Redshift Quantization?

by Suede
Tags: quantization, redshift
 PF Gold P: 1,129 Hi Suede - welcome to Physics Forums! That's an interesting list of references (some of which weren't working just now but seem to be improving). It appears that they are mostly concerned with the quantization of intrinsic redshift (for galaxies within a cluster as well as quasars) rather than its existence, and I think that whole idea is on somewhat dodgy ground as it seems that the statistical strength of the arguments is getting weaker as the amount of data is getting larger. If quasars aren't black holes, then it would certainly be possible that some the evolution of new quasars could proceed in a stepwise way (for example regularly blowing off layers on reaching certain critical energy density levels). However, I'd prefer to consider the evidence for or against decreasing intrinsic redshift without getting into the issue of quantization. Do you have a similar list of references which claim to prove that there is no intrinsic redshift, or which criticize the articles you've listed? I'd like to see a more balanced view.
P: 71
 Quote by Jonathan Scott Hi Suede - welcome to Physics Forums! That's an interesting list of references (some of which weren't working just now but seem to be improving). It appears that they are mostly concerned with the quantization of intrinsic redshift (for galaxies within a cluster as well as quasars) rather than its existence, and I think that whole idea is on somewhat dodgy ground as it seems that the statistical strength of the arguments is getting weaker as the amount of data is getting larger. If quasars aren't black holes, then it would certainly be possible that some the evolution of new quasars could proceed in a stepwise way (for example regularly blowing off layers on reaching certain critical energy density levels). However, I'd prefer to consider the evidence for or against decreasing intrinsic redshift without getting into the issue of quantization. Do you have a similar list of references which claim to prove that there is no intrinsic redshift, or which criticize the articles you've listed? I'd like to see a more balanced view.
Yeah I just fixed a bunch of those links.

They got hosed up when I collected them.

As for the counter arguements, you can look up intrinsic redshift on wiki which is dominated by people opposed to the idea. You're not going to find many published papers refuting it though, just a lot of ad hom attacks and pontification.

I haven't seen any published papers refuting the findings to date.

Emeritus
PF Gold
P: 4,005

Redshift Quantization?

Welcome to PF, Suede.

It looks quite an impressive list, doesn't it?

And as many of the papers on your list have been around for a long time, I'm sure you won't be at all surprised to learn that Tifft et al.'s claims (3, 4, 5, 6, 7, 8, 9)* have been given a pretty thorough working over. And some curious things emerge from these papers - and the couple of dozen or so that you don't cite:

a) despite the apparent similarity in findings, between papers, a closer read of them shows that most are, in fact, inconsistent - would you like to go through a sample in detail?

b) there is a paper which points out that the statistical methods used, in most of the early papers, is wrong, rendering the stated conclusions invalid (I'll see if I can dig it up, if anyone is interested)

c) the later the paper, generally, the weaker or more limited the 'redshift periodicity' reported. For example, Weak redshift discretisation in the Local Group of Galaxies? (2005):
 We discuss the distribution of radial velocities of galaxies belonging to the Local Group. Two independent samples of galaxies as well as several methods of reduction from the heliocentric to the galactocentric radial velocities are explored. We applied the power spectrum analysis using the Hann function as a weighting method, together with the jackknife error estimation. We performed a detailed analysis of this approach. The distribution of galaxy redshifts seems to be non-random. An excess of galaxies with radial velocities of $\sim 24 {km} \cdot {s}^{-1}$ and $\sim 36 {km} \cdot {s}^{-1}$ is detected, but the effect is statistically weak. Only one peak for radial velocities of $\sim 24 {km} \cdot {s}^{-1}$ seems to be confirmed at the confidence level of 95%.
2) uses SDSS DR3, and the SDSS team cited Bell's paper in their own, announcing DR5; here's what they had to say:
 Repeating the analysis of Richards et al. (2006) for the DR5 sample reveals no structure in the redshift distribution after selection effects have been included (see lower histogram in Figure 3); this is in contrast to the reported redshift structure found in the SDSS quasar survey by Bell & McDiarmid (2006).
IOW, a more careful analysis, using a larger set of data (a superset of DR3), found no signal.

1) is in its second version, and AFAICS is not yet published, despite going up in arXiv over a year ago. Maybe we should wait until it appears in a relevant peer-reviewed journal before commenting?

Oh, and 10), the book by Arp? Well, anyone can write a book, can't they? No peer-review required wrt any claims made, is there?

* not all these have Tifft as an author (Guthrie and Napier are an independent pair, for example), but they all address the ~24/36/72 km/s apparent redshift periodicity
Emeritus
PF Gold
P: 4,005
 Quote by Suede Yeah I just fixed a bunch of those links. They got hosed up when I collected them. As for the counter arguements, you can look up intrinsic redshift on wiki which is dominated by people opposed to the idea. You're not going to find many published papers refuting it though, just a lot of ad hom attacks and pontification. I haven't seen any published papers refuting the findings to date.
Our posts crossed Suede, and as only the first two on your list have anything to do with quasars, how about we ask a mentor to split out the quantisation/periodicity/discretisation of galaxy redshifts into a separate thread?

As I indicated in my earlier response, there isn't much need to "refut[e] the findings to date" ... largely because they are mutually inconsistent!
 P: 71 Oh here's one: A 2005 paper done by two chinese atstronomers. http://adsabs.harvard.edu//abs/2002MNRAS.336L..13H We have used the publicly available data from the 2dF Galaxy Redshift Survey and the 2dF QSO Redshift Survey to test the hypothesis that there is a periodicity in the redshift distribution of quasi-stellar objects (QSOs) found projected close to foreground galaxies. These data provide by far the largest and most homogeneous sample for such a study, yielding 1647 QSO-galaxy pairs. There is no evidence for a periodicity at the predicted frequency in log(1 +z), or at any other frequency. To which I quote Scott: Recently mainstream astronomers have joyfully announced that they can find no quantization effects in the observed redshift values of quasars. Of course not! The raw measured total redshift values of the universal set of all known quasars are not quantized. It is the inherent redshift z values that are!
Emeritus
PF Gold
P: 4,005
 Quote by Suede Oh here's one: A 2005 paper done by two chinese atstronomers. http://adsabs.harvard.edu//abs/2002MNRAS.336L..13H We have used the publicly available data from the 2dF Galaxy Redshift Survey and the 2dF QSO Redshift Survey to test the hypothesis that there is a periodicity in the redshift distribution of quasi-stellar objects (QSOs) found projected close to foreground galaxies. These data provide by far the largest and most homogeneous sample for such a study, yielding 1647 QSO-galaxy pairs. There is no evidence for a periodicity at the predicted frequency in log(1 +z), or at any other frequency. To which I quote Scott: Recently mainstream astronomers have joyfully announced that they can find no quantization effects in the observed redshift values of quasars. Of course not! The raw measured total redshift values of the universal set of all known quasars are not quantized. It is the inherent redshift z values that are!
Who is "Scott"?

Where is this quote from (source please)?
 P: 71 Lets not be so choosy when selecting our quotes. Looking at the full quote that refutes the Bell findings: http://arxiv.org/PS_cache/arxiv/pdf/...704.0806v1.pdf Repeating the analysis of Richards et al. (2006) for the DR5 sample reveals no structure in the redshift distribution after selection effects have been included (see lower histogram in Figure 3); this is in contrast to the reported redshift structure found in the SDSS quasar survey by Bell & McDiarmid (2006). To construct the lower histogram we have partially removed the effect of host galaxy contamination (by excluding extended objects), limited the sample to a uniform magnitude limit of i < 19.1 (accounting for emission-line effects), and have corrected for the known incompleteness near z ∼ 2.7 and z ∼ 3.5 due to quasar colors lying close to or in the stellar locus. Accounting for selection effects significantly reduces the number of objects as compared with the raw, more heterogeneous catalog, but the smaller, more homogeneous sample is what should be used for statistical analyses. What? So they limited the magnitude of the selection, "corrected" for incompleteness (how they "corrected" I'd love to know), and then make the claim that a small sample size is what should be used for statistical analyses. Where did these guys learn stats? Larger sample sizes always reflect a higher degree of accuracy in statistical analysis.
Emeritus
PF Gold
P: 4,005
 Quote by Suede Lets not be so choosy when selecting our quotes. Looking at the full quote that refutes the Bell findings: http://arxiv.org/PS_cache/arxiv/pdf/...704.0806v1.pdf Repeating the analysis of Richards et al. (2006) for the DR5 sample reveals no structure in the redshift distribution after selection effects have been included (see lower histogram in Figure 3); this is in contrast to the reported redshift structure found in the SDSS quasar survey by Bell & McDiarmid (2006). To construct the lower histogram we have partially removed the effect of host galaxy contamination (by excluding extended objects), limited the sample to a uniform magnitude limit of i < 19.1 (accounting for emission-line effects), and have corrected for the known incompleteness near z ∼ 2.7 and z ∼ 3.5 due to quasar colors lying close to or in the stellar locus. Accounting for selection effects significantly reduces the number of objects as compared with the raw, more heterogeneous catalog, but the smaller, more homogeneous sample is what should be used for statistical analyses. What? So they limited the magnitude of the selection, "corrected" for incompleteness (how they "corrected" I'd love to know), and then make the claim that a small sample size is what should be used for statistical analyses.
Yes, that's right.

Selection effects are the bane of astronomers' lives, and recognising (first), characterising (second), and correcting for them (last) takes up an enormous amount of their professional life.

I'm not sure how familiar you are with astronomy, Suede, so please don't take the following example as being condescending ....

If you want to know how stars are distributed by inherent brightness ('intrinsic luminosity' or 'absolute luminosity') - how many are within each narrow range of brightness, within a sufficiently large volume - how would you go about it?

Well, for starters you'd measure the observed brightness of all the stars in the sky, wouldn't you.

But many stars emit most of their 'light' in parts of the EM spectrum that either doesn't get through the atmosphere, or which is partially blocked by it ... so you have to find a way to convert your observed brightness numbers to 'above the atmosphere' numbers (and that's very difficult to do; in fact until instruments could be put on rockets, or satellites, close to impossible).

But many of the stars in the sky are clearly dimmed by stuff between us and the stars, especially dust, so you have to find a way to convert your estimated 'above the atmosphere' brightness numbers to 'removing the effect of dust absorption'.

But the stars in the sky are clearly not all at the same distance from us, so you have to find a way to account for this, by estimating each one's distance.

When you've done all this - and to get this far took astronomers several centuries (!) - you find something quite interesting: there are lots and lots of really (intrinsically) faint stars near to us, and very few intrinsically bright ones! So much so that even today we're quite unsure of the details of the lowest part of the 'stellar luminosity function', because we can't be sure we've even seen all the really faint stars within ~100 pc (say), let alone been able to estimate their absolute luminosity.

One of the terrific things about SDSS is the huge amount of effort that went in, at the design stage, to addressing systematic effects (a superset of selection effects). Such care lead to the paper you quoted from - the survey data allows for correction of systematic effects relatively easily.

Now, would you really love to know how ("how they "corrected" I'd love to know")? I'd be happy to walk you through the details (they're all out in the open, in papers published in relevant peer-reviewed journals), but it will require a considerable investment of your time (unless you've already got an advanced degree in astrophysics).
Emeritus
PF Gold
P: 4,005
 Quote by Suede [...] Where did these guys learn stats? Larger sample sizes always reflect a higher degree of accuracy in statistical analysis.

May I ask what your professional training is, in physics, astronomy, or statistics?

I ask because this part of your post, that I'm quoting, seems to reveal gross ignorance, at least about astronomy and statistics.
 P: 71 I submit that "significantly" reducing the sample size as they have done will of course alter the findings in their favor. When looking for redshift periodicity, it makes no sense to limit the selection down to a handful of QSOs when an entire raw catalogue of valid data is present. Selection effects in statistics are inherent biases in the collection methods. There are no biases in raw redshift data, it is what it is. The larger the sample, the more accurate your results will be when looking for periodicity. You'd have to convince me otherwise in order for me to accept that papers findings, and considering every stats textbook on the planet says otherwise, I'll go with the paper using the larger sample.
P: 71
 Quote by Nereid You added this after I quoted your post ... May I ask what your professional training is, in physics, astronomy, or statistics? I ask because this part of your post, that I'm quoting, seems to reveal gross ignorance, at least about astronomy and statistics.
Gross ignorance is eliminating valid data to make findings fit preconcieved notions of redshift distribution.
 Sci Advisor PF Gold P: 9,185 Large numbers of contaminated data is not more statistically meaningful than a smaller, more carely selected and properly corrected data set. Contaminated data will always yield statistical anomalies. In that sense I suggest the larger data sets are little more than 'gee whiz' statistics. If these findings were truly valid and significant, wouldn't you think every grad student in the universe would be publishing supporting papers? Everyone wants a piece of any new discovery if one is to had - especially aspiring doctoral students looking to jump start their careers. Oddly enough, few such papers are to be found, despite the many years since quantized redshift was first 'discovered'. The silence is deafening. So I ask, how many of the papers you cite been further refined - as in several papers built one upon the other - that systematically affirm the original 'conclusions'?
P: 71
 Quote by Chronos Large numbers of contaminated data is not more statistically meaningful than a smaller, more carely selected and properly corrected data set. Contaminated data will always yield statistical anomalies. In that sense I suggest the larger data sets are little more than 'gee whiz' statistics. If these findings were truly valid and significant, wouldn't you think every grad student in the universe would be publishing supporting papers? Everyone wants a piece of any new discovery if one is to had - especially aspiring doctoral students looking to jump start their careers. Oddly enough, few such papers are to be found, despite the many years since quantized redshift was first 'discovered'. The silence is deafening. So I ask, how many of the papers you cite been further refined - as in several papers built one upon the other - that systematically affirm the original 'conclusions'?

I agree that "Large numbers of contaminated data is not more statistically meaningful than a smaller" however, if you took the time to look at the SDSS datasets, you'd see they have built in confidence definitions for each redshift listed.

In fact, if you took the time to read some of the papers I posted, you'd see things like:

http://arxiv.org/PS_cache/arxiv/pdf/...712.3833v2.pdf
I obtained 80,398 quasar data from the SDSS BestDR6 database (Adelman-McCarthy et al 2007) available at cas.sdss.org/astro/en/. The data (downloaded on the 7th December 2007) were selected by specClass = 3 for QSO and 4 for HIZ_QSO, low redshift and high redshift quasars, respectively. These include any objects with spectra that have been classified by the spectroscopic pipeline as quasars (specClass = QSO or HIZ_QSO). The DR6 data used here were not filtered as was the DR5 quasar catalog, described in Schneider et al. (2007), found at http://www.sdss.org/dr6/products/val...socat_dr5.html. In that case the DR5 catalog quasars were chosen from those that have apparent i-band PSF magnitudes fainter than 15, absolute i-band magnitudes brighter than -22, contain at least one emission line or are unambiguously broad absorption line quasars, and have highly reliable redshifts. In the latter cosmological assumptions were required to obtain absolute magnitudes. For this analysis such assumptions were avoided.

But hey, why take his word that he found quantized intervals. Read his paper and follow the instructions. Its a pretty simple procedure. I bet we could walk through it right now.

The SDSS has published the data on publicly accessable SQL server.

You can query all 80,000 qso's yourself, pull out only those that have a scientific rating and a 95% or greater redshift confidence rating and do a fourier analysis on the data right now.

http://cas.sdss.org/astrodr7/en/tools/search/sql.asp

--This query selects top X number of QSOs that are rated research grade with a redshift confidence rating of .95 or higher. Lists Dataset 7 redshift and dataset 5 redshift for all QSOs that appear in the DR5 catalog meeting that criteria.

select top 10 A.z as DR7z, A.zConf as DR7zConfidence, B.z as DR5z, a.sciencePrimary
from SpecObjAll as A, DR5QuasarCatalog as B
where A.specObjId = B.specObjId
and A.sciencePrimary = 1
and A.zConf >= 0.95

To get everything back meeting that criteria, just remove the "top 10" constraint from the query.
Emeritus
PF Gold
P: 4,005
 Quote by Suede Oh here's one: A 2005 paper done by two chinese atstronomers. http://adsabs.harvard.edu//abs/2002MNRAS.336L..13H We have used the publicly available data from the 2dF Galaxy Redshift Survey and the 2dF QSO Redshift Survey to test the hypothesis that there is a periodicity in the redshift distribution of quasi-stellar objects (QSOs) found projected close to foreground galaxies. These data provide by far the largest and most homogeneous sample for such a study, yielding 1647 QSO-galaxy pairs. There is no evidence for a periodicity at the predicted frequency in log(1 +z), or at any other frequency.

Do you have the right paper?

Your link takes you to a paper with this abstract, but it's a 2002 (not 2005) paper, by E. Hawkins, S.J. Maddox and M.R. Merrifield, whose affiliation is stated as "School of Physics & Astronomy, University of Nottingham, Nottingham NG7 2RD, UK".

 To which I quote Scott: Recently mainstream astronomers have joyfully announced that they can find no quantization effects in the observed redshift values of quasars. Of course not! The raw measured total redshift values of the universal set of all known quasars are not quantized. It is the inherent redshift z values that are!
I don't know who Scott is, nor where you got this quote from, but again it seems there's a disconnect ... Hawkins et al. set out to test a specific 'intrinsic redshift' explanation (or model, if you prefer), as presented in papers they reference. Scott seems to be saying that such a test should surely fail because it's testing the wrong hypothesis. If so, then it's not germane.

Do you know if Scott has published his own hypotheses re 'intrinsic redshift'? If so, where?

To conclude: the redshift model Hawkins et al. tested is, it seems, the same one that one of the papers in the OP presents (or very similar to it).
Emeritus
PF Gold
P: 4,005
 Quote by Suede I agree that "Large numbers of contaminated data is not more statistically meaningful than a smaller" however, if you took the time to look at the SDSS datasets, you'd see they have built in confidence definitions for each redshift listed. In fact, if you took the time to read some of the papers I posted, you'd see things like: http://arxiv.org/PS_cache/arxiv/pdf/...712.3833v2.pdf I obtained 80,398 quasar data from the SDSS BestDR6 database (Adelman-McCarthy et al 2007) available at cas.sdss.org/astro/en/. The data (downloaded on the 7th December 2007) were selected by specClass = 3 for QSO and 4 for HIZ_QSO, low redshift and high redshift quasars, respectively. These include any objects with spectra that have been classified by the spectroscopic pipeline as quasars (specClass = QSO or HIZ_QSO). The DR6 data used here were not filtered as was the DR5 quasar catalog, described in Schneider et al. (2007), found at http://www.sdss.org/dr6/products/val...socat_dr5.html. In that case the DR5 catalog quasars were chosen from those that have apparent i-band PSF magnitudes fainter than 15, absolute i-band magnitudes brighter than -22, contain at least one emission line or are unambiguously broad absorption line quasars, and have highly reliable redshifts. In the latter cosmological assumptions were required to obtain absolute magnitudes. For this analysis such assumptions were avoided.
Indeed, that's the first paper you provide a link to, in the OP.

As I have already noted, it's not yet been published (apparently), nor even accepted for publication* ... and reading it carefully I found several rather big shortcomings (whether any relevant journal reviewer would recommend publication, edits, or rejection I cannot say).

 But hey, why take his word that he found quantized intervals. Read his paper and follow the instructions. Its a pretty simple procedure. I bet we could walk through it right now. The SDSS has published the data on publicly accessable SQL server. You can query all 80,000 qso's yourself, pull out only those that have a scientific rating and a 95% or greater redshift confidence rating and do a fourier analysis on the data right now. http://cas.sdss.org/astrodr7/en/tools/search/sql.asp --This query selects top X number of QSOs that are rated research grade with a redshift confidence rating of .95 or higher. Lists Dataset 7 redshift and dataset 5 redshift for all QSOs that appear in the DR5 catalog meeting that criteria. select top 10 A.z as DR7z, A.zConf as DR7zConfidence, B.z as DR5z, a.sciencePrimary from SpecObjAll as A, DR5QuasarCatalog as B where A.specObjId = B.specObjId and A.sciencePrimary = 1 and A.zConf >= 0.95 To get everything back meeting that criteria, just remove the "top 10" constraint from the query.
It almost seems like you are the author - are you?

In any case, the main problems with the paper, that I can see, have little to do with how to query the SDSS databases and perform analyses on the data so obtained; rather, they have to do with selection effects.

In a nutshell, the relevant astronomy literature has many papers on the difficulties of obtaining samples of quasars whose selection biases are both well-understood and well-characterised; the Schneider et al. 2007 paper discusses some of these challenges, and cites other papers which also address them. Unfortunately, Hartnett seems to have chosen to ignore all these, in his paper.

Would you, or any other reader, be interested in discussing some of the possible selection effects that Hartnett seems to ignore?

* if you have information to the contrary, please share it.
 P: 71 No, I'm not the author, and again, selection effects are a result of bias in collection methods. Redshift data that is collected has no bias. IMHO, all that is required in a search for quantization is that the redshift recorded has a high confidence in it's accuracy. I posted 11 papers and a book in support of quantization, papers that refute the posted work are few and far between, and I call their methods into question, just as you seem to have a problem with the way the quantization papers were conducted. If you have a problem with the way the first paper conducted the Fourier analysis, then do your own right here and now. The raw data is simple enough to obtain, I even posted a link to it. Whether the paper is published in a journal or simply accepted as by Arxiv as a scholarly work seems inconsequential to me considering access to the raw data is readily available. I'll wager you can constrain the data down to less than a thousand records, if you are so inclined, and still see peak formation in the analysis.
Emeritus
PF Gold
P: 4,005
I missed this earlier post (bold added)
 Quote by Suede I submit that "significantly" reducing the sample size as they have done will of course alter the findings in their favor. When looking for redshift periodicity, it makes no sense to limit the selection down to a handful of QSOs when an entire raw catalogue of valid data is present. Selection effects in statistics are inherent biases in the collection methods. There are no biases in raw redshift data, it is what it is. The larger the sample, the more accurate your results will be when looking for periodicity. You'd have to convince me otherwise in order for me to accept that papers findings, and considering every stats textbook on the planet says otherwise, I'll go with the paper using the larger sample.
You're being serious, aren't you; this isn't a joke, right?

The SDSS team made their procedures, operations, and observations open; there are several papers on them, plus detailed notes, tutorials, etc on their website.

Let's start with the huge topic of "quasar target selection"*, which is how the SDSS operation (team, telescopes, cameras, software, etc, etc, etc) selected objects for observation by the spectroscopes. By definition, if the spectrum of an object was not taken, then it cannot have been included in the database with a SpecClass flag of either 3 or 4, can it? So if an object in the "astrometrically and photometrically calibrated catalog of objects found in the data from the imaging camera" is NOT selected by the SDSS quasar target selection algorithm but IS, in fact, a quasar, then we have a bias, don't we? And that's not counting the objects in the sky which should have been included in that (imaging) catalogue but weren't (and which, in fact, are quasars).

Here's a key paragraph (I've replaced the link to Richards et al. (2002) with one to the arXiv preprint):

The quasar target selection algorithms are summarized in this schematic flowchart. Because the quasar selection cuts are fairly numerous and detailed, the reader is strongly recommended to refer to Richards et al. (2002) [...] for the full discussion of the sample selection criteria, completeness, target efficiency, and caveats.

Hmm ... "sample selection criteria, completeness, target efficiency, and caveats" ... that sure reads like a list of biases to me, as in "inherent biases in the collection methods"!

ETA: Suede's post #17 and mine (this one) crossed.

* more detail here and here.

 Related Discussions Special & General Relativity 7 Cosmology 1 General Physics 1 Introductory Physics Homework 8 Quantum Physics 4