I wrote:

analog to digital converter, I have designed with many of them.
Look up the Weak Law of Large Numbers. You are "cooking up" precision

do not have.

weak law of large numbers applies here so show everyone
big boy a real example that clearly demonstrates that averaging doesn't
eliminate more noise, that averaging won't work.

I am just a computer programmer.

psss. I have already done it. The higher the samples results in a more
stable answer. Ask yourself why big boy. Oh, could it be that the
noise is canceling itself???

I wrote:

In math, how many digits accuracy do you have if you have 1289

divided

by 10.00?

Is that 1289.1 or 1289.2 or 1289.3 or

From the kindness of their own heart, lol. Kindness is the last word I

In sci.physics,

Wow, did I actually find the second sober person in sci.physics.pub?

No, you found a hard-assed empiricist with some theory,
a good knowledge of how to find stuff using Google, and
maybe some tolerance, at least online. :-) (In person,
that's a different story.)

There is guy who created a device that measures the earths magnetic
field. It's not a big discovery or anything, but it shows what even a
*little* amount of cleverness can create.

I have no idea what you're referring to beyond an ordinary
compass though I could see a rather simple device consisting
of three or six coils connected to appropriate hardware, which
would carefully calculate where it is and how it is oriented
based on the voltages appearing at the terminals from the
coils. (It's probably simpler to just use a gyroscope or
a standard compass, though.)

So how many Teslas does it
require? Try 1/20000 of a Tesla. Sure, there are the magnetizing
coils that first align the magnetic moments, but that does not need to
be at any fixed strength. It is the magnetic field of earth that
changes the frequency from the hydrogen in the bottle of water that he
uses. If the magnetizing coils generate less current that only means
the EM that comes from the hydrogen will less. So he has a magnetizing
coil on the outside. Inside is a bottle of water that is wrapped
inside a receiving coil.

Erm...the water bottle is inside a sealed shipping container.
Or, perhaps, *is* the sealed shipping container. Will we require
that all shipping containers contain detector coils? An
interesting notion.

He magnetizes the outside coil and then
releases it. This aligns some of the protons and when this field is
released then some of the protons realign back to the earths magnetic
field or some align in random positions. Anyhow, this realignment
generates the EM that varies according the earths magnetic field, which
is ~1/20000 of a Tesla.

You are extremely confused as to how NMI works. First off, it's
*neutrons*. Second off, my guess is that the atoms do rotate,
and then rotate back, emitting photons in the process. However,
I'd have to look, and it gets complicated.

[2] Signal processing may involve a time/quality tradeoff. What
parameters of time, false positives, false *negatives*, and
power consumption are you envisioning here?

Now you want me to give out the secret?

You've patented it, right? :-)

How do you think the U.S. government would think if China or
better yet Iran learns of it. I know that's not what you want
to hear ... who wouldn't. If you people can provide help in
the area of NMR or if you have a better way of detecting the
element type then I'll take care of the software end.
Here is what I can presently do with my software breakthrough and my
limited knowledge of physics ->

Ah, of course. Well, in that case, what you need to do is get
some seed money, construct a working prototype that can process
a few containers a second -- then sell your services to the
US government after gaining a security clearance, going through
all requisite testing (presumably the US government knows how to
make a dirty bomb, fake dirty bomb, C4 explosive briefcase, etc.
etc. for "debugging" purposes when testing a security system).

~~~~~~~
There are two designs. Design #1 can aquire a 3D image of the area
from just one 180 degree rotation of the devices antennas.

OK, how fast does it rotate? I said 50 boxes per minute for a reason.
How many units will be online?

In design #1 we can get various false artifacts
imbedded in the 3D image. Basically this is caused by ground that
contains large changes in ground type. For example, if the antenna is
right over a split in ground type that goes from say granite to sand.

Construction artifact. The thing's sitting somewhere in port, remember?

The longer the signal is
the better the software can hear the incoming EM reflection.

I already mentioned the speed/quality tradeoff. Duh!

It seems
most people in this thread cannot believe that any device could receive
a reflected signal that reversed through 10's of thousands of feet
through solid earth at say 100MHz much less 10GHz.

I have no knowledge at this time regarding the conductivity frequency
spectrum of solid rock. (Apart from the rather obvious note that
I can't see through it, which means that it won't transmit EM
radiation in the 300 - 700 nm region of the spectrum, translating
into 1,000 THz to 429 THz frequency range.)

At 10GHz most will
tell you that signal does not penetrate the earth very deep at all--
few inches. That is an incorrect way of thinking. Perhaps at 10GHz
99% of the signal is absorbed and reflected in the first foot, but at
10 feet there is still a fraction of a % of the signal and at 10000
feet there is still a fraction of the signal.

Remember, the software
can pick up on the faintest signal amongst a sea of noise.

Depends on the signal strength, the length of time used to
process the signal, the characteristics of the signal, and
the sensitivity of the ADC.

The
incoming signal may be 1 millionth of the ADC's single bit-- it doesn't
matter.

If it's 1 millionth of the ADC's bit, then it will take a
million acquisition cycles to pick up that bit of data.
And that's if you're lucky.

Then again, I'm not that up on infotech. However, if one
models a signal as

Math.random() * 999999e-6 + Math.sin(t * omega) * 1e-6

and then tries to detect the signal using, say, a
threshold of 0.5, one will get 1 part signal and
999,999 parts noise each time.

As long as there is noise in the ADC input then the software
can pick up on as small of signal as it wishes. So in that sense,
noise helps. :-)

You're misconstruing the problem, then. Of course, noise might
very well help in certain areas of pattern recognition if one
has a million 1-bit ADCs (basically a crude CCD) and some
sort of focusing mechanism. However, it's far from clear to
me that it'll help the *computer*.

On the other hand the computer does have limitations
such as memory. For example, obviously it would be very time consuming
for the software to utilize a signal code that's say 1E+14 bits.

You'd be surprised.

30 frames/second * 10 megapixels * 24 bits/pixel = 7.2 gigabits/s --
and that's with somewhat cheap equipment.

That's a lot of memory.

100 Terabits. Might fit in a large shoebox. :-) I'm seeing 3 1/2"
quarter-terabyte (250 GB) drive units. Or one can use RAMsticks;
each RAMstick can hold 4 GB so you'd need about 25,000 of them.
The main problem is mounting and cooling them properly, and
feeding the processors.

In such a case the software would have to use
some type of memory such as a huge real of memory tape rather than fast
RAM.

Tape is sequential access; that'll slow your system *way* down.

The more time the software has to sample the signal then the
better it can hear the signal.

And the more things back up behind the container under test.
I have my doubts your system here will handle almost a container
a second, unless you have a *lot* of test units.

Also the stronger the emitted signal
obviously equates to a stronger received signal.
The EM frequencies used depends on the situation. Obviously lower
frequencies traverse better through solid matter, but it requires more
time to generate a 3D of the same resolution. Believe it or not, the
wavelength of the radio is not a limitation of resolution. For
example, such a device could generate a 3D picture down to 1mm in
resolution with a radio wavelength of 1000mm.
~~~~~~~

Maybe, if one has enough time.

(There's a *lot* of issues here, and getting more than a few microounces
of the stuff would be extremely difficult anyway.
I'd expect Al Qaeda would have some major problems
doing it right, but not much problem doing it wrong,
although ideally they'd blow themselves up in the
attempt. [One can hope! ;-) ])

Al Qaeda has the same philosophy as China as far as patience. I think
they learned long ago that a rushed job doesn't pay off. They have
TIME on there side. Thankfully you scientists also have time on your
side. PLEASE, get to work and start thinking. I am also working on
this, but I have not taken the years of my life to study science, just
computer programming and some basic science. I am pretty darn good at
classical EM though.

I *am* thinking. I'm thinking your idea needs more work. :-)
However, a functional prototype able to handle a shoebox,
resulting in a picture within 1 second would be nice.

You do have one, right?