Galaxy Zoo project (maybe) finds a "lopsided universe".
Kent Paul Dolan
project (in which I'm one participant among at least
100,000 participants), modeled on the Seti at Home
project, but using human eyes as processors rather
than merely handing off stray computer power has
received some publicity in the Daily Telegraph:
That URL is probably volitile and will go away
eventually; newspaper online archives seem
impermanent.
It seems that the universe-as-we-see-it may not be
as isotropic as expected; spiral galaxies may have a
tiny bias in their preferred chirality. Work is
currently underway to check the classifications for
sources of bias to make sure this is a universe bias
and not a human classifier bias.
This anisotropy, if confirmed, leaves room for all
kinds of cosmology speculations.
See also
in case any readers here would like to join the
effort. Clicking for galaxy type is a fun way to
relax for an hour or two a day.
There is a payoff in doing this to astronomy, of
course, as the above URLed article attempts to
convey.
One more personal payoff is that every rare once in
a while, one discovers something astonishingly
beautiful. I've made five screen wallpapers from my
finds among the currently 11,400 or so galaxies
classified by me [not at all in contention for "a
large number", there are some few participants who
have submitted over 100,000 classifications].
Another personal payoff is the chance that some
individual discovery one makes, reported "out of
band" in email or on the Galaxy Zoo forum, may
especially advance astronomy.
One "train wreck" I found is leaving debris looking
very like a rare "irregular" galaxy, giving one
possible scenario for how such galaxies are formed.
I've aslo found half a dozen or more "ring galaxies"
helping add to the evidence that these galaxy types,
thought to be the result of face to face collisions
of similar sized spiral galaxies, if I understand
correctly, aren't nearly as rare as previously
thought.
xanthian.
Oh No
>The Galaxy Zoo internet cooperative astronomy research
>project (in which I'm one participant among at least
>100,000 participants), modeled on the Seti at Home
>project, but using human eyes as processors rather
>than merely handing off stray computer power has
>received some publicity in the Daily Telegraph:
>
>http://tinyurl.com/24qr2o
>
>That URL is probably volitile and will go away
>eventually; newspaper online archives seem
>impermanent.
>
>It seems that the universe-as-we-see-it may not be
>as isotropic as expected; spiral galaxies may have a
>tiny bias in their preferred chirality. Work is
>currently underway to check the classifications for
>sources of bias to make sure this is a universe bias
>and not a human classifier bias.
>
>This anisotropy, if confirmed, leaves room for all
>kinds of cosmology speculations.
>
by human eye is much smaller than the scale on which we make the
assumptions of homogeneity and isotropy in the cosmological principle.
We do not expect isotropy or homogeneity within the local supercluster,
for example. To get a handle on the degree of homogeneity up to
redshifts of 0.2, take a look at the results from the 2dFGRS, for
example.
Regards
--
Charles Francis
moderator sci.physics.foundations.
substitute charles for NotI to email
Kent Paul Dolan
> Note that the proportion of the universe which is
> available to analysis by human eye is much smaller
> than the scale on which we make the assumptions of
> homogeneity and isotropy in the cosmological
> principle.
Well, yes, but we're talking about a sample of
1,000,000 galaxies/whatever out of a potential
125,000,000,000 or so galaxies suspected to exist
out to the limits of our seeing, so the sample size
is way big enough _to be_ representative, although
distributed fairly locally to us just due to limits
of our technology.
It seems to me to be "special pleading" to say that
the sample our technology is capable of providing is
somehow not representative of the mostly less
accessible or inaccessible whole.
Granted, a "galaxy" of a few image pixels isn't
going to be classifiable, it only takes a hundred
pixels or so to distinguish some kinds from other
kinds.
> We do not expect isotropy or homogeneity within
> the local supercluster, for example.
"We" are a long way past that, IIUC.
> To get a handle on the degree of homogeneity up to
> redshifts of 0.2, take a look at the results from
> the 2dFGRS, for example.
> http://www.roe.ac.uk/~jap/2df
I'm not sure you understand just how far out this
task is looking. Here's an example right on the
ragged edge of what I'd consider worth classifying
as other than an "I don't know", and it has a z of
0.504, if I'm reading its accompanying labels
correctly.
http://cas.sdss.org/astro/en/tools/explore/obj.asp?id=588017977812255245
Sure, we'd _like_ that kind of galaxy image
resolvability to the limits of what the speed of
light has brought to our seeing, but that would
probably need a telescope with an aperture "the
size of a planet" or at least of a large moon
crater, and no one is willing to pay for such a
beast before we have a constant presence in space.
I don't think you can dismiss that big a sample as
"not pertinent to questions of cosmological
isotropy"; the professional astronomers doing the
real analysis work with the classifications as done
by us amateurs certainly think it _is_ applicable,
which is what has them so excited by their first cut
looks at the data appearing to support anisotropy.
[The Sloan Digital Sky Survey apparently had some
excellent seeing at its command.]
xanthian.
Oh No
>
> > Note that the proportion of the universe which is
> > available to analysis by human eye is much smaller
> > than the scale on which we make the assumptions of
> > homogeneity and isotropy in the cosmological
> > principle.
>
>Well, yes, but we're talking about a sample of
>1,000,000 galaxies/whatever out of a potential
>125,000,000,000 or so galaxies suspected to exist
>out to the limits of our seeing, so the sample size
>is way big enough _to be_ representative, although
>distributed fairly locally to us just due to limits
>of our technology.
>
>It seems to me to be "special pleading" to say that
>the sample our technology is capable of providing is
>somehow not representative of the mostly less
>accessible or inaccessible whole.
I'm not saying that. I am just trying to get a handle on scales here.
The size of the sample is less relevant than the distribution of the
sample. When you talk of human eye, I guess you mean people looking at
plates from the largest telescopes, not what they may see in smaller
telescopes, which would necessarily produce a bias toward near galaxies.
>
>Granted, a "galaxy" of a few image pixels isn't
>going to be classifiable, it only takes a hundred
>pixels or so to distinguish some kinds from other
>kinds.
>
> > We do not expect isotropy or homogeneity within
> > the local supercluster, for example.
>
>"We" are a long way past that, IIUC.
>
> > To get a handle on the degree of homogeneity up to
> > redshifts of 0.2, take a look at the results from
> > the 2dFGRS, for example.
>
> > http://www.roe.ac.uk/~jap/2df
>
>I'm not sure you understand just how far out this
>task is looking.
Possibly not.
>Here's an example right on the
>ragged edge of what I'd consider worth classifying
>as other than an "I don't know", and it has a z of
>0.504, if I'm reading its accompanying labels
>correctly.
>
>http://cas.sdss.org/astro/en/tools/explore/obj.asp?id=588017977812255245
ok. I drew attention to the 2dFGRS because it shows that, for a
comprehensive survey out to ~0.2, albeit for a smallish field, the
universe is still fairly lumpy. Consider that, in a closed model
universe, a full sky survey at z=1 would give a region less than 7% of
the whole to which the assumptions of homogeneity and isotropy apply.
z=0.5 would be less than 2%. In an open universe, of course, the whole
is infinite, so a simplistic measure is not possible. For the kind of
modelling we are doing here, in constructing Friedmann Cosmologies, I
wouldn't think irregularities within a region of under 2% of the whole
need necessarily be considered of great importance, unless they show a
high level of anisotropy.
>
>Sure, we'd _like_ that kind of galaxy image
>resolvability to the limits of what the speed of
>light has brought to our seeing, but that would
>probably need a telescope with an aperture "the
>size of a planet" or at least of a large moon
>crater, and no one is willing to pay for such a
>beast before we have a constant presence in space.
Indeed. I look forward, however, to the next generation of very large
telescopes, which will tell us more about this kind of thing.
>
>I don't think you can dismiss that big a sample as
>"not pertinent to questions of cosmological
>isotropy"; the professional astronomers doing the
>real analysis work with the classifications as done
>by us amateurs certainly think it _is_ applicable,
>which is what has them so excited by their first cut
>looks at the data appearing to support anisotropy.
As I say, it is not the size of the sample that concerns me, but its
distribution. The fact remains that we are, at best, only able to
observe a tiny proportion of the universe. I don't have a calculation to
hand, but I believe even the microwave background would represent less
than half of a closed model universe, and a zero proportion of an
infinite one! I believe in the cosmological principle, but at the same
time I think we have to accept that we make the assumptions of isotropy
and homogeneity because it is difficult to see that there is another
meaningful or reasonable way forward, not because we can ever have
empirical evidence of what takes place outside the light cone.
>.
Phillip Helbig---remove CLOTHES to reply
Dolan <xant...@well.com> writes:
> Oh No wrote:
>
> > Note that the proportion of the universe which is
> > available to analysis by human eye is much smaller
> > than the scale on which we make the assumptions of
> > homogeneity and isotropy in the cosmological
> > principle.
>
> Well, yes, but we're talking about a sample of
> 1,000,000 galaxies/whatever out of a potential
> 125,000,000,000 or so galaxies suspected to exist
> out to the limits of our seeing, so the sample size
> is way big enough _to be_ representative, although
> distributed fairly locally to us just due to limits
> of our technology.
>
> It seems to me to be "special pleading" to say that
> the sample our technology is capable of providing is
> somehow not representative of the mostly less
> accessible or inaccessible whole.
Every experiment needs a good control. What about classifying
computer-generated "galaxies" generated such that the sample is
symmetric? Most people are right-handed, and from the way we walk in
the supermarket to the way we ice skate there are norms (though they are
different in different countries). The Earth spins in a certain
direction, which defines the way a clock moves etc. Just like humans
tend to overestimate vertical distance compared to horizontal (an
obvious throwback to apes living in trees), perhaps there are in-built
prejudices here. Has this been looked into?
Kent Paul Dolan
Helbig---remove CLOTHES to reply) wrote:
> Kent Paul Dolan <xanth...@well.com> writes:
>> ...but we're talking about a sample of
>> 1,000,000 galaxies/whatever out of a potential
>> 125,000,000,000 or so galaxies suspected to exist
>> out to the limits of our seeing, so the sample size
>> is way big enough _to be_ representative, although
>> distributed fairly locally to us just due to limits
>> of our technology.
>> It seems to me to be "special pleading" to say
>> that the sample our technology is capable of
>> providing is somehow not representative of the
>> mostly less accessible or inaccessible whole.
> Every experiment needs a good control. What about
> classifying computer-generated "galaxies"
> generated such that the sample is symmetric?
Supposedly, [and having looked at so many real
galaxies in all their variety, I believe it]
computer generated galaxies aren't nearly "good
enough" yet to produce a representative sample as
varied as the real universe is. You do, after all,
have to know what a pretty good size set of galaxies
look like, the project currently underway, to know
what they _should_ look like, to model them, a boot
strap problem without an obvious solution.
> Most people are right-handed, and from the way we
> walk in the supermarket to the way we ice skate
> there are norms (though they are different in
> different countries). The Earth spins in a
> certain direction, which defines the way a clock
> moves etc. Just like humans tend to overestimate
> vertical distance compared to horizontal (an
> obvious throwback to apes living in trees),
> perhaps there are in-built prejudices here. Has
> this been looked into?
Yes, except that _has this been_ is more _this is
being_.
Whenever a result looks fair to overturn the
received wisdom of a science, here "cosmological
isotropy", the sane scientist or here group of
scientists goes into a frantic "check for errors"
mode, trying very hard to prove to themselves that
they are wrong. As soon as the early results started
looking slightly skewed chirally (IIUC, it's only a
single percent or so skewed), the organizers of
Galaxy Zoo started looking for just the kinds of
biases you describe.
The first "control" used was to check the amateur
classifiers against the professionals, with
pre-existing (small, couple of thousand objects)
professionally classified samples tossed into the
mix to see if amateurs agreed with professionals as
to the classification of those objects. It seems we
amateurs are pretty even with the professionals at
picking out the simple "what shape is this object"
kind of classifications we're doing in Galaxy Zoo
"phase one".
There will be a Galaxy Zoo "phase two", with a
richer set of classification options,
implemented sometime next year, and probably
"early next year" IIUC. The operators are still
trying to puzzle out which are the best
questions to have we amateurs attempt to answer
for them. I know I have been often gritting my
teeth for lack of in-band ways to add flavor to
my classifications.
Next, checks for handedness biases and for galaxy
color biases were added; we're now being shown a
several-fold bigger set of images, which include
"the originals", "the originals rotated, 90, 180,
270", "the originals reflected (and maybe also
rotated)", and "the originals reduced to monochrome
(and sometimes reflected and / or rotated)".
This is to check for "if it's blue, it's probably
spiral" bias as well as clock bias and handedness
bias. One thing that's really, really hard to avoid
doing is convincing oneself a galaxy _must_ be a
spiral one, and then forcing oneself to "see" a
chirality that exists only because of that
conviction.
This whole enterprise is a really, really fuzzy
classification problem, there are smooth blends from
the obvious easy to classify cases to the "what the
heck is that" nearly impossible to classify cases,
because every galaxy is a child of a different
formation, consolidation, and interaction history.
For example, there are galaxies that have the long
curved sweeping "arms" one looks for in deciding
"hey, this must be a spiral galaxy", but have them
curved both directions and embedded in a typical
elliptical galaxy smoothly elliptical cloud of
stars, so that they are neither clockwise nor
counterclockwise spirals, just big pressure waves
stirred into the ellipticals, I guess.
One fellow, talking about this, described the
density waves in such galaxies as resembling the
suds in a top loading agitator washing machine, and
warned others not to turn those into "spirals".
Since the average participant among 100,000 of us
has classified (as of one point in time) a sample of
about 300 objects, and the whole set of objects is
around 1,000,000 "galaxies or other", there's a
pretty rich availability of MxN re-classification
opportunities to be had, and that "check for biases"
started maybe two weeks ago now.
It's nice that there are so many participants, and
such a big dataset that none of us will ever
classify even half of it individually. We can safely
be treated, in checking for bias, as a mob, without
worrying about our individual idiosyncracies in
doing classification, and there are lots of those.
For example, I'm pretty sure I've been classifying
far too many "fuzzy stars" as "elliptical galaxies",
which I realized after finding a way to check my
work against objects already known to be "stars"
versus "galaxies". Oops.
The Galaxy Zoo forum is also a good place to see
people carefully explaining (arguing for, might be a
better description) their misconceptions about how
to do classification, which lets the site operators
see bias sources they might not have guessed to be
there themselves, and then to build in checks for
the admitted biases to see if they are common or
rare.
This is very much a classic "Delphi" voting system,
where the average vote of the mob will be much
better at classifying on average than the average
classification success of any single participant.
Delphi voting works, but the theory as to why it
does is pretty unsettled, so far as I know.
xanthian.
[By the way, note to the moderators, Google Groups
is currently incapable of posting to this particular
moderated newsgroup, at least from the browser I
use, SeaMonkey, and has been for quite a while. I've
sent their software support staff a report, but the
usual turnaround I see on bugs I report is in excess
of a year to find them fixed. I just use another
Usenet access provider when this happens. FYI]
Oh No
>Supposedly, [and having looked at so many real
>galaxies in all their variety, I believe it]
>computer generated galaxies aren't nearly "good
>enough" yet to produce a representative sample as
>varied as the real universe is.
Hardly a problem. One would just randomise images of real galaxies,
flipping half of them in the process.
>Whenever a result looks fair to overturn the
>received wisdom of a science, here "cosmological
>isotropy", the sane scientist or here group of
>scientists goes into a frantic "check for errors"
>mode, trying very hard to prove to themselves that
>they are wrong. As soon as the early results started
>looking slightly skewed chirally (IIUC, it's only a
>single percent or so skewed), the organizers of
>Galaxy Zoo started looking for just the kinds of
>biases you describe.
As I have explained, a 1% bias is hardly likely to overturn cosmological
isotropy.
Kent Paul Dolan
> As I have explained, a 1% bias is hardly likely to
> overturn cosmological isotropy.
Further review of the SDSS images shows that there
are galaxies capable of being classified by eye for
chirality out to at least z=20 (obviously those are
very large galaxies).
I can't see how a trained statistician can dismiss a
1% bias in a sample of one million as anything less
than paradigm-shattering, though, brought to bear
against an isotropy claim on whicn most of cosmology
depends for its very existence. Perhaps you'd like
to explain why the odds against that happening don't
matter?.
xanthian.
[Mod. note: pretty sure you don't mean z=20... -- mjh]
Oh No
>Oh No <N...@charlesfrancis.wanadoo.co.uk> wrote:
>
> > As I have explained, a 1% bias is hardly likely to
> > overturn cosmological isotropy.
>
>Further review of the SDSS images shows that there
>are galaxies capable of being classified by eye for
>chirality out to at least z=20 (obviously those are
>very large galaxies).
This can't be true. You are misunderstanding the titling. I noticed that
something other than redshift is also labelled z on the example you gave
earlier. The furthest galaxies are, I am fairly sure, still at about
z=6. The images are extremely unclear. There was at least one candidate
at z=10, but there are serious problems in correctly identifying
redshift from the available data, and I believe it was eventually
determined that this was not correct. Your previous figure of z=0.5 is
much closer to a realistic range for this kind of identification.
>
>I can't see how a trained statistician can dismiss a
>1% bias in a sample of one million as anything less
>than paradigm-shattering, though, brought to bear
>against an isotropy claim on whicn most of cosmology
>depends for its very existence. Perhaps you'd like
>to explain why the odds against that happening don't
>matter?.
>
I think I have already explained. The cosmological principle applies to
the universe as a whole, but we can only observe a tiny proportion of
the universe inside the light cone. The motion of the galaxies is
modelled as an ideal fluid, in which galaxies are treated as point-like
particles in the fluid. This is not a fundamental assumption of physics,
but simply a modelling assumption, and is only intended as an
approximation to reality. The chirality of galaxies has no particular
bearing on the motion of the fluid, taken as a whole. Nor do local
motions of galaxies and clusters, such as the motion of local clusters
toward the Great Attractor. One might regard such things as small scale
turbulence in the fluid. With respect to the cosmological principle, the
region we can see, inside the light cone, is small scale. Remember also
we are working in units in which the speed of light is 1, so the
velocity of turbulent motion is very low.
I already pointed you at the images from 2dFGRS. A glance at these
should tell you that the level of inhomogeneity we observe up to z=0.2
is much more than 1%, and would still be much more than 1% up to z=0.5.
Your exercise is interesting enough, but saying that it has a bearing on
the cosmological principle is like looking at a small section of a road
surface under a magnifying glass and declaring that the road is not
flat.
Richard Saam
> The motion of the galaxies is
> particles in the fluid. This is not a fundamental assumption of physics=
,
> but simply a modelling assumption, and is only intended as an
> approximation to reality. The chirality of galaxies has no particular
> bearing on the motion of the fluid, taken as a whole. Nor do local
> motions of galaxies and clusters, such as the motion of local clusters
> toward the Great Attractor. One might regard such things as small scale=
> turbulence in the fluid. With respect to the cosmological principle, th=
e
> region we can see, inside the light cone, is small scale. Remember also=
> we are working in units in which the speed of light is 1, so the
> velocity of turbulent motion is very low.
> =
This idea of 'particles in a fluid' cosmic or otherwise
has its roots at about the same time as Einstein's theories =7E1910s
with the work of:
Miron Smoluchowski, Drei Vortr=E4ge =FCber Diffusion, Brownsche Molekular=
bewegung =
und Koagulation von Kolloidteilchen (Three Lectures on Diffusion, Brownia=
n =
Motion, and Coagulation of Colloidal Particles), Phys. Z., 17, 557 (1916)=
; =
Versuch einer Mathematischen Theorie der Koaguationskinetik Kolloider L=F6=
sugen =
(Trial of a Mathematical Theory of the Coagulation Kinetics of Colloidal =
=
Solutions), Z. Physik. Chem., 92, 129, 155 (1917).
Albert Einstein's son 'Hans Albert' worked in Berkeley
as professor for hydraulics at the University of California
from 1947 to 1971 and knew this subject very well.
One interesting concept is the idea of fluid shear dv/dx
and in hyraulics is usually designated by the letter G
and is defined as sqrt(Dissipating Power/(viscosity*volume)).
It is used in the design of mixing applications
wherein the power required to keep particles
in fluid (with a particular viscosity) suspension
at a particular dv/dx is computed.
It has the units of /time, the same as Hubble constant 'H'
so for argument in cosmic dimension use 'H'.
Another concept is the idea of particle kinetic collision:
J =3D (1/6)*n1*n2*H*(d1+d2)=5E3
J =3D particle collisions/volume/time
n1 =3D number of particles 1 per volume
n2 =3D number of particles 2 per volume
d1 =3D diameter of particle 1
d2 =3D diameter of particle 2
A natural development of this logic is that
an equilibrium particle diameter d =7E 1/H.
The basic question is:
Can Hubble constant be looked upon as a cosmic fluid with shear dv/dx
which plays a fundamental role in establishing galactic size and distribu=
tion?
Given the visible universe size (limited by c)
and H =3D 2.31E-18 /sec (cm/sec/cm)
then the extremely low cosmic viscosity =3D 8.8E-32 poise (g/(cm sec))
all congruent with a cosmic density of =7E6E-30 g/cc.
Like in any mixing situation, local variation is expected.
Richard Saam
Richard Saam
> The motion of the galaxies is
> modelled as an ideal fluid, in which galaxies are treated as point-like
> particles in the fluid. This is not a fundamental assumption of physics,
> but simply a modelling assumption, and is only intended as an
> approximation to reality. The chirality of galaxies has no particular
> bearing on the motion of the fluid, taken as a whole. Nor do local
> motions of galaxies and clusters, such as the motion of local clusters
> toward the Great Attractor. One might regard such things as small scale
> turbulence in the fluid. With respect to the cosmological principle, the
> region we can see, inside the light cone, is small scale. Remember also
> we are working in units in which the speed of light is 1, so the
> velocity of turbulent motion is very low.
Francis and Dolan
This idea of 'particles in a fluid' cosmic or otherwise
has its roots at about the same time as Einstein's theories ~1910s
with the work of:
Miron Smoluchowski
Three Lectures on Diffusion, Brownian Motion
and Coagulation of Colloidal Particles
Phys. Z., 17, 557 (1916);
Trial of a Mathematical Theory
of the Coagulation Kinetics of Colloidal Solutions)
Z. Physik. Chem., 92, 129, 155 (1917).
Albert Einstein's son 'Hans Albert' worked in Berkeley
as professor for hydraulics at the University of California
from 1947 to 1971 and knew this subject very well.
One interesting concept is the idea of fluid shear dv/dx
and in hyraulics is usually designated by the letter G
and is defined as sqrt(Dissipating Power/(viscosity*volume)).
It is used in the design of mixing applications
wherein the power required to keep particles
in fluid (with a particular viscosity) suspension
at a particular dv/dx is computed.
It has the units of /time, the same as Hubble constant 'H'
so for argument in cosmic dimension use 'H'.
Another concept is the idea of particle kinetic collision:
J = (1/6)*n1*n2*H*(d1+d2)3
J = particle collisions/volume/time
n1 = number of particles 1 per volume
n2 = number of particles 2 per volume
d1 = diameter of particle 1
d2 = diameter of particle 2
A natural development of this logic is that
an equilibrium particle diameter d ~ 1/H.
The basic question is:
Can Hubble constant be looked upon
as a cosmic fluid with shear dv/dx
which plays a fundamental role
in establishing galactic size and distribution?
Given the visible universe size (limited by c)
and H = 2.31E-18 /sec (cm/sec/cm)
then the extremely low cosmic viscosity = 8.8E-32 poise (g/(cm sec))
all congruent with a cosmic density of ~6E-30 g/cc.
Phillip Helbig---remove CLOTHES to reply
<xant...@well.com> writes:
> Oh No <N...@charlesfrancis.wanadoo.co.uk> wrote:
>
> > As I have explained, a 1% bias is hardly likely to
> > overturn cosmological isotropy.
>
> Further review of the SDSS images shows that there
> are galaxies capable of being classified by eye for
> chirality out to at least z=20 (obviously those are
> very large galaxies).
There have never been galaxies observed at this high a redshift.
> I can't see how a trained statistician can dismiss a
> 1% bias in a sample of one million as anything less
> than paradigm-shattering, though, brought to bear
> against an isotropy claim on whicn most of cosmology
> depends for its very existence. Perhaps you'd like
> to explain why the odds against that happening don't
> matter?.
Extraordinary claims require extraordinary evidence. Remember Nordland
and Ralston? http://www.jb.man.ac.uk/~jpl/screwy.html
In other words, a 1% bias, if real, would be very interesting, but a 1%
bias might be explained by other things. Of course, the same would
apply to a 50% bias, but it is more difficult to imagine this being due
to subtle effects.
Jonathan Thornburg [remove -animal to reply]
> I can't see how a trained statistician can dismiss a
> 1% bias in a sample of one million as anything less
> than paradigm-shattering, though, brought to bear
> against an isotropy claim on whicn most of cosmology
> depends for its very existence.
*If* the effect is genuinely present in light arriving at the top
of the atmosphere, then it might telling be us something interesting
about galaxies. If it's present all over the sky, over a wide range
in galaxy redshifts, it might be a lot more interesting.
But before getting too excited, I'd like to see solid evidence that
the effect is (for example)
(a) present in independent classifications of the same objects,
(b) not present if galaxies are randomly and double-blind-ly flipped
before being classified,
(c) not an artifact of atmospheric refraction or seeing,
(d) not an artifact of non-symmetric telescope optics, and
(e) not an artifact of the anisotropic CCD clocking combined with
imperfect charge transfer between pixels in the CCD readout
process.
For some ideas on how to do data analysis to look for tiny distortions
in the shapes of huge numbers of galaxies, it's useful to consider
measurements of weak gravitational lensing. These go back to Tyson
et al 1990 (well, they go back earlier, but I've seen that cited as
the first generally-agreed-to-be *successful* measurement). The weak
lensing measurements had to (painfully) overcome plenty of similar
error sources before they became reliable. [See, for example, the
chapter by P. Schneider in "Gravitational Lensing: Strong, Weak and
Micro", edited by P. Schneider, C. Kochanek, and J. Wambsganss,
proceedings of the Saas-Fee Advanced Course 33, Spring-Verlag 2006,
hardcover ISBN-10 3-540-30309-X, ISBN-13 978-3-540-30309-1.]
ciao,
--
-- Jonathan Thornburg (remove -animal to reply) <J.Tho...@soton.ac-zebra.uk>
School of Mathematics, U of Southampton, England
"Trying to learn modern physics from popularizations is like trying to
learn to dance by watching the shadows flit by under the closed door
of a ballet school." -- John Baez, sci.physics.research, 22.Jan.2003
Phillip Helbig---remove CLOTHES to reply
<wil...@cfa.harvard.edu> writes:
> Resolution is not as bad a problem as that. Remember that distant
> galaxies are magnified; the angular size distance reaches a maximum at
> z=1.7 or so.
That's true, but the luminosity distance doesn't have such a maximum.
As a result, the surface brightness drops quite a bit. For observations
in a finite band (i.e. all real observations), that means that the
signal-to-noise ratio goes like (1+z)^5.
> The real problem is that visible light observations of
> high redshift galaxies detect rest-frame UV, in which local galaxies
> look quite different than in rest-frame B or V and would not be
> classifiable.
Indeed, another problem.
Kent Paul Dolan
> Kent Paul Dolan <xanthian @ well.com> writes:
>> Further review of the SDSS images shows that there
>> are galaxies capable of being classified by eye for
>> chirality out to at least z=20 (obviously those are
>> very large galaxies).
> There have never been galaxies observed at this high
> a redshift.
My error, you're quite correct. The site uses "z"
ambiguously, and in this case it seems to have meant
"optical magnitude in the far infrared", as can be
seen by hovering the cursor over the "z" entry in
the upper right hand widget here:
http://cas.sdss.org/astro/en/tools/chart/navi.asp?ra=188.58447&dec=8.19761&opt=
Sorry to be the bearer of misinformation.
> Extraordinary claims require extraordinary evidence.
Which is precisely why the operators have extended
the Galaxy Zoo project with months more classifying
time and a target of another 8,000,000
classifications (on a smaller set of candidates to
get lots more classifications per candidate (to
eliminate "sampling noise" as one possible source of
bias, I suspect)) in an attempt to eliminate
recognized possible sources of bias from the
classifications. These are real scientists running
the project, though we volunteers are mostly no such
creatures.
xanthian.
Kent Paul Dolan
> Phillip Helbig` wrote:
>> Extraordinary claims require extraordinary
>> evidence.
> Which is precisely why the operators have extended
> the Galaxy Zoo project with months more
> classifying time and a target of another 8,000,000
> classifications (on a smaller set of candidates to
> get lots more classifications per candidate (to
> eliminate "sampling noise" as one possible source
> of bias, I suspect)) in an attempt to eliminate
> recognized possible sources of bias from the
> classifications. These are real scientists running
> the project, though we volunteers are mostly no
> such creatures.
And, precisely why good scientists check their work,
_especially_ in the case of "extraordinary claims",
the universe seems not to be lopsided, just the
habits of the classifiers. This TouTube video
explains, more or less, that the bias check showed
that we classifiers are indeed biased, though not,
yet, why that is so.
http://chrislintott.net/2008/01/10/aas-look-im-on-youtube/
To the tune of some large sighs of relief, isotopy
still lives.
xanthian.