I'm self learning quantum computing through IBM's Qiskit's learning section (which I really like), and I've stumbled across an inequality that I don't quite understand fully.
This must be really easy, I just don't get this next inequality.

In their of the QFT circuit implementation, they write:
enter image description here

What interests me is the last equation, and specifically in the last equation the x circled in blue. This interest me because of their next proof of the QFT's circuit implementation:
enter image description here enter image description here Again, the last equation is what interests me.

Disregarding the order of qubits (Which I know is reversed in the equations), I do not understand how the first picture x's are equal to the second picture x's.
For example, in their example for a 3-qubit QFT, their output of the circuit is as follows: enter image description here Which is to my understanding not equal to their equation in the first picture (Focusing on the x's - in the first picture the x's are complete in each tensor product of the equation while in the last its only x's partial components in each tensor product).


1 Answer 1


I suppose there are two sources of confusion here -- one is the reversed order of the qubits between the circuit and the unitary, and the other is the use of $x$.

In terms of qubit ordering, I think your $|\psi_6\rangle$ is the output of the $\text{QFT}$ unitary rather than the output of the circuit, i.e. the first and last qubits are swapped.

In terms of the notation, let's consider the three-qubit example. Write $$x = (x_1 x_2 x_3) = x_1 2^{2} + x_{2} 2^1 + x_3 $$ in binary, where $|x_1 x_2 x_3\rangle$ is the input to the circuit. Then we have for example that $$\frac{x}{2^2} = x_1 + x_2 2^{-1} + x_3 2^{-2} $$ and so the middle term of $|\psi_6\rangle$ should be $\frac{1}{\sqrt{2}} \left[ |0\rangle + \exp\left( 2\pi i \frac{x}{2^2} \right) |1\rangle \right]$, where $$\exp\left( 2\pi i \frac{x}{2^2} \right) = \exp\left( 2\pi i (x_1 + x_2 2^{-1} + x_3 2^{-2}) \right)$$ $$= \underbrace{\exp\left( 2\pi i x_1 \right)}_{=1} \cdot \exp\left( 2\pi i ( x_2 2^{-1} + x_3 2^{-2}) \right) $$ $$ = \exp\left( \frac{2\pi i}{2^2} x_3 + \frac{2\pi i}{2} x_2 \right),$$ which is what we see in the expression.

In other words, we only see the partial components $x_i$ of $x$ in the last expression because the integer part of $x / 2^i$ gets cancelled out in the exponent. Hopefully you can convince yourself that the first and last terms follow similarly?


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.