"Henri Wilson" HW@.... wrote in message
...
On Sat, 21 Apr 2007 10:33:43 +0100, "George Dishman"
wrote:
"Henri Wilson" HW@.... wrote in message
. ..
On Thu, 19 Apr 2007 19:05:42 +0100, "George Dishman"
wrote:

Your figure for the linear change corresponds to a velocity
of over 43000 km/s instead of the 300 km/s measured.

Only if you apply the VDoppler equation.

That's right Henry. Try to get your head round this, please.
When astronomers take a series of spectra from a star in a
binary system they calculate the ratio of the frequency or
wavelength shift to the mean value for any spectral lines
and multiply by the speed of light to get the value that is
published as a velocity curve.

Yes I know that George.

Then you also appreciate that we need that curve on your
program in order to do any comparison. I can take the
300 km/s value from the graph of EF Dra below and convert
that into "Linear = 1.002" manually but it would be easier
if you just displayed it.

Have you any idea how much extra work that would involve?

A one-line change to convert the value you currently
display as the "linear" number from unitrs of magnitude
to units of km/s. It could read:

Mag. Change =0.32019 (Log) = 87772 km/s (peak to peak)

Obviously a linear (red) plot as well as the green
log plot would be nice but just putting the value
on the screen would be a start.

Presenting hte output so that hte curves can be magnified has given me
quite a
few headaches already, particularly for the log curves.

...but just for you I will have a go...

You could just add a text line giving the value but
displaying the curve with a horizontal axis and a scale
would avoid the doubt you had some time ago about a factor
of two difference if the figure is orbital velocity or
peak-to-peak variation.

True.
The magnitude change figure is actually the ratio maximum/minimum
brightness....so it's 'peak to peak'.

OK, so above I have doubled the 43886 km/s I posted before.
Your value should be around that ballpark. Anyway you then
simply adjust your orbital parameters to match that number
and the curve shape to the published velocity curve and you
have your match. Whether you reduce the speed, alter the
pitch or reduce the equalisation distance is up to you but
my feel is that you will find using the same equalisation
distance as for the pulsars gets you a solution and logically
we should expect that to be the case.

No, we have found it is dominant for pulsars and we don't
know for stars but we will in a moment.

The only evidence I have that it is dominant for pulsars is that it is
the
only
way I can match the curve of PSR1913+16 accurately.

The first evidence we found was J1909-3744 where the Shapiro
delay gives the orbital phase but both are consistent and I
suspect J0737-3039 will give much the same though you haven't
tried it yet.

However, the way such
curves are statistically produced, leaves a great deal of room for
error.
Maybe
a constant light speed is even assumed in their making.

More excuses Henry? I love your approach - "I can't match
the curve but there is room for error in the measurement
therefore I can claim I match it."

I CAN match the curve very accurately George.

I haven't seeen you match the amplitude of any velocity
curve yet, your example of EF Dra is too high by a factor
of nearly 150.

...but I can also produce a very similar curve with ADoppler completely
dominant.

TDoppler Henry, your ballistic theory equations produce
TDoppler and there is no room in it for alternatives.

If you are familiar with the way these curves are produced ...'epoch
matching'
or something like that. You gave me a reference to the technique

It was someone else I think but basically all it means is
that any one point is the average of all the readings at
the same phase over many cycles.

...then you
will be aware of the potential for inaccuracy.

Sure, things like sensor drift can cause problems. The
scatter of the readings gives the error bars but there
should always be some indication of systematics.

It is quite possible that this
curve was actually fiddled a bit to match the anticipated VDoppler one.

Accusing people of fraud again Henry? Sorry but if your
theory can't handle reality, don't go blaming the guys
who took the readings.

That's OK, I have consistently been saying you have no
choice but to include both VDoppler and ADoppler, they
aren't really different things, just different terms in
the single TDoppler equation.

It is possible that VDoppler is needed to match pulsar curves but it
doesn't
apear to be required in the case of star curves.

It doesn't matter what is needed, ballistic theory includes
both factors, period.

It appear that 'photon compressibility' is VDoppler determined but
the curve shape matches ADoppler.

No, what appears is that it seems the equalisation distance is
small for Cepheids too and most of the luminosity variation
is intrinsic.

Right, I think I have a possible picture now. We have two possibilities.
What
you just could be correct...and any variations from the pure Keplerian
curve
are also intrinsic.

Variations from a Keplerian orbit could only be due to other
bodies and you are severely limited as to what configurations
could be stable. Post-Newtonian factors will likewise be small
other than systematic drift of parameters over many years like
PSR1913+16. Over a few orbits these are negligible.

Wait, I stated that wrongly. I am assuming that these are huff puff stars.
..in
which case the radial velocities of expansion and contraction follow a
similar
curve to that of a star on Keplerian orbit.

No, you still don't seem to undertsand how they work. I didn't
get a chance to respond to an earlier post of yours about this.
the mechanism was described in the presentation I cited some
days back. Basically as the star contracts, the pressure and
density rise and importantly it reaches a point where it becomes
opaque. The radiation is trapped, pressure builds rapidly and the
outer layers start to expand. They become more transparent again
but momentum carries the material out until stopped by gravity
and it starts to fall back. The process is more like someone on
a trampoline.

What I meant was that the little peak at the top of the RT Aur curve could
be
intrinsic and VDoppler could produce the curve.
I can produce it very accurately with just ADoppler....BUT the velocity
range
is then seemingly too large.

Ballistic theory only allows you to use TDoppler, you
can't choose to use only some parts of an eqaution.

There cannot be a difference Henry, basic physics and even
pure maths rules it out (Fourier analysis). Ballistic
theory says that any electromagnetic disturbance moves at
"c+v" (that's shorthand as usual) and that applies to any
waveform shape whether a simple sine wave or a complex
signal. If you think of a burst of a sine wave of some
duration, it carries some finite energy and contains a
number of cycles. There is therefore an amount of energy
in each cycle. The brightening occurs because more cycles
arrive in a given time than were transmitted in the same
time due to bunching, but that increased number of cycles
in a given time is also obviously the frequency change
factor.

Whatever you do to modify the speed of propagation, the
green and red curves we have talked about are rigidly
linked. As you have said yourself in justifying not
drawing the extra curve, the two are identical with just
a different scale.

George I am just realising how amusing you are.

You keep telling me that if one piece of evidence refutes a theory then
the
theory is wrong. You often refer to sagnac as an example.

Yes, ballistic theory is undoubtedly wrong, this is purely
a hypothetical discussion to see what the consequences would
be if it were correct.

Yet you are trying to use the classical wave theory of light to prove my
ballistic theory wrong when it has been firmly established that the wave
theory
itself is wrong. .. IT BREAKS DOWN AT THE P.E. EFFECT.

Classical wave theory was around for a long time before it
was quantised. Ballistic is still in the wave theory region,
it has not yet been quantised. Before you can attempt that,
you need to know what the wave interpretation says and then
when you try to quantise it one of the constraints you must
meet is getting your quantum version equations to give the
same result as the wave version when averaged over large
numbers of photons.

Sure it can be used to
model interference patterns and such but it cannot explain anything about
the
particle nature of light.

Of course not, it doesn't attempt to yet.

Wave theory might work well in relation to the behaviour of groups of
photons...but it doesn't tell us anything about the partical nature of
individual photons or the way they travel through space. Nor does LET, SR
or
any other current theory ..

The current full theory is QED which amalgamates SR and QM in
flat space.

.. except the ballistic one.

Nonsense, you don't have any quantum equations equivalent to
ballistic theory at all. Perhaps we should revise ballistic
theory as you seem to be losing the plot a bit. Ritz said light
was emitted at c relative to the source. De Sitter suggested
that was untenable because binaries would show multiple images.
That is true for spectroscopic binaries so to try to get round
that a new theory was produced which included "speed equalisation"
The current theory has three parts:

1) The universe is Galilean invariant

2) Any electromagnetic disturbance initially propagates
at c relative to the source.

3) The speed relative to the medium through which the
disturbance propagates varies as it progresses as
described by the differential equation:

dv/ds = (c/n - v)/R

where n is the refractive index
R is a characteristic distance

Both n and R depend on the medium. n also depends
on the frequency of the wave however it seems R
does not since the orbital parameters of a binary
would then vary with the optical band in which it
was measured.

My current view is that wave theory should work for pulsars because the
pulses
are made of groups of photons. Both the pulse widths and the spacing
between
pulses should behave as wave theory says. Similarly, my brightness curves
are
based on the relative movements and aggregations of vast numbers of
photons and
should be an accurate simulation.

In the case of individual photons however, there is absolutely no reason
to
believe the same principle applies.
It cannot be assumed that the actual 'wavelengths' of photons making up
the
light in each 'macro pulse' experience the same compression or
rarification
that the pulses do.

It is not an assumption, it is a mathematical identity.
Think about the Fourier transform of a pulsed waveform.
Also remember the fact that the peak intensity from a
grating cannot be different if you analyse it by two
different approaches. The experiment is often done in
undergraduate work, get a line from a grating with a
bright source and then dim it until a photomultiplier
shows individual photons arriving. You can watch a
histogram build up to show exactly the same probability
as the wave theory intensity.

To match the brightness/velocity phase relationship of cepheids, I need
ADoppler to dominate.

Tough luck, the universe isn't going to change to suit
your needs. Ballistic theory says the result is due to
TDoppler, end of story.

Therefore I propose that photons are indeed ADoppler
sensitive but to a much smaller degree than are the pulses themselves.

You don't get to"propose" anything Henry, you write down
the equations for ballistic theory, which I have done
above, and then you calculate what they predict.

To quantise the theory, you write down a new set of
equations that apply to photons and then you calculate
what those predict.

For instance if the pulses compress by a factor of 1.5, the photons may do
so
by only 0.0015 or less.

That's not what the equations say.

If you think of photons as discrete entities then this is a quite
conceivable
theory ...not just a wild guess.
In summary, individual photon 'compression' DOES occur when the source is
accelerating ... but the effect is much smaller than it is in the case of
the
relative movements of all the photons in a group.

This way, the brightness/velocity phasing of star curves is correct and
the
doppler shifts are of the right order.

So, even though a brightness curve might match perfectly and vary by 1.5
or so
(linear), the actual spectral shift is diluted to only a very small
fraction of
that value.
I'm wondering if there isn't a time dependent term in the
'compressibility
equation'. Does a second order effect come into play? (da/dt)

da/dt is the derivative of the acceleration. For a
circular orbit, v = sin(wt) where w is the angular
frequency or 2 pi / P where P is the orbital period.

a = dv/dt = w cos(wt)

da/dt - -w^2 sin(wt)

It would come in as one more order than acceleration so
would be roughly a factor of 1/P less than the basics
we have discussed. For a one day orbit it would be four
orders of magnitude smaller or make a change of about
0.01% to your curves, not worth worrying about.

'a' goes from -1 through zero to +1.

The difference ratio can be infinite.

You need to revise your calculus Henry.

...The time interval between the emission of the 'ends of a photon' is
much
smaller than that between the pulses of a pulsar. Therefore, for a
source
in
orbit, the velocity difference is much larger across the pulse that
across
the
photon.

So the pulse gap should compress relatively much more that will the
photon.

No, the speed difference is delta_v = a * t

so the ratio of speed difference to time is linear.

Only if a is constant..

No, a can vary in any way you like in which case
delta_v is the integral of a.dt

Anyway let's not delve too deeply into this because I suspect there is a
far
more important factor involved. INDIVIDUAL PHOTONS simply do not compress
as
much as the groups do.

When (if ever) you quantise ballistic theory, you will start
by taking the wave theory result and showing that the quantum
version produces identical results for large aggregates.

What I mean is that you can rely on the measurement of the
shift. Photometry needs careful attention to comtaminating
light, getting accurate calibration, CCD pixel sensitivity
and so on but spectra have far fewer sources of systematic
errors, especially when only the relative shift is needed
and not an absolute wavelength measurement.

...well ther is still argument as to whether gratings are sensitive to
'wavelength' (absolute distance between wavecrests) or 'frequency' (rate
of
wavecrest arrival)

Gratings produce a peak of intensity at the point on the
screen where the reflected waves from each ruling arrive
in phase. You should be able to answer the question simply
by calculating that condition for an incoming plane wave.

Yes, the classical theory is well established....but, without an aether,
it is
still not clear why gratings detect doppler shifts from moving sources.

As I said, you need to analyse the behaviour of a grating in
ballistic theory, but since you include speed equalisation
it becomes quite simple, as the speed varies the waves are
compressed (for v initially c/n) or stretched (for v
initially c/n) and that changes the wavelength.

SR
certainly doesn't provide a 'physical' explanation.

Of course it does, it even explains the second order
effect which Ritz and aether theory get wrong. Don't
waste your time trying to change the subject again
Henry.

The only requirements are that it should be a Keplerian
orbit and the _TDoppler_ should match the published curve.

Yes, I understand what you are saying.

But in the simulation we DO know the correct blue curve...because that's
what
we start with.

The blue curve represents the actual motion of the body
which is an unknown under the Ritzian theory, in terms
of the model it is not "known", it is the "independent
variable" (technically a small number of variables,
semi major axis and eccentricity for example).

.....but in the simulation it is completely defined.

No, in the sim it is the "independent variable". That
like saying the operating frequency of a radio is not
known but the position of the tuning knob is "completely
defined". It is defined only in that there is a specific
angle you turn it to for each station.

We need a model that produces a brightness curve that is correct
in both shape and mag. change but which also accommodates the
compressible
photon concept to a small extent.

No, the theory requires that the waves change in a certain
way. If you then go on from there to try to quantise the
theory then it starts with the fact that a grating will
produce a peak of intensity at some location and the photons
are then partly defined by the fact that a photomultiplier
will produce individual flashes with a distribution curve
that is identical to the intensity curve.

That's a bit of an oversimplification George.

Not really Henry. If you try to quantise the theory you will
probably start by adopting Planck's solution to the black
body problem, then you will define the properties of a photon
such that energy is conserved and wavelengths and frequencies
match the wave theory you have at the moment and so on.

I maintain that groups of photons behave differently from individual ones.

The behaviour of groups is defined by the statistical formula
that will form the basis of you method of quantisation.

Have a think about what I said above.
Should individual photons compress as much as the gaps between groups of
photons or pulses?

The relationship between acceleration and velocity is
linear so the effect must be identical. We could talk
about this forever Henry, there is a huge amount of
evidence. For example we see the sodium doublet in the
lab and also in stellar spectra. That doublet is the
same as a modulated sine wave (what is called suppressed
carrier) and any discrepancy would show up as harmonics
or a change in the ratio of the separation to the centre
frequency.


that's ancient stuff.... side bands... I assume that's what you are
saying...

Exactly. We don't see any.

The doublet is just due to a fine quantum level separation. I has nothing
to do
with what I'm taking about.

The fact that we don't see extra sidebands tells you how
waves behave and quanta are simple the smallest amount of
energy for any given wave.

Conservation of energy means wave cycles must
arrive at a higher rate if the brightness is to increase,
and cycle arrival rate is the definition of frequency.

You're talking about 'E=h.nu' type brightness variation.

No! E=h.nu will not work in ballistic theory as a general
principle. It will berquired to solve the black body
problem but can only apply at the point of emission.

My guess is that ballistic theory will require a kinetic
approach such as if a photon is emitted with speed c+v and
arrives at speed c then the energy will be h*nu*(c/(c+v))^2
where nu is the frequency of emission in the source frame.
However I think you have a _lot_ of work to do to make that
work consistently.

My brightness variation is caused by numbers of photons arriving per unit
time.

Yes, that's what I meant too.

There is no room for any argument Henry, the red curve
must be directly related to the green and your claimed
match the luminosity curve of EF Dra gives a gross
mismatch to the velocity curve.

Not any more.

Yes Henry, you haven't changed the ballistic theory equations
so that's what they predict.

The solution is simple, it doesn't falsify your theory
and it works for pulsars and stars. Add the red curve
and match it to the velocity curve. You can do that
three ways. First you could reduce the size of the blue
curve. That ties up with what you have said many times,
that astronomers overestimate the velocity of orbits
because they don't take account of ADoppler. Second
you could simply increase the pitch. A nearly face-on
orbit can reduce the radial velocity by a factor of 140
without too much trouble, though I think you'll find
that all binary systems have to be face-on which raises
difficult questions. The third method which I have been
suggesting for some time is that the speed equalisation
is a property of space and is similar regardless of the
source of the light travelling through it. We had a six
hour figure for the pulsar and I think it needs to be
much less than that. there is no reason to think that
the effect for other types of star would be any
different and I anticipate that putting that number into
the model for EF Dra will solve the problem. You haven't
given me any reason at all why that modelling approach
is not valid and it fully complies with ballistic theory
while your suggestions are all at odds with it.

I am claiming that, ...

I am not interested in handwaving "claims", the equations
do not predict what you are saying, your application of
them is wrong.

For the BaTh, ADoppler has to dominate otherwise photon waveshift will be
all
VDoppler and the phasing between the brightness and velocity will be 90
out.

For the current ballistic theory, the equations say that
the Doppler effect and the luminosity variation are both
the TDoppler factor, end of story. If you quantise the
theory, and I doubt you are capable of doing that, then
there will probably be an additional variation of (c/(c+v))^2
to take into account, but that is not part of the current
set of equations. At present, your simulation for EF Dra has
the velocity grossly too high but you can correct it by
reducing your equalisation distance to a value comparable to
that for the pulsars. The conclusion is that most of the
Cepheid variation is intrinsic and for EF Dra it is mostly
due to eclipsing.

George