Question about noise in entanglement experiments

[Isa]

Hi everyone

While learning about quantum mechanics, I became curious about the real-life experimental data. Wikipedia says that entanglement experiments require coincidence counters, because the majority of the signal received by detectors is noise. It further says, that coincidence counters reduce this noise down to a workable level, but do not fully eliminate it. https://en.wikipedia.org/wiki/Coincidence_counting_(physics)

Could someone with experience in the field educate me about the rough size of the remaining noise?

a) Is this fully negligible, or does it account for a few percent of the real data, or is it more than that?

b) And if the remaining noise is significant: what techniques are used to eliminate it?

c) Is this “normalization” done pre-or post-data acquisition?
Many thanks!

 

[jfizzix]

The remaining noise is what is called "accidental coincidences", which happens when the detector is triggered by a photon that is not due to the entangled pair (e.g., room light or another laser stimulated optical process), or due to thermal fluctuations in the detector itself (dark counts)

For a good source of photon pairs, a coincidence count rate might be of the order 10^6 per second, while the accidentals will be of the order 10^4. This may be negligible if the experiment you're doing allows you to subtract background noise.

The normalization would be done post-acquisition, since photon detectors and coincidence counters do exactly that, regardless of the source.

 

[f95toli]

Note also that this is only relevant for optical entanglement experiment. There are many, many other systems where entanglement can be studied/used.
That said. all systems will of course have their own sources of noise; there is no such thing as an experiment without noise.

 

[zonde]

Wikipedia says that entanglement experiments require coincidence counters, because the majority of the signal received by detectors is noise.

This wikipedia article is quite misleading but anyways it does not say that purpose of coincidence counters is reduction of noise. Coincidence counters are required for measurement of correlation between two detections - the main result of experiment.
This sentence from article is not correct in the light of modern experiments and quite misleading concerning older experiments:
"In any experiment studying entanglement, the entangled particles are vastly outnumbered by non-entangled particles which are also detected; patternless noise that drowns out the entangled signal."
There are two photon entanglement experiments where unpaired detections where not subtracted for inequality calculations.
http://arxiv.org/abs/1511.03189
http://arxiv.org/abs/1511.03190

In experiments there are four sources of unpaired photon detections:
- stray photons
- so called detector "dark counts" (counts that detectors produce when input is blocked);
- unpaired entangled photon detections, when partner photon is lost before reaching detector;
- unpaired entangled photon detections, when partner is not detected in detector due to limited detection efficiency.
And only the first two I would call noise. Other two are rather imperfect detection of entangled photons.

b) And if the remaining noise is significant: what techniques are used to eliminate it?

Stray photons are reduced using narrow-band filters before detectors and maintaining the room dark.
"Dark counts" are reduced by high quality detectors or even transition edge sensors (superconductor based detectors).
Losses in transit to detector are reduced by using high quality equipment, careful setup, short distance to detector (if possible) and possibly a lot of experience.
Detector efficiency can be increased by using high efficiency detectors (obviously). Say transition edge sensors can have very high efficiency (up to 99%) but they have bigger "dead time" (larger time before they can detect next photon).

I hope I gave fairly accurate list by not being experimentalist myself.