Previous Article in event
Previous Article in session
Next Article in event
Cosmology in Tension
Published:
17 November 2019
by MDPI
in 5th International Electronic Conference on Entropy and Its Applications
session Astrophysics, Cosmology, and Black Holes
Abstract:
The Cosmic Microwave Background (CMB) temperature and polarization anisotropy measurements from the Planck mission have provided a strong confirmation of the Lambda Cold Dark Matter (LCDM) model of structure formation. However, there are a few interesting tensions with other cosmological probes that leave the door open to possible extensions to LCDM. I will review some interesting extended cosmological scenarios, in order to find a new concordance model that could explain and relieve tensions in current cosmological data.
Keywords: CMB; Hubble constant; tension cosmology; dark energy
Comments on this paper
Geert Verdoolaege
19 November 2019
Multimodal distributions and a question of methodology
Dear author,
Many thanks for an interesting presentation. I have two questions, one specific and one more general.
First, when seeing the various data sets that seem to have systematic discrepancies, I am somewhat surprised that the joint distribution is always unimodal. Are there circumstances under which instead you get multimodal distributions when jointly analyzing these data?
Second, for me, working in the field of data science for nuclear fusion experiments, it is always pleasant to see that, in astrophysics and cosmology, the application of modern data science methods, like Bayesian inference, seems to be well embedded in the mainstream research. This is not really the case in fusion science, where still only a minority of researchers use such methods (although the situation is improving). Would you say that, in your field, Bayesian methods are widely used and have been generally accepted as an important and necessary analysis tool? If so, do you have any suggestions towards the fusion community on how to help accelerating the adoption of such methods (Bayesian inference, machine learning, etc.)?
Many thanks for your replies.
Many thanks for an interesting presentation. I have two questions, one specific and one more general.
First, when seeing the various data sets that seem to have systematic discrepancies, I am somewhat surprised that the joint distribution is always unimodal. Are there circumstances under which instead you get multimodal distributions when jointly analyzing these data?
Second, for me, working in the field of data science for nuclear fusion experiments, it is always pleasant to see that, in astrophysics and cosmology, the application of modern data science methods, like Bayesian inference, seems to be well embedded in the mainstream research. This is not really the case in fusion science, where still only a minority of researchers use such methods (although the situation is improving). Would you say that, in your field, Bayesian methods are widely used and have been generally accepted as an important and necessary analysis tool? If so, do you have any suggestions towards the fusion community on how to help accelerating the adoption of such methods (Bayesian inference, machine learning, etc.)?
Many thanks for your replies.
Eleonora Di Valentino
19 November 2019
Thanks for your questions.
A multimodal distribution can be present on a particular parameter when the different peaks in the posterior have almost the same probability. Here we have a multiparameters analysis and the cosmological probes, different from the CMB, only constrain some of them. Their constraining power is then not enough to create a multimodal distribution (because of the correlations between the parameters), but enough to shift the posterior of the parameters they are constraining.
On the second question, in cosmology the Bayesian method is the most important analysis tool. I believe the difficulty in the adoption of this method in other fields is the computational effort. We have just one realization of the Universe, while in the nuclear fusion field you can reproduce your experiment several time making use of quicker analysis methods.
A multimodal distribution can be present on a particular parameter when the different peaks in the posterior have almost the same probability. Here we have a multiparameters analysis and the cosmological probes, different from the CMB, only constrain some of them. Their constraining power is then not enough to create a multimodal distribution (because of the correlations between the parameters), but enough to shift the posterior of the parameters they are constraining.
On the second question, in cosmology the Bayesian method is the most important analysis tool. I believe the difficulty in the adoption of this method in other fields is the computational effort. We have just one realization of the Universe, while in the nuclear fusion field you can reproduce your experiment several time making use of quicker analysis methods.
