The uncertainty principle (derivation)

Heisenberg’s uncertainty principle states that the position and momentum of a particle cannot be determined simultaneously with unlimited precision.

The uncertainty not only applies to the position and momentum of a particle, but to any pair of complementary observables, e.g. energy and time. In general, the uncertainty principle is expressed as:

\Delta A\Delta B\geq \frac{1}{2}\left\vert\langle\varphi\vert[\hat{A},\hat{B}]\vert\varphi\rangle\right\vert\; \; \; \; \; \; \; \; \; 12

where \hat{A} and \hat{B} are Hermitian operators and A and B are their respective observables.

The derivation of eq12 involves the following:

  1. Deriving the Schwarz inequality
  2. Proving the inequality \left ( \Delta A \right )^{2}\left ( \Delta B \right )^{2}\geq -\frac{1}{4}\left ( \left \langle \varphi \left | \left [ \hat{A},\hat{B} \right ] \right |\varphi \right \rangle \right )^{2}
  3. Showing that \left \langle \varphi \left | \left [ \hat{A},\hat{B} \right ] \right |\varphi \right \rangle =-\left \langle \varphi \left | \left [ \hat{A},\hat{B} \right ] \right |\varphi \right \rangle^{*}

 

Step 1

Let

f(\lambda)=\langle\phi-\lambda\psi\vert\phi-\lambda\psi\rangle\; \; \; \; \; \; \; \; 13

where \phi and \psi are arbitrary square integrable wavefunctions and \lambda is an arbitrary scalar.

Since \langle\phi-\lambda\psi\vert\phi-\lambda\psi\rangle=\int (\phi-\lambda\psi)^{*}(\phi-\lambda\psi)d\tau=\int \vert\phi-\lambda\psi\vert^{2}d\tau\geq 0

f(\lambda)\geq 0\; \; \; \; \; \; \; \; \; 14

Expanding eq13, we have

f(\lambda)=\langle\phi\vert\phi\rangle-\lambda\langle\phi\vert\psi\rangle-\lambda^{*}\langle\psi\vert\phi\rangle+\lambda^{*}\lambda\langle\psi\vert\psi\rangle\; \; \; \; \; \; \; \; 15

Since \lambda is an arbitrary scalar, substituting \lambda=\frac{\langle\psi\vert\phi\rangle}{\langle\psi\vert\psi\rangle} and \lambda^{*}=\frac{\langle\phi\vert\psi\rangle}{\langle\psi\vert\psi\rangle} in eq15 gives:

f(\lambda)=\langle\phi\vert\phi\rangle-\frac{\langle\phi\vert\psi\rangle}{\langle\psi\vert\psi\rangle}\langle\psi\vert\phi\rangle

Substituting eq14 in the above equation and rearranging yields \langle\phi\vert\psi\rangle\langle\psi\vert\phi\rangle \leq\langle\phi\vert\phi\rangle\langle\psi\vert\psi\rangle. Since \langle\phi\vert\psi\rangle= \langle\psi\vert\phi\rangle^{*}

\vert\langle\psi\vert\phi\rangle\vert^{2}\leq\langle\phi\vert\phi\rangle\langle\psi\vert\psi\rangle\; \; \; \; \; \; \; \; 16

Eq16 is called the Schwarz Inequality.

 

Step 2

Let \psi=(\hat{A}-\langle A\rangle)\varphi and \phi=(\hat{B}-\langle B\rangle)\varphi, where \varphi is normalised, and \hat{A} and \hat{B} are Hermitian operators, which implies that \hat{A}-\langle A\rangle and \hat{B}-\langle B\rangle are also Hermitian operators (see this article for proof). The variance of the observable of \hat{A} is

(\Delta A)^{2}=\frac{\sum_{i=1}^{N}(\hat{A}-\langle A\rangle)^{2}}{N}=\langle(\hat{A}-\langle A\rangle)^{2} \rangle

=\langle \varphi\vert(\hat{A}-\langle A\rangle)[(\hat{A}-\langle A\rangle)\varphi]\rangle=\langle [(\hat{A}-\langle A\rangle)\varphi]\vert[(\hat{A}-\langle A\rangle)\varphi]\rangle=\langle\psi\vert\psi \rangle\; \; \; \; \; \; \; \; 17

Note that the 2nd last equality uses the property of Hermitian operators (see eq36). Similarly,

(\Delta B)^{2}=\langle\phi\vert\phi\rangle\; \; \; \; \; \; \; \; 18

Substituting eq17 and eq18 in eq16 results in

(\Delta A)^{2}(\Delta B)^{2} \geq\vert \langle\psi\vert\phi\rangle\vert^{2} \; \; \; \; \; \; \; \; 19

Let z= \langle\psi\vert\phi\rangle where z=x+iy. So, \left | z \right |^{2}=x^{2}+y^{2}\geq y^{2}. Since y=\frac{z-z^{*}}{2i}, we have \left | z \right |^{2}\geq \left ( \frac{z-z^{*}}{2i} \right )^{2}, which is

\left | \langle\psi\vert\phi\rangle\right |^{2}\geq \left ( \frac{\langle\psi\vert\phi\rangle-\langle\psi\vert\phi\rangle^{*}}{2i} \right )^{2}=-\frac{1}{4}\left (\langle\psi\vert\phi\rangle-\langle\phi\vert\psi\rangle\right )^{2}\; \; \; \;\; \; \; \; 20

Substituting eq19 in eq20 gives

(\Delta A)^{2}(\Delta B)^{2}\geq -\frac{1}{4}\left ( \langle\psi\vert\phi\rangle-\langle\phi\vert\psi\rangle\right )^{2}\; \; \; \; \; \; \; \; 21

Next, we have

\langle\psi\vert\phi\rangle= \langle (\hat{A}-\langle A\rangle)\varphi\vert (\hat{B}-\langle B\rangle )\varphi\rangle=\langle\varphi\vert(\hat{A}-\langle A\rangle)(\hat{B}-\langle B\rangle)\varphi\rangle

=\langle\varphi\vert\hat{A}\hat{B}\vert\varphi\rangle-\langle B\rangle\langle\varphi\vert\hat{A}\vert\varphi\rangle-\langle A\rangle\langle\varphi\vert\hat{B}\vert\varphi\rangle+\langle A\rangle\langle B \rangle\langle\varphi\vert\varphi\rangle

=\langle\varphi\vert\hat{A}\hat{B}\vert\varphi\rangle-\langle A\rangle\langle B \rangle \; \; \; \; \; \; \; \; 22

Similarly,

\langle\phi\vert\psi\rangle=\langle\varphi\vert\hat{B}\hat{A}\vert\varphi\rangle-\langle B\rangle\langle A \rangle \; \; \; \; \; \; \; \; 23

Substituting eq22 and eq23 in eq21 yields

(\Delta A)^{2}(\Delta B)^{2}\geq -\frac{1}{4}(\langle\varphi\vert\hat{A}\hat{B}\vert\varphi\rangle-\langle\varphi\vert\hat{B}\hat{A}\vert\varphi\rangle )^{2}

=-\frac{1}{4}(\langle\varphi\vert\hat{A}\hat{B}-\hat{B}\hat{A}\vert\varphi\rangle )^{2}=-\frac{1}{4}(\langle\varphi\vert[\hat{A},\hat{B}]\vert\varphi\rangle )^{2}\; \; \; \; \; \; \; \; 24

 

Step 3

\langle\varphi\vert[\hat{A},\hat{B}]\vert\varphi\rangle=\langle\varphi\vert\hat{A}\hat{B}\vert\varphi\rangle-\langle\varphi\vert\hat{B}\hat{A}\vert\varphi\rangle=\langle\hat{A}\varphi\vert\hat{B}\varphi\rangle-\langle\hat{B}\varphi\vert\hat{A}\varphi\rangle

=\langle\varphi\vert\hat{B}(\hat{A}\varphi)\rangle^{*}-\langle\varphi\vert\hat{A}(\hat{B}\varphi)\rangle^{*}=- \{\langle\varphi\vert\hat{A}(\hat{B}\varphi)\rangle^{*}-\langle\varphi\vert\hat{B}(\hat{A}\varphi)\rangle^{*}\}

=-\{\langle\varphi\vert\hat{A}(\hat{B}\varphi)\rangle-\langle\varphi\vert\hat{B}(\hat{A}\varphi)\rangle\}^{*}=-\langle\varphi\vert[\hat{A},\hat{B}]\vert\varphi\rangle^{*}\; \; \; \; \; \; \; \; 25

We have used eq37 for the 2nd equality and eq35 for the 3rd equality. Substituting eq25 in one of the \langle\varphi\vert[\hat{A},\hat{B}]\vert\varphi)\rangle in eq24 results in

(\Delta A)^{2}(\Delta B)^{2}\geq\frac{1}{4} \{\langle\varphi\vert[\hat{A},\hat{B}]\vert\varphi\rangle^{*}\}\langle\varphi\vert[\hat{A},\hat{B}]\vert\varphi\rangle

Therefore,

(\Delta A)^{2}(\Delta B)^{2}\geq \frac{1}{4}\vert\langle\varphi\vert[\hat{A},\hat{B}]\vert\varphi\rangle\vert^{2}

\Delta A\Delta B\geq \frac{1}{2}\vert\langle\varphi\vert[\hat{A},\hat{B}]\vert\varphi\rangle\vert\; \; \; \; \; \; \; \; 26

which is eq12, the general form of the uncertainty principle.

For, the observable pair of position x and momentum p, we have

\Delta x\Delta p\geq \frac{1}{2}\vert\langle\varphi\vert[\hat{x},\hat{p}]\vert\varphi\rangle\vert

Since [\hat{x},\hat{p}]\varphi=x\frac{\hbar}{i}\frac{d}{dx}\varphi-\frac{\hbar}{i}\frac{d}{dx}(x\varphi)=i\hbar\varphi

\Delta x\Delta p\geq \frac{1}{2}\vert\langle\varphi\vert i\hbar\varphi\rangle\vert=\frac{\hbar}{2}\vert i\vert

\Delta x\Delta p\geq \frac{\hbar}{2}\; \; \; \; \; \; \; \; 27

 

 

Next article: spectral decomposition of an operator
Previous article: commuting operators
Content page of quantum mechanics
Content page of advanced chemistry
Main content page

Leave a Reply

Your email address will not be published. Required fields are marked *