Tomás E. Müller Bravo
Type Ia (read as “one-a”) Supernovae (SNe Ia) are very energetic explosions with White Dwarf stars as their progenitors. These events display a light curve which luminosity increases rapidly in time, reaching its maximum (with a brightness comparable to that of galaxies!) in a few weeks, and fading away afterwards. However, the exact mechanism of explosion is still under debate.
Despite the fact of not having a complete picture of these events, astronomers noticed a few decades ago that the brighter the explosion is intrinsically, the longer it lasts, and vice-versa. Knowing this, one can standardise the light curves of SNe Ia so they all look (almost) identical, which gives these objects the name of “standardisable candles”. Once this is done, we can apply the same physical principles as with a light bulb here on Earth. For example, we know that the further a light bulb is from us, the fainter it looks and we can estimate its distance by knowing its apparent brightness and intrinsic brightness (which we know beforehand by reading how many Watts it produces). Therefore, we can use the apparent brightness of very distant SNe Ia to measure distances in the Universe, by knowing their intrinsic brightness from very close-by SNe Ia. In fact, thanks to these explosions astronomers discovered the accelerated expansion rate of the Universe (caused by what is known as Dark Energy), by comparing distances with the velocity at which things are moving away from us.
With the years, further improvement has been done to the standardisation of these objects by analysing their light curves, which has resulted in more precise measurement of distances. My research is focused on using machine learning techniques to have a data-driven approach on the standardisation of SNe Ia to improve the precision of distance measurement even more.
What makes these explosions so interesting for astronomers is not only their high luminosity, but also the fact that they can be used to measure distances in the Universe. Their lightcurves can be corrected using empirical relations to display the same luminosity, to later on be used to measure distances. In other words, the lightcurves of SNe Ia are standardised to have the same luminosity, which gives these objects the name of “standardisable candles”. In fact, thanks to these explosions, astronomers discovered two decades ago the accelerated expansion rate of the Universe (caused by Dark Energy) by comparing distances with the velocity at which things are moving away from us. This research field is one of the main focus in several current and future telescope surveys.
Back in 1995, astronomer Mark Phillips noticed that the brighter the SNIa intrinsically is, the longer it lasts, and vice-versa. This “length” in the duration is called stretch. This means that the intrinsic brightness of SNe Ia can be corrected for these stretch parameters to standardise their luminosity. This is known as the Phillips relationship.
To measure distances astronomers apply the same principles as here on Earth. We know that the further a light bulb is from us, the fainter it looks. We can estimate its distance using these physics principles by knowing its apparent (observed) brightness and intrinsic (true) brightness, which we know beforehand by reading how many Watts it produces. So, in a similar way, one can use the apparent brightness of very distant SNe Ia to measure distances in the Universe, by knowing their intrinsic brightness from very close-by SNe Ia.
Since SNe Ia started being used for distance measurement, astronomers have worked on different ways of further correcting their lightcurves to improve the accuracy of their measurements. This is precisely the aim of my research.
Many of the parameters used for SNIa lightcurve correction have an empirical origin, so I decided to use a different approach, a data-driven one. I use Non-negative Matrix Factorization (NMF) to analyse SNIa lightcurves. This is an algorithm that extracts the most relevant features of a dataset. With NMF I decompose SNIa lightcurves and extract their main components. These can be studied to understand the underlying physics of the lightcurves, but additionally, they can be used for their standardisation.
To test my method I used SNe Ia from the Pantheon sample, a compilation of different surveys. The measured distances are compared against distances from a theoretical model of the Universe. The lower the difference between measured and theoretical values, the better. We also compare our values against those obtained with SALT2, a SNIa lightcurve fitting code that extracts its own set of parameters for their standardisation (this code is widely used nowadays). From the comparison one can notice that my method obtains similar results to those of SALT2. However, there is plenty of room for improvement as this is a new approach. Thus, my goal is to improve this method in future work.