Uncertainty relation demonstration

[andresordonez]

HI, I will not used the template provided since this is not a textbook problem. It's a problem I have with a demonstration in Quantum Mechanics Vol 1, Cohen-Tannoudji

"
Complement C III

[Q,P] = i \hbar

Consider the ket:

|\phi \rangle = (Q +i\lambda P)|\psi \rangle

where \lambda is an arbitrary real parameter. For all \lambda, the square norm \langle \phi | \phi \rangle is positive. This is written:

\langle \phi | \phi \rangle = \langle \psi | (Q - i \lambda P)(Q + i \lambda P) | \psi \rangle

=\langle \psi | Q^2 | \psi \rangle + \langle \psi | (i \lambda Q P - i \lambda P Q) | \psi \rangle + \langle \psi | \lambda ^2 P^2 | \psi \rangle

=\langle Q^2 \rangle + i \lambda \langle [Q,P] \rangle + \lambda ^2 \langle P^2 \rangle

=\langle Q^2 \rangle - \lambda \hbar + \lambda ^2 \langle P^2 \rangle \geqslant 0

The discriminant of this expression, of second order in \lambda, is therefore negative or zero:

\hbar ^2 - 4 \langle P^2 \rangle \langle Q^2 \rangle \leqslant 0

and we have:

\langle P^2 \rangle \langle Q^2 \rangle \geqslant \frac{\hbar ^2}{4}
"

I don't understand why the discriminant must be negative or zero for \lambda to be real. As far as I know, if you have:

a x^2 + b x + c = 0

with a, b, c reals, then:

x = \frac{-b \pm sqrt{b^2 - 4 a c}}{2 a}

and x is real if:

b^2 - 4 a c \geqslant 0

In this case:

a = \langle P^2 \rangle

b = -\hbar

c = \langle Q^2 \rangle

so the discriminant should be positive or zero instead of negative or zero right?

 

[vela]

If f(λ)=aλ²+bλ+c ≥ 0 for all λ, then its graph either lies completely above the λ axis or is tangent to the axis, so that it has, respectively, no (real) roots or just one.

 

[andresordonez]

Right, thanks!