, $m$ and $μ$ are all $m$, the mass, with all 11 dimensions of the symmetry group kept intact.

In Relativity, the following is invariant: $$E dt - 𝐩·d𝐫,$$ where $d𝐫 = (dx,dy,dz)$. If you make the above replacements, you get $$H dt - 𝐩·d𝐫 + μc^2 dt.$$ To make a meaningful non-relativistic limit out of this, you have to throw in the proper time, which we can denote $s$ and rewrite it as: $$H dt - 𝐩·d𝐫 + μc^2 dt - μc^2 ds.$$ Then, you can define a new coordinate $u = c^2(s - t)$ - i.e. the time dilation, itself, ramped up by $c^2$ and write: $$H dt - 𝐩·d𝐫 - μdu.$$

Now, you have a new invariant that you can take the non-relativistic limit for. In place of the Minkowski invariant $$ds^2 = dt^2 - \frac{|d𝐫|^2}{c^2}$$ are now two invariants: $$|d𝐫|^2 + 2 dt du + \frac{1}{c^2}du^2, \quad ds = dt + \frac{1}{c^2}du.$$ This, too, has a non-relativistic limit: $$|d𝐫|^2 + 2 dt du, \quad ds = dt.$$ The 11th generator in the extended Poincaré group is actually the translation on the $u$ coordinate. Since everything else is $u$-invariant, then ... by the Noether Theorem ... there is a conserved quantity. That's $μ$.

In the non-relativistic limit, $μ$ is the "central charge" that arises when you centrally extend the Galilei group to the Bargmann group - which is actually what the correct symmetry group for non-relativistic theory is, not Galilei!

Extend the operator correspondences appropriately: $$H ⇔ iħ \frac{∂}{∂t}, \quad 𝐩 ⇔ -iħ∇, \quad μ ⇔ -iħ\frac{∂}{∂u},$$ replace the $𝝰$ and $β$ matrices to rewrite $$δH - 𝝲·𝐩 - εμ ⇔ iħ\left(δ\frac{∂}{∂t} + 𝝲·∇ + ε\frac{∂}{∂u}\right),$$ for units $δ$, $𝝲 = \left(γ^1, γ^2, γ^3\right)$, $ε$, endowing them with a suitably-defined algebra (which will - in fact - be isomorphic to the complexified Dirac algebra), and add in the eigenvalue equation $$μψ = -iħ\frac{∂ψ}{∂u},$$ for the central charge. Restricting to the ordinary Poincaré group, for Relativity, by setting $U = 0$ and $μ = m$ produces: $$iħ\left(δ\frac{∂ψ}{∂t} + 𝝲·∇ψ + ε\frac{∂ψ}{∂u}\right) = 0, \quad mψ = -iħ\frac{∂ψ}{∂u}$$ which will be equivalent to the Dirac equation. Taking the limit $1/c^2 = 0$, on the other hand, will produce a result equivalent to the Schrödinger equation.

You can also add in a coupling with the electromagnetic potential. In that case, the non-relativistic version will be the Pauli-Schrödinger equation.

Note that shifting from $E$ to $H$ by subtracting off $mc^2$ corresponds adding a phase factor $\exp(±imc^2/ħ)$ to $ψ$, depending on which direction you're going in.

I demonstrate this, in detail - here Morphing The Dirac-Maxwell Equation To The Pauli-Schrödinger Equation and in the opposite direction, here Morphing The Schrödinger Equation To The Klein-Gordon Equation.

Needless to say, none of this was available in during the time Schrödinger devised his equation. He didn't know that Bargmann was the correct symmetry group for non-relativistic theory, rather than Galileo, nor that its natural geometric interpretation is 5D, not 4D; and even up to the present day there is still not a proper understanding that the correct symmetry group for Relativity is not the Poincaré group, but its (trivial) central extension to the extended Poincaré group.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.