Why quantum erasers cannot communicate backwards in time

From: Bruce Zweig (userzweig_at_hotmail.com)
Date: 06/29/04 

Date: Tue, 29 Jun 2004 21:33:49 +0000 (UTC)

A number of posters have wondered why delayed choice quantum erasers
cannot communicate backwards in time. I myself wondered about it
after reading popular reports about the phenomenon (especially Brian
Greene's "The Fabric of the Cosmos"). To review, you first split a
photon beam into idlers and signals using a spontaneous parametric
down-converter. Then you run the beams through a double slit,
recording which-path information on the the idler. Finally, you
observe diffraction on the signal beam depending on whether or not the
which-path information on the idler is erased. The erasure of the
idler's information can occur after the signal is detected, and this
creates the temptation to contemplate backwards time communication
possibilites.

Unfortunately, it turns out that the signal photons cannot diffract on
their own. Diffraction only occurs when a coincidence counter pairs
up the signal photons with their original idler companions during
detection. There's a paper by Shimizu et. al.
(http://arxiv.org/abs/quant-ph/0210142) in which the authors get
normal photons to diffract, and they get SDLC coincident pairs to
diffract, but they just can't get the darn signal photons by
themselves to diffract. A paper by Altschul & Altschul
(http://arxiv.org/abs/quant-ph/0106113) which describes the erasure
experiment makes the statement "This process produces double-slit
interference if and only if it is subjected to coincidence counting
with an A+B idler beam." (The two slits are labeled 'A' and 'B').

So, although the erasure can occur after the signal photon registers
on its detector, the diffraction pattern cannot emerge until after the
signal photon measurements have been combined with the idler photon
measurements in order to produce coincidence information, which can
only happen in a causal time frame.

If this were not the case, then things would get pretty complicated if
the idler photons were sent off on an infinitely long journey in which
it might never be determined whether erasure occurred or not.

If anyone knows of a way to get diffraction out of the signal photons
themselves, I'd love to hear about it. There's a thought experiment
by R. Srikanth (http://arxiv.org/abs/quant-ph/0101022) that seems to
be after this possibility, but I don't think anyone has tested it.