Delayed quantum eraser with double slit

[Edgardo]

Hello all,

I saw this setup for a quantum eraser:
http://grad.physics.sunysb.edu/~amarch/PHY5656.gif
In this setup, we get no interference:
http://grad.physics.sunysb.edu/~amarch/PHY5658.gif

Then, by inserting a polarizer we can erase the which way information:
Setup with polarizer:
http://grad.physics.sunysb.edu/~amarch/PHY5657.gif
With this setup there is interference:
http://grad.physics.sunysb.edu/~amarch/PHY5653.gif

Let's forget about the coincidence counter.
We put our detector for photon p (idler photon) on the Mars with Bob, such that we have a delayed choice experiment. And Alice is the one who looks at the screen to observe the photon distribution.

Now my question: Alice writes down the photon distribution BEFORE Bob measures the idler photon. Since we have not erased the which-way information yet, we get NO interference pattern, so Alice writes down: "I have not observed interference".
Then, when the idler photons arrive Mars, Bob decides to erase which-way information of EVERY idler photon by putting in the polarizer. According to the "delayed quantum erasure" the photons should show an interference pattern, but we know that Alice already has written down: "I have no interference pattern"

How can this be?

-------------------------------------------
Here the whole description to that experiment
http://grad.physics.sunysb.edu/~amarch/

 

[DrChinese]

How can this be?

As I read the setup, the interference pattern is seen when the coincidence results are plotted. There is no "Alice pattern" apart from the "Bob" pattern. The pattern appears only when the results are compared.

 

[Edgardo]

Thx Dr. Chinese!
I think I now found the mistake in my gedankenexperiment. It's NOT possible to erase the which-way-information of EVERY single idler photon. There's only a certain probability that the idler photon's state will collapse to the polarizer's eigenstate.
I forgot about the quantum probabilities.