Are high-z SNe Ia data reliable for measuring the evolution of dark energy?

[Garth]

R. G. Vishwakarma has replaced his paper Recent Supernovae Ia observations tend to rule out all the cosmologies? with the accepted version. He does not say accepted by whom.

4. Conclusion

Supernovae Ia observations have profoundly changed cosmology by predicting an accelerated rate of cosmic expansion, and thus a repulsive dark energy component - an issue which is regarded as an almost mature science now.
However, as more and more accurate data get accumulated, thanks to the remarkable progress made in various types of astrophysical and cosmological observations in recent years, they do not seem to fit any cosmology. The recent observations, taken at their face values, seem to rule out all the cosmologies at fairly high confidence levels. Though these probabilities may not be regarded sufficient to rule out the models completely, however, they are high enough to point out towards the alarming trend of the recent data: as you add newer data to the older samples, the goodness-of-fit-probability from the resulting samples successively decreases. Though the fit improves in some cases if we do not stick to the concordance model, however, this is inconsistent with the anisotropy measurements of CMB which predict a flat space.
The situation has worsened to the extent that the most recent SNe Ia observations made by the Supernova Legacy Survey [12] are analyzed in a way which does not address the goodness-of-fit of the data to the models, rather it assumes that the data have a good fit, and just goes on estimating the parameters of the models. However, it may be noted that unless we have a credible goodness-of-fit, the whole parameter estimation becomes suspect. It must be noted that our result (that the recent observations seem to rule out all the cosmologies at fairly high confidence levels) is deduced from those observations only which, unlike the SNLS data, have already included the intrinsic scatter of the SN absolute magnitude (estimated by reasonable ways) in their error bars.
Assuming that the standard big bang cosmology is correct, the present situation is pointing out towards some flaws in our understanding of the SN Ia phenomenon and towards the futility of the use of SNe Ia in order to constrain cosmological models. We need better understanding of the entire SN Ia phenomenon in order to test the empirical calibrations that are so confidently extrapolated at high redshifts. Similar conclusions have also been drawn by Clocchiatti et al. [11] from a smaller sample of data. However, this is more evident from the present analysis of a bigger sample of data. This view is also supported by Middleditch [13] who argues that SNe Ia seem to be affected by some systematic effects which alone, without invoking any dark energy, could make them too faint for their redshifts. It is argued that it may be impossible to get a clean sample of SNe Ia which are free from this kind of effects [14].

(Emphasis mine)

Does the high-z SNe Ia data record the evolution of Dark Energy, i.e. $\omega$, or the evolution of the SNe Ia?

Garth

 

[Locrian]

Well, how exactly could our understanding of Ia sn be wrong. . . as a function of distance?

 

[turbo]

Well, how exactly could our understanding of Ia sn be wrong. . . as a function of distance?

Arp, the Burbidges, Narlikar and others believe that at least some component of an object's redshift is intrinsic - that is, it is not due to the peculiar motion of the object, nor is it attributable to the Hubble relation. Theirs is one cosmology that may be supported by the SNe Ia data. Here is why:

If galaxies have intrinsic redshifts, the distances to nearby galaxies will be overestimated because all of the redshift is attributed to the Hubble redshift-distance relationship. The redshifts of more distant galaxies will be less contaminated by intrinsic redshift as a percentage of the total, and therefore, their calculated distances (as estimated by the Hubble redshift-distance relationship) will actually be more accurate than the distances of local galaxies. If this is true, the question becomes not "Why are the absolute luminosities of distant SNe Ias fainter than local ones?" but "Why are the absolute luminosities of local SNe Ia brighter than their more distant counterparts?" In the Arp-Burbidge-Narlikar cosmology, the answer is that local galaxies are closer than we think because we have attributed all their redshift to the Hubble relationship and we have failed to recognize that intrinsic redshifts are adding to the cosmological redshift. It may very well be that distant SNe Ia are good standard candles and the over-luminosity of local SNe Ia are a measure of the error in our distances to local galaxies.

The SNe Ia data are not ruling out all cosmologies, but they are supportive of a cosmology that is not very popular or widely-accepted.

 

[Garth]

Well, how exactly could our understanding of Ia sn be wrong. . . as a function of distance?

Simply, because it is assumed that the intrinsic luminosity of the SNe Ia is thought to be well known, and 'on average' does not change over cosmological time or distance.

It is from their apparent magnitude and the luminosity function that their distance is calculated, and that distance compared with their red shift.

This distance-red shift relationship gives a 'handle' on how the universe is expanding and, as the distant SNe Ia are observed to be fainter than previously expected, the universe is consequently thought to be accelerating in its expansion.

This SNe Ia data set is one part of the bedrock of the standard $\Lambda$CDM model.

However if the intrinsic luminosity is not 'on average' constant then that 'bedrock' becomes 'shifting sand'...

Garth

 

[matt.o]

Arp, the Burbidges, Narlikar and others believe that at least some component of an object's redshift is intrinsic - that is, it is not due to the peculiar motion of the object, nor is it attributable to the Hubble relation. Theirs is one cosmology that may be supported by the SNe Ia data. Here is why:

If galaxies have intrinsic redshifts, the distances to nearby galaxies will be overestimated because all of the redshift is attributed to the Hubble redshift-distance relationship. The redshifts of more distant galaxies will be less contaminated by intrinsic redshift as a percentage of the total, and therefore, their calculated distances (as estimated by the Hubble redshift-distance relationship) will actually be more accurate than the distances of local galaxies. If this is true, the question becomes not "Why are the absolute luminosities of distant SNe Ias fainter than local ones?" but "Why are the absolute luminosities of local SNe Ia brighter than their more distant counterparts?" In the Arp-Burbidge-Narlikar cosmology, the answer is that local galaxies are closer than we think because we have attributed all their redshift to the Hubble relationship and we have failed to recognize that intrinsic redshifts are adding to the cosmological redshift. It may very well be that distant SNe Ia are good standard candles and the over-luminosity of local SNe Ia are a measure of the error in our distances to local galaxies.

The SNe Ia data are not ruling out all cosmologies, but they are supportive of a cosmology that is not very popular or widely-accepted.

And at which redshift does a galaxy cease to be local?

 

[turbo]

And at which redshift does a galaxy cease to be local?

There is not a border, but a reduction in the contamination of intrinsic redshift that shrinks with increasing distance.

 

[Garth]

Note: I didn't intend this thread to be primarily a discusion about other cosmologies, just the OP question arising from the OP link paper, i.e. Does the present data support the concept that SNe Ia are standard candles or not?

Garth

 

[SpaceTiger]

Well, how exactly could our understanding of Ia sn be wrong. . . as a function of distance?

Garth gave the correct answer. It is very difficult to constrain the properties of supernovae at high redshift, so we often assume that their properties are similar to those at low redshift. However, there are a number of perfectly good reasons (such as evolving chemical content) to think that their properties might change with redshift, so the question of systematics will always hang over the heads of anyone doing cosmological measurements with SNe Ia.

The initial measurements of the accelerating universe were made with SNe Ia and were greeted with much skepticism for this very reason. However, a number of independent cosmological probes have since lent support to the cosmological model suggested by these SNe measurements, so it's likely that the systematics are not sizable at moderate redshifts (z <~ 1).

 

[Garth]

However, a number of independent cosmological probes have since lent support to the cosmological model suggested by these SNe measurements, so it's likely that the systematics are not sizable at moderate redshifts (z <~ 1).

I agree ST, however does Vishwakarma's paper in the OP link seriously challenge this conclusion?

He claims:

as more and more accurate data get accumulated, thanks to the remarkable progress made in various types of astrophysical and cosmological observations in recent years, they do not seem to fit any cosmology.

The standard model he claims is ruled out at the 96.6% confidence level and

The recent observations, taken at their face values, seem to rule out all the cosmologies at fairly high confidence levels.

quoting various papers from which "fairly high" is referring to over the 90% confidence level.

His conclusion seems to be that systematic errors in the SNe Ia analysis are important at z > 0.5

There is also the question of a correct model for SNe Ia as questioned in John Middleditch's eprint (Report-no: LAUR 06-5685): Core-collapse, GRBs, Type Ia Supernovae, and Cosmology

Type Ia Supernovae (SNe) have been used by many to argue for an accelerated expansion of the universe. However, high velocity and polarized features in many nearby SNe Ia, and their inverse relation to luminosity, particularly for polarization, are consistent with an extreme version of the axisymmetry seen in SN 1987A, which could be the result of double degenerate merger-induced core-collapse. This could be the correct paradigm for many SNe Ia and Ic, where Ia's are both thermonuclear and core-collapsed objects, which leave weakly-magnetized, rapidly spinning (~2 ms) pulsars. In this paradigm Ia/c is produced from the merger of two degenerate cores of common envelope WR stars, or of two CO white dwarfs. Its polar blowouts produce the observed high velocity and polarized spectral features in Ia's, and its equatorial bulge is much brighter in Ia's, due to the greater fraction of 56Ni contained within it. These become classified as Ia's when viewed from the merger equator, and Ic's when viewed from the poles, where a hypernova signature and a gamma-ray burst (GRB) will be observed for lines of sight close to the merger axis. Thus cosmology determined strictly from Ia's alone may be flawed because the local sample is selectively biased.

(emphasis mine)

Now I am not advocating Middleditch's model in particular, just using it to show that we do not actually know what SNe Ia are for certain, only that they exhibit weak hydrogen lines (therefore Type I - stripped hydrogen envelope progenitors?) and a strong Si II line with other intermediate 'metals': Nickel-56 through Cobalt-56 to Iron-56, (accreting white dwarf detonation?) and the local ones have very similar absolute luminosity profiles suggesting their role as 'standard candles'.

However, they could also be the result of stellar mergers as proposed in the Middleditch paper. In which case their luminosity may be a function of directionality as well.

Other systematics may also be involved, such as chemical evolution with cosmological age as ST mentioned, or possibly a selection effect whereby there may be a delay in detecting the more distant and fainter SN until later on in their light curve when the luminosity has died down somewhat.

As an example of chemical evolution: the radio-active decay of the Nickel–56 is responsible for the brightest part of the SNe Ia luminosity curve, so a Nickel deficiency in early high-z SNe Ia progenitors could also explain their relative faintness.

Garth

 

[oldman]

Statistical sophistry

Does the present data support the concept that SNe Ia are standard candles or not?

Garth

When this kind of data is simply plotted (as in Fig 10.7 of Kirschner's Extravagent Universe), rather than statistically analysed, it is difficult to ignore its considerable scatter and become convinced that it is of model-deciding significance.

Small effects based on statistical analyses often dissolve as observations and/or analysis improve. Weber's original claims to have observed gravitational waves are one ancient example I can think of.

 

[Locrian]

Garth gave the correct answer. It is very difficult to constrain the properties of supernovae at high redshift, so we often assume that their properties are similar to those at low redshift. However, there are a number of perfectly good reasons (such as evolving chemical content) to think that their properties might change with redshift, so the question of systematics will always hang over the heads of anyone doing cosmological measurements with SNe Ia.

Well obbviously we understand IaSN at least well enough such that we aren't getting wild fluctuations that prevent us from making a plot at all. It seems clear to me they are some sort of standard candle - all one can say is that we're misreading them, which to me is more specific. In reference to changing chemistry - is there any reason to believe that all objects that go IaSN have the same chemistry? I mean, is the variation in chemistry over the distance of low to high z SN greater than random variation between Ia dwarfs of the same z?

 

[Locrian]

Arp, the Burbidges, Narlikar and others believe that at least some component of an object's redshift is intrinsic - that is, it is not due to the peculiar motion of the object, nor is it attributable to the Hubble relation. Theirs is one cosmology that may be supported by the SNe Ia data.

That's very interesting. However, I don't think it has any value until this "intrinsic" redshift has been measured separately. I'm sorry for not doing the research on my own - do they argue this has been independently measured?

 

[Locrian]

As an example of chemical evolution: the radio-active decay of the Nickel–56 is responsible for the brightest part of the SNe Ia luminosity curve, so a Nickel deficiency in early high-z SNe Ia progenitors could also explain their relative faintness.

Garth

Thanks for mentioning this, I will look into it.

 

[matt.o]

There is not a border, but a reduction in the contamination of intrinsic redshift that shrinks with increasing distance.

Ok, I'll re-phrase: Where should the effects of the intrinsic redshift be most prevalent? Or, how much does this intrinsic redshift contribute to the observed redshift and is the intrinsic redshift the same for each galaxy?

I think it is important to note that whilst SNIa evolution is an ongoing research subject (it isn't being swept under the rug; just do an author search for 'salvo' on adsabs.harvard.edu), I think it is also important to note that other studies such as baryon acoustic ocillations and WMAP all show convergence to a dark energy dominated universe. Hopefully in the near future the error bars will be tightened and there will be better constraints.

 

[Garth]

Hi matt.o.

I think it is also important to note that other studies such as baryon acoustic ocillations and WMAP all show convergence to a dark energy dominated universe. Hopefully in the near future the error bars will be tightened and there will be better constraints.

But this analysis is model dependent, and the OP link paper suggested that no model could be made to fit if the SNe Ia were standard candles. What should we make of this strong assertion?

Garth

 

[Garth]

In Tuesday's ArXiv, Mazzali, Ropke, Benetti and Hillebrandt's paper (Journal-ref: Science 315, 825 (2007) DOI: 10.1126/science.1136259): A Common Explosion Mechanism for Type Ia Supernovae.

Abstract
Type Ia supernovae, the thermonuclear explosions of white dwarf stars composed of carbon and oxygen, were instrumental as distance indicators in establishing the acceleration of the universe's expansion. However, the physics of the explosion are debated. Here we report a systematic spectral analysis of a large sample of well observed type Ia supernovae. Mapping the velocity distribution of the main products of nuclear burning, we constrain theoretical scenarios. We find that all supernovae have low-velocity cores of stable iron-group elements. Outside this core, nickel-56 dominates the supernova ejecta. The outer extent of the iron-group material depends on the amount of nickel-56 and coincides with the inner extent of silicon, the principal product of incomplete burning. The outer extent of the bulk of silicon is similar in all SNe, having an expansion velocity of ~11000 km/s and corresponding to a mass of slightly over one solar mass. This indicates that all the supernovae considered here burned similar masses, and suggests that their progenitors had the same mass. Synthetic light curve parameters and three-dimensional explosion simulations support this interpretation. A single explosion scenario, possibly a delayed detonation, may thus explain most type Ia supernovae.

(emphasis mine)
These progenitors were of the same mass and were all 'nearby'.

We derive the distribution of the principal elements in 23 nearby SNeIa (distances <40Mpc) with good spectral coverage extending from before maximum to the late nebular phase, about one year later.

But how can we be sure that at further distances z ~ 1 the masses of the progenitors and/or the N-56 abundance are the same as the nearby ones?

Garth

 

[Garth]

There may also be two separate populations of SNe Ia Two populations of progenitors for type Ia SNe?.

Furthermore, the Fe abundance seems to be evolving over cosmological history: On the evolution of the Fe abundance and of the Type Ia SN rate in clusters of galaxies. Note this is an example of chemical evolution over cosmological time. The Ni-56 is produced from Carbon and Oxygen burning. Evolution of one element, Iron, might also signify evolution of these SNe Ia fuels thus affecting the amount of Ni-56 and the intrinisic maximum of the luminosity curve.

Garth

 

[Garth]

From today's ArXiv:

SN 2005hj: Evidence for Two Classes of Normal-Bright SNe Ia and Implications for Cosmology by Robert Quimby, Peter Höflich & J. Craig Wheeler., accepted for publication in Ap.J.

HET Optical spectra covering the evolution from about 6 days before to about 5 weeks after maximum light and the ROTSE-IIIb unfiltered light curve of the "Branch-normal" Type Ia Supernova SN 2005hj are presented. The host galaxy shows HII region lines at redshift of z=0.0574, which puts the peak unfiltered absolute magnitude at a somewhat over-luminous -19.6. The spectra show weak and narrow SiII lines, and for a period of at least 10 days beginning around maximum light these profiles do not change in width or depth and they indicate a constant expansion velocity of ~10,600 km/s. We analyzed the observations based on detailed radiation dynamical models in the literature. Whereas delayed detonation and deflagration models have been used to explain the majority of SNe Ia, they do not predict a long velocity plateau in the SiII minimum with an unvarying line profile. Pulsating delayed detonations and merger scenarios form shell-like density structures with properties mostly related to the mass of the shell, M_shell, and we discuss how these models may explain the observed SiII line evolution; however, these models are based on spherical calculations and other possibilities may exist. SN 2005hj is consistent with respect to the onset, duration, and velocity of the plateau, the peak luminosity and, within the uncertainties, with the intrinsic colors for models with M_shell=0.2 M_sun. Our analysis suggests a distinct class of events hidden within the Branch-normal SNe Ia. If the predicted relations between observables are confirmed, they may provide a way to separate these two groups. We discuss the implications of two distinct progenitor classes on cosmological studies employing SNe Ia, including possible differences in the peak luminosity to light curve width relation.

Which highlights uncertainties in the modelling of the SNe Ia.

Garth

 

[Chronos]

Not terribly discordant. The physics behind SNe 1a are fairly well understood, so the candle model is pretty robust. What is not well understood is extinction effects.

 

[Garth]

Not terribly discordant. The physics behind SNe 1a are fairly well understood, so the candle model is pretty robust. What is not well understood is extinction effects.

A recent paper accepted for publication in a special edition of General Relativity and Gravitation by Professor Sarkar (Oxford University) Is the evidence for dark energy secure? suggests alternative explanations for the standard interpretation of cosmological data, including SNe Ia, which normally support the hypothesis of DE.

He says:

According to the second Friedmann equation (6) an accelerating expansion rate requires the dominant component of the universe to have negative pressure. The more mundane alternative possibility, namely that the SNe Ia appear fainter because of absorption by intervening dust, can be observationally constrained since this would also lead to characteristic reddening, unless the dust has unusual properties [2]. It is more difficult to rule out that the dimming is due to evolution, i.e. that the distant SNe Ia (which exploded over 5 Gyr ago) are intrinsically fainter by ∼ 25% (e.g. [26]. Although SNe Ia are believed to result from the thermonuclear explosion of a white dwarf, there is no “standard model” for the progenitor(s) (see [38]), hence there may well be luminosity evolution which would complicate the use of SNe Ia as ‘standard candles’.

He continues:

However it is known (using nearby SNe Ia with independently measured distances) that their time evolution is tightly correlated with their peak luminosities such that the intrinsically brighter ones fade faster. This can be used to make corrections to reduce the scatter in the Hubble diagram using various empirical methods such as a ‘stretch factor’ to normalise the observed apparent peak magnitudes [58] or the ‘Multi-colour Light Curve Shape’ method [60]. Such corrections are essential to reduce the scatter in the data sufficiently so as to allow meaningful deductions to be made about the cosmological model. It is a matter of concern that the corrections made by different methods do not always correlate with each other when applied to the same objects (see [51]), especially since there is no physical understanding of the observed correlations.

(Emphasis mine)

Garth

 

[Garth]

In my post #17 I reiterated my point about iron abundance affecting SNe Ia luminosity and in #18 the possibility of there being two distinct classes of SNe Ia. (This latter point has resonance with the PopI and PopII Cepheid Variable confusion that led to the pre-war evaluations of Hubble's constant being a factor~2.5 time too large)

In today's ArXiv we have a little more light thrown on the subject: The Outermost Ejecta of Type Ia Supernovae

The properties of the highest velocity ejecta of normal Type Ia supernovae (SNe Ia) are studied via models of very early optical spectra of 6 SNe. At epochs earlier than 1 week before maximum, SNe with a rapidly evolving Si II 6355 line velocity (HVG) have a larger photospheric velocity than SNe with a slowly evolving Si II 6355 line velocity (LVG). Since the two groups have comparable luminosities, the temperature at the photosphere is higher in LVG SNe. This explains the different overall spectral appearance of HVG and LVG SNe. However, the variation of the Ca II and Si II absorptions at the highest velocities (v >~ 20,000 km/s) suggests that additional factors, such as asphericity or different abundances in the progenitor white dwarf, affect the outermost layers. The C II 6578 line is marginally detected in 3 LVG SNe, suggesting that LVG undergo less intense burning. The carbon mass fraction is small, only less than 0.01 near the photosphere, so that he mass of unburned C is only <~ 0.01 Msun. Radioactive 56Ni and stable Fe are detected in both LVG and HVG SNe. Different Fe-group abundances in the outer layers may be one of the reasons for spectral diversity among SNe Ia at the earliest times. The diversity among SNe Ia at the earliest phases could also indicate an intrinsic dispersion in the width-luminosity relation of the light curve.

So it seems their 'standardness' is vindicated.(?)

Garth