Type 1a SN -not so standard candles?

[kzb]

Copied from the Fun Papers in Arxiv thread (by Antoniseb):

*Brightness of Double Degenerate Type 1a SN* http://arxiv.org/abs/1209.0645 I wondered about this about half a year ago when it starting seeming like the majority of type 1a supernovae were pairs of white dwarfs merging, rather than the previous main idea, a single WD accreting the straw that broke the camels back. This makes them a bit less of a precisely standard candle, but just how much deviation should we be seeing? This paper examines that.

If there are different mechanisms potentially responsible for an observation of a Type 1a supernova, how does that affect its status as a standard candle?

If we also allow that either the average relative frequency, or the average size of, these second types, has evolved over cosmic time....well you can see where I am going. Is this not a hole in the accelerating expansion of the universe theory?

[antoniseb]

...If there are different mechanisms potentially responsible for an observation of a Type 1a supernova, how does that affect its status as a standard candle?

If we also allow that either the average relative frequency, or the average size of, these second types, has evolved over cosmic time....well you can see where I am going. Is this not a hole in the accelerating expansion of the universe theory?

Your question about evolution is what the paper cited is working on... and it appears that there is some evolution, but not enough to seriously alter the results so far... As to the variation, the answer is as Jerry recently pointed out, you can't get a bead on the distance from a single event, but you can get a really good sense of distance-redshift relationship from statistically large samples, which is what we do now.

[StupendousMan]

We observe a range of luminosities for Type Ia SNe now. Fortunately, there appear to be other observable quantities which correlate with the peak luminosity. That means that we can recognize that a particular SN is more or less luminous than average and account for the difference in our analysis.

As long as there are other observable quantities which correlate with luminosity, it may be possible to use Type Ia SNe for cosmological purposes.

[Tensor]

We observe a range of luminosities for Type Ia SNe now. Fortunately, there appear to be other observable quantities which correlate with the peak luminosity. That means that we can recognize that a particular SN is more or less luminous than average and account for the difference in our analysis.

As long as there are other observable quantities which correlate with luminosity, it may be possible to use Type Ia SNe for cosmological purposes.

Do you know of a reason why critics of using SNe Ia for distance, continue to fixate on peak luminosity and not get into the other correlating features?

[kzb]

Your question about evolution is what the paper cited is working on... and it appears that there is some evolution, but not enough to seriously alter the results so far... As to the variation, the answer is as Jerry recently pointed out, you can't get a bead on the distance from a single event, but you can get a really good sense of distance-redshift relationship from statistically large samples, which is what we do now.

But this does affect statistically large samples, and in either (or both) of two ways. If you have two types of 1a SNs, either the number ratio of one type to the other type changes over time, or the average size of one type changes over time, thereby affecting the bulk average.

[antoniseb]

But this does affect statistically large samples, and in either (or both) of two ways. If you have two types of 1a SNs, either the number ratio of one type to the other type changes over time, or the average size of one type changes over time, thereby affecting the bulk average.

Right... but the paper you noted in the OP here shows that this evolution is very slow, and for the epochs where we have SN1a observations, not very big compared to the effect they have demonstrated.

[StupendousMan]

Do you know of a reason why critics of using SNe Ia for distance, continue to fixate on peak luminosity and not get into the other correlating features?

Because they have other reasons for disagreeing with the SNe-based cosmologies?

[Jerry]

Do you know of a reason why critics of using SNe Ia for distance, continue to fixate on peak luminosity and not get into the other correlating features?

Because they have other reasons for disagreeing with the SNe-based cosmologies?

When I was a chemist, my manager, who was also an engineer, dragged me into his office and complained that I kept changing the test methods and screwing up his baseline. He didn't want to hear my argument that he was getting a better number, or that it was our charter as a 'Method Development Laboratory' to develop and improve upon methods; he wanted to be able to compare yesterday's results with todays.

Supernova researchers are in the same boat: The evolution in optics and survey strategy has led to progressively more complex, and hopefully more accurate data reduction routines. For those of us sitting on the sidelines scoring the progress, it is frustrating to handed a new set of rules with each and every paper. The peak luminosity of local events is one of the easier metrics to score, and every time an event occurs that broadens the known absolute magnitude of type Ia Supernovae, the shakier the current distance ladder gets. The known range of events is much broader than it was when 'dark energy' resurfaced. That is important.

Personally, I am much more interested in the light curve length; and how that length changes with red-shift distance, because I still think that at least a portion of the redshift is not relativistic - I think it is due to radiation transfer. If our observations continue to progressively identify shorter light curves with increasing distance; this may be an evolutionary factor that runs opposite to known selection effects, but it may also be stark proof that there is a measurable contribution to light curve redshift from radiation transfer.

[Jerry]

http://arxiv.org/pdf/1209.1339v1.pdf Spectral modelling of the “Super-Chandra” Type Ia SN 2009dc – Testing a 2 M⊙ white dwarf explosion model and alternatives
Yet another superlumious beast with a light curve well beyond the maximum length assumptions used by Leibengut to rule-out all non-relativistic explanations for the cosmic redshift.

[George]

Peanut Gallery asks....

Thinking about Cepheid history, is this deja vu all over again?

Wouldn't the time dilation of the light curves, however, greatly constrain the amount of variance in the expansion constant? [Or are the Super Chandeskehar SNs mentioned too close for this? I didn't see anything (cursory scan) about time dilation mentioned.]

Here is more on the SC 2009dc showing sphericty due to their polorization results, yet the ejecta may be a bit clumpy for IMEs (eg Si) above the 56Ni layer. I don't know what it means, but it's somethin'.

[Jerry]

Peanut Gallery asks....

Thinking about Cepheid history, is this deja vu all over again?

Sometimes I think it is in nature's nature to conspire.

Here[/URL]]Despite the large 56Ni production, the expansion velocity of SN 2003fg (~ 8, 000 km s−1 near the maximum brightness) is slower than normal Type Ia SNe (~ 10, 000 km s−1). From these facts, they suggested that the progenitor of SN 2003fg has a super-Chandrasekhar mass(~ 2M(sun)).

One of the supporting evidences of time dilation in the most distant supernova events, is that the expansion velocity is slower than in your garden-variety local events, thus demonstrating time dilation. Although I started arguing ten years ago that a case could be made for a class of more massive superluminous events; it didn't seem to me like a more luminous event should have a slower expansion velocity. Yet this is one of the defining signitures of superluminous events: They have longer light curves, expanding and chemically evolving more slowly. It would take a combination of all three of these attributes to fool researchers into thinking that the most distant events are normal supernova. The other defining feature in superluminous events is that the differential UV light curves are brighter. Since the most distant events studied have brighter UV signatures, a good case can be made for questioning the current estimates of their absolute magnitudes...and true light curve lengths.

Wouldn't the time dilation of the light curves, however, greatly constrain the amount of variance in the expansion constant? [Or are the Super Chandeskehar SNs mentioned too close for this? I didn't see anything (cursory scan) about time dilation mentioned.

These are too close for time dilation effects to be discriminatory.

[antoniseb]

... They have longer light curves, expanding and chemically evolving more slowly. ...

I'm not sure what you're saying here. Most of the light from the peak weeks of the light curve comes from the decay of Ni56 into Fe. The Ni56 arrives because of the decay of Cobalt, and disappears as it becomes Iron. This curve is not going to change, regardless of how much Cobalt is made initially. The decay rates of these things are constants.

[George]

I'm not sure what you're saying here. Most of the light from the peak weeks of the light curve comes from the decay of Ni56 into Fe. The Ni56 arrives because of the decay of Cobalt, and disappears as it becomes Iron. This curve is not going to change, regardless of how much Cobalt is made initially. The decay rates of these things are constants.

So in georgeeze -- and for the sake of my fellow peanuters -- you are saying the luminosity is fixed because the half life is constant for these decaying elements, thus the dimming seen in the light curve is unaffected by the possible difference in size of the SC SN, compared with regular Type Ia supernovae.

Yet the bigger the ball of a glowing gas, the brigher it looks (i.e. apparent magnitude). And Jerry seems to be saying that size does matter, namely that a slower expansion of the extremely bright shell will appear more dim than normal and will take longer to fade away.

I would guess, however, that the SED (spectral energy distribution) would have a lot to say about this. Perhaps the outer layers beyond Ni56 suppress the light, but this too should be quite noticeable in the SED as strong absorption lines. Sphericity also becomes important since a spherical blast will have the same apparent magnitude regardless of what side we are on. The SC SN mentioned seem to have spherical expansions (thankfully).

Intriguing stuff.

[StupendousMan]

Most of the light from the peak weeks of the light curve comes from the decay of Ni56 into Fe. The Ni56 arrives because of the decay of Cobalt, and disappears as it becomes Iron. This curve is not going to change, regardless of how much Cobalt is made initially. The decay rates of these things are constants.

While the first half of the above is true -- the radioactive decay of iron-group elements provides the majority of the energy to SNe Ia -- the second half is not so true. The rate of decline of a Type Ia SN after maximum can vary quite a bit, depending on the speed and density profile and chemical composition of the ejecta; and those can depend on the details of the explosion, which can vary from event to event.

Compare, for example, the light curve of SN 1991bg with that of SN 1991T.

[antoniseb]

... -- the second half is not so true. ...

My understanding it that the time at which other factors can have major impacts on the shape of the light curve comes some time after the peak luminosity for most type 1a SN. So, yes, other factors play a role, but the first 24 days are pretty consistent (Stupendousman, please correct me if I am wrong about even this). For George, let me say that the total luminosity might be varying, but the shape of the curve is very similar every time while this Ni56 thing is the dominant contributor.

[StupendousMan]

Ah, well, we're into a land of qualifiers now, I guess. The light curves are very similar, and yet do show some measureable differences. I guess we'd have to get quantitative to settle the issue, and that would take a long time and go into specific cases which would become complicated quickly.

We're all correct! Yay!

:-)

[Jerry]

My understanding it that the time at which other factors can have major impacts on the shape of the light curve comes some time after the peak luminosity for most type 1a SN. So, yes, other factors play a role, but the first 24 days are pretty consistent (Stupendousman, please correct me if I am wrong about even this). For George, let me say that the total luminosity might be varying, but the shape of the curve is very similar every time while this Ni56 thing is the dominant contributor.

The first thirty days of a supernova ia event; the rise time to peak magnitude and the immediate fall-off are highly consistent for the the very reason you describe. The difficulty, is that the earliest stage is rarely captured and especially hard to quantify in the most distant sample. There are several methods to quantify the peak magnitude and the light curve width, which must be quantified to correct for the well-known correlation between light curve width and luminosity.

One method of quantifying the light curve width is the "Magnitude in blue/15"; which is a curious normalization that quantifies how many days it takes an event to lose half the peak magnitude. The quanification is inversely normalized around 15 days. (An 'M(B)15' of '1.0' means an event loses half the peak magnitude in exactly 15 days.) The smaller the number, the longer the light curve lasts, so a light curve that loses half of its peak in thirty days has an M(B)15 value of 1-(15/30)=0.5.

Most Supernovae Ia have an M(B)15 of between 0.8 and 1.1. Any change in the average M(B)15 with increasing supernova distance would be indicative of either: 1) Evolution of Supernova events 2) selection effects 3) systemic error in the data reduction, due to [k corrections, dust extinction, curve fitting, galactic contamination...]. One of the reasons researchers have concluded the rate of expansion of the universe is increasing, is that virtually all of the known data reduction errors (and including selection effects) would fall on the side of greater acceleration, rather than slower expansion. In a worst case scenario, the combined errors do make the current 'consensus model' of the universe untenable.

Another method is the 'Stretch Factor'; which is also a inverse normalized value following roughly the same convention as the M(b)15. Researchers could be much more straight forward about this: Why not just publish how many days it takes for the event to lose half of the peak luminosity? This science is difficult enough without being deliberately obtuse.

The M(B)15 value of SN2009cd is less than 0.5, which is truly extraordinary. (The rise time is constrained to no less than 21 days, which is also way outside the curve for type Ia events: typically 8 to 14 days, imcmc). If a SN2009cd-like event was observed at a redshift of 1, the rise time, after time dilation (but before correction for time dilation) would appear to be at least 42 days. This observation would prove to me that time dilation is as fundamental as astrophysicists assume it to be. Selection effects should favor the observation of this type of highly luminous events at the greatest cosmological distances. Why can't we find them?

[StupendousMan]

One method of quantifying the light curve width is the "Magnitude in blue/15"; which is a curious normalization that quantifies how many days it takes an event to lose half the peak magnitude. The quanification is inversely normalized around 15 days. (An 'M(B)15' of '1.0' means an event loses half the peak magnitude in exactly 15 days.) The smaller the number, the longer the light curve lasts, so a light curve that loses half of its peak in thirty days has an M(B)15 value of 1-(15/30)=0.5.

Actually, the parameter M(B)15 is defined as the difference in magnitude in the B filter between the peak of the light curve (in B) and the brightness 15 days later. In other words, if a SN reaches its peak B-band magnitude in June 1 at B=10.0, and then has magnitude B=11.2 on June 16, M(B)15 = 11.2 - 10.0 = 1.2. This quantity is simply defined and simple to measure: look at the light curve in B -- how much does it drop in the 15 days past maximum?

You can find this definition in the paper by Phillips, ApJ 413, L105 (1993).

http://adsabs.harvard.edu/abs/1993ApJ...413L.105P

[Jerry]

Actually, the parameter M(B)15 is defined as the difference in magnitude in the B filter between the peak of the light curve (in B) and the brightness 15 days later. In other words, if a SN reaches its peak B-band magnitude in June 1 at B=10.0, and then has magnitude B=11.2 on June 16, M(B)15 = 11.2 - 10.0 = 1.2. This quantity is simply defined and simple to measure: look at the light curve in B -- how much does it drop in the 15 days past maximum?

You can find this definition in the paper by Phillips, ApJ 413, L105 (1993).

http://adsabs.harvard.edu/abs/1993ApJ...413L.105P

Opps - I have the M(B)15 mixed up with another system, introduced by Hamuy...I think. M(B)15's for 'typical' events run about 1.1 , while real fast decliners (subluminous events) can lose ~ 2 magnitudes in 15 days. Superluminous Slow decliners have values considerably less than one, so the smaller numbers do reflect more luminous events.