Which is Correct? (delta p) (delta x) >= hbar or hbar/2?

[greatscott]

hbar or hbar/2??

Regarding the uncertainty principle, some books say

(delta p) (delta x) >= hbar

and others say

(delta p) (delta x) >= hbar/2.

Which is right? This matters because I get different results when I
let p x=hbar(or hbar/2), plug into the expression for energy, and
minimize it to get the ground state energy of the system.

 

[zefram_c]

Sigh... unless exact definitions are given for those deltas, the statement of the Uncertainty Principle should be taken as an order-of-magnitude estimate only. I believe that when one employs standard deviations, which is the correct way to go about it, the answer is hbar/2.

 

[marlon]

Sigh... unless exact definitions are given for those deltas, the statement of the Uncertainty Principle should be taken as an order-of-magnitude estimate only. I believe that when one employs standard deviations, which is the correct way to go about it, the answer is hbar/2.

zefram is correct

 

[meteor]

Cohen-Tannoudji's "Quantum mechanics" writes the HUP as \Deltax\Deltap(>~)hbar
Hameka's "Quantum mechanics" as \Deltax\Deltap>hbar

 

[zefram_c]

There is a simple way to settle the issue. The inequality becomes equality for the SHO in its ground state. So if anyone wants to calculate the the standard deviations \sigma_x=\sqrt{ <x^2>-(<x>)^2} and similar for p, then multiply them out, we have the answer. Actually it's easy:
<x>=<p>=0 \; so \; \sigma_x \sigma_p = \sqrt { <x^2> <p^2> }

By the virial theorem:
\frac{<p^2>}{2m} = \frac{1}{2} k<x^2> = \frac{E_0}{2} = \frac{\hbar \omega(0+1/2)}{2}

<p^2>=\frac{m \hbar \omega}{2}, <x^2>=\frac{\hbar \omega}{2k}

\sigma_x \sigma_p = \frac{\hbar}{2} \sqrt {\frac{m \omega ^2}{k}} \; but \; \omega^2 = k/m

Hence \sigma_x \sigma_p = \hbar /2 is the final answer. Note to self: learn Latex. It took me far longer to compose this than to actually solve the problem.

 

[meteor]

...is the final answer

That is how the HUP appears in Griffith's "Introduction to Quantum mechanics"

 

[humanino]

It took me far longer to compose this than to actually solve the problem.

Maybe not appearing in your derivation is the fact that the equality holds when the wavefunction has the same "shape" in postion and momentum variables. That is, when the Fourier transform of the wavefunction is analogous to the initial wavefunction : in the gaussian case, which applies to Harm. Oscill. you used.

Maybe I'll give a try to latex too...

 

[humanino]

OK, the all story is simple : the Heisenberg undeterminacy principle simply follows from the Schwarz inequality. Let us see how. Consider a state \psi and two observables \hat{A} and \hat{B}.
Now the standard deviation is given by :
<br /> ( \Delta a )^2 = \langle \psi|(\hat{A} - \langle a\rangle )^2 |\psi\rangle = \langle (a - \langle a\rangle )^2 \rangle <br />
This seems natural. Why bother an overall factor at this stage ?
Let \hat{A&#039;} = \hat{A} - \langle a\rangle
Then ( {\Delta}a )^2 = \langle\psi|\hat{A&#039;}^2|\psi\rangle
Likewise for \hat{B} one gets
( {\Delta}b )^2 = \langle\psi|\hat{B&#039;}^2|\psi\rangle

Now the real argument : Schwarz inequality. I redemonstrate.
Consider the norm of the vector
(\hat{A&#039;} + i\lambda \hat{B&#039;} )|\psi\rangle
This vector has positive norm :
<br /> \langle\psi|(\hat{A&#039;} - i\lambda \hat{B&#039;} )(\hat{A&#039;} + i\lambda \hat{B&#039;} )|\psi\rangle\geq 0 <br />
From this follows simply :
<br /> (\Delta a)^2 + \lambda^2 (\Delta b)^2 + i \lambda \langle \psi |[\hat{A&#039;},\hat{B&#039;}]|\psi\rangle\geq 0 <br />

As you can see, a 2nd order polynomial in \lambda which is always positive will lead to :
<br /> (\Delta a)(\Delta b) \geq \frac{1}{2}\langle \psi |[\hat{A},\hat{B}]|\psi\rangle <br />
and I did not bother about the primes, since the commutators are equal :
<br /> [\hat{A},\hat{B}]=[\hat{A&#039;},\hat{B&#039;}]<br />
This is the general way of deriving the \frac{1}{2} factor.
____________________________________________________________
Let me add the HO argument's origin : let us see how gaussian functions appear. The inequality becomes an equality iff the second order polynomial vanishes, that is when
\lambda = \lambda_0 = \frac{\hbar}{2(\Delta b)^2}=\frac{2(\Delta a)^2}{\hbar}
in which case the vector has vanishing norm, so :
[\hat{A}-\langle \hat{A}\rangle+i\lambda_0(\hat{B}-\langle \hat{B}\rangle)]|\psi\rangle = 0
Therefore, the condition for the inequality to become an equality is that the vectors [\hat{A}-\langle \hat{A}\rangle]|\psi\rangle = 0 and [\hat{B}-\langle \hat{B}\rangle]|\psi\rangle = 0 be proportional to each other (linearly dependent).
Let us take \hat{A}=\hat{x} (position) and

\hat{B}=\frac{\hbar}{i}\widehat{\frac{d}{dx}}
We collect the equation :
<br /> \left[ x + \hbar\lambda_0\frac{d}{dx} -\langle \hat{A}\rangle - i \lambda_0 \langle \hat{B} \rangle \right] \psi(x)<br />
with \langle\hat{x}|\psi\rangle.
We furthermore eliminate mean values :
<br /> \psi(x) = e^{i\langle\hat{B}\rangle x/\hbar}\phi(x- {\langle \hat{A}\rangle} )<br />
in order to get :
\left[ x + \lambda_0\hbar\frac{d}{dx}\right]\phi(x)=0
whose solution is :
\phi(x) = C e^{-x^2/2\lambda_0\hbar}
C is an arbitrary compex constant.
Finally :
\psi(x) = \left[2\pi(\Delta x)^2\right]^{-\frac{1}{4}}e^{i\langle p\rangle x/\hbar}e{-\left[ \frac{x-\langle x\rangle}{2\Delta x} \right]^2}
We note that the same lines can be carried out in the momentum representation, where one gets :
\bar\psi(p) = \left[2\pi(\Delta p)^2\right]^{-\frac{1}{4}}e^{i\langle x\rangle p/\hbar}e{-\left[ \frac{p-\langle p\rangle}{2\Delta p} \right]^2}
credit : Jean-Louis Basdevant "Mecanique quantique, cours de l'X"