X-Message-Number: 19378
Date: Sun, 30 Jun 2002 10:21:32 EDT
Subject: Intensity interferometer 3

Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit

Intensity interferometer has been invented by Robert Hanbury Brown (who died 
January 16 2002). When presented with the theory of the instrument, many 
physicists said it can't work, that strong view continued unabated even as 
the Australian Narrabri interferometer worked for 20 years. Today, that 
viewpoint seens to have been restated as: It can't produce picture because it 
throw aways the phase information. That is plain nonsens as what allows an 
interferometer to build a picture is the spatial sampling of information, not 
the phase in a single detector. I know that if I continue on that subject, 
particularly if I build some hard wired systems, I'll be "informed" by 
knowing people that it can't work, can't produce pictures,... and so on. So I 
choose to put things in place at start: I know the story, I know too that 
someone gave a "science" conference in November 1959 about the "proof" that 
artificial satellites are impossible, two months after Sputink 1. So thank 
you, I know the Earth is flat, at least for some.:-)

I have said before in part 2 that more we have beams in a 2 waves 
interferometer, the best the picture quality at the end. Optical systems with 
lenses and mirrors are limit cases of interferometers with nearly infinite 
entry beams. In the same way, insect composed eye are thousand beams 
intensity interferometers.

Intensity interferometers have a strange property: they are wave lenght 
insensitive. To see that in a simple way, assume for example that we look at 
one micrometer radiation (near infra-red) and we sample it at 3 Ghz or 10 cm 
wavelength = 100 000 micrometers; If we used a 2 wave instrument, we would 
directly get information at incoming wavelength scale, that is one 
micrometer. If we sample information at a scale x time larger, we need x 
squared experiments to reduce the uncertainty to the initial value. There x = 
100 000 and so we need 10 billions experiments. If we had started with 10 
micrometer radiation (far infra-red) only 10 000 x 10 000 = 100 milions 
experiments would have bring us down to wavelength precision, but we would 
need 100 times more experiments to go down to the micrometer scale. And there 
we hit what make intensity interferometer bad looking: It can go beyond the 
wavelength: it can work as a near field instrument at long distance.

So, that instrument sample a signal at a scale, use a radiation at a second 
scale and look at a third one. What define the number of experiments is the 
difference between sample and observed scales.  With long wavelengths, 
experiments are done on short bunch tails, with short wavelengths, longuer 
tails are sampled and the energy efficiency drops.

The Narrabri astronomical intensity interferometer had a sampling frequency 
near 100 Mhz or 3 m (9') and used starlight near half a micrometer. The big 
difference between these two values constrained it to look at long tails 
giving a poor energy efficiency. That limited it to look at only bright 
stars, and observing at an intermediate scale between these extremes. Going 
from 3 m to 30 micrometers asked for 10 billions experiments and 3 months of 
observing time.

Some years ago, I had undertook a survey of high speed photometers, I could 
not found systems far better that the one used by Hanbury Brown fifty years 
ago. May be someone could tell me what is the state of the art today?

Using balistic gate transistors, I assume 100 Ghz is a possibility. Moving 
from 100 Mhz to 100 Ghz would divide observing time on an  intensity 
interferometer by one million. For low power light sources such stars, the 
benefit is even larger: Oberving shorter tails would enhence energy 
efficiency and so allows observation of dimer sources. 

Going back to brain reader, a 100 Ghz sampling frequency gives a sample scale 
of 3 millimeters, if the observed scale is 100 000 times smaller, it is 30 
nanometers, what is requested to see molecular sized objects. The observing 
time would be 3 months, something workable for an experiment, not for 
business constrained device.

Yvan Bozzonetti.


 Content-Type: text/html; charset="US-ASCII"


Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=19378