How did the re-analysis of supernova data differ from previous studies?

The re-analysis used a principled statistical method called the maximum likelihood estimator and did not assume isotropy on the sky. It also reconsidered and undid corrections made for peculiar velocities in earlier analyses, finding them to be based on out-of-date and potentially unphysical models.

What is the significance of the dipole found in the re-analyzed supernova data?

The re-analysis found that the apparent acceleration of the universe is not isotropic but rather a dipole, with acceleration in one direction and deceleration in the opposite. This suggests the universe might not have an axis of acceleration and challenges the assumption of isotropy and homogeneity in standard cosmological models.

What are the issues with using Baryon Acoustic Oscillations as evidence for dark energy?

The speaker argues that measuring Baryon Acoustic Oscillations requires millions of galaxies and that current sample sizes are insufficient. Claims of detection often rely on fitting to the standard model (Lambda-CDM) rather than independently verifying the evidence, suggesting a confirmation bias.

Why is the concordance between different cosmological datasets potentially misleading?

The speaker suggests concordance might be manufactured due to confirmation bias and that many techniques, while consistent with the standard model, lack independent statistical power. The reliance on simplified models and the cosmic sum rule for CMB data could also contribute to this apparent agreement.

What are the problems with current supernova catalogs, and how do they affect the Hubble tension?

Discrepancies have been found between different supernova catalogs (e.g., JLA and Pantheon) in measured redshifts, exceeding quoted uncertainties. These inconsistencies and the lack of detailed data on peculiar velocity corrections make it premature to claim a 'tension' in Hubble constant measurements, as different catalog choices yield different inferred values.

Can the standard cosmological model be easily modified to account for new findings?

Modifying the standard model is mathematically very complicated because exact solutions to Einstein's equations are hard to find except in highly symmetric situations. While simpler models like the Lemaître–Bondi model can remove evidence for acceleration, they may not fully explain structure formation, highlighting the advantage of the standard model's mathematical simplicity.

Key Moments

How good is the evidence for Dark Energy?

Sabine Hossenfelder

Science & Technology4 min read46 min video

Mar 2, 2020|143,412 views|5,060|764

physics astrophysics universe dark energy does the universe expand does dark energy exist is dark energy real subir sarkar evidence for dark energy cosmology expansion of the universe hossenfelder

Save to Pod

Key Moments

TL;DR

Evidence for dark energy is questioned, with analysis suggesting acceleration might be inferred, not fundamental.

Key Insights

The statistical analysis of supernova data, used to infer dark energy, may be flawed, showing only marginal evidence for acceleration.

A significant dipole anisotropy observed in high-redshift radio galaxies challenges the assumption of universal isotropy.

The assumption that the universe becomes isotropic on large scales (beyond 100 megaparsecs) may be incorrect.

The interpretation of cosmological data, including baryon acoustic oscillations and CMB, relies heavily on a pre-assumed standard model.

Discrepancies in supernova catalogs and potential errors in data analysis raise concerns about the robustness of current cosmological conclusions.

The 'concordance' of evidence for dark energy may be partially manufactured due to confirmation bias and adherence to a century-old cosmological model.

THE SUPERNOVA DATA AND THE MARGINAL EVIDENCE FOR COSMIC ACCELERATION

The Nobel Prize in 2011 was awarded for the discovery of dark energy based on supernova data. However, a reanalysis of this data, made public in 2014, revealed methodological issues. Specifically, the statistical analysis employed by previous researchers involved adjusting error bars to fit the assumed standard model of cosmology, which is problematic for model selection. Using a principled statistical method, the maximum likelihood estimator, researchers found evidence for acceleration to be marginal, less than three standard deviations, falling short of the five-sigma threshold expected for a fundamental discovery. This suggests the evidence for acceleration from supernovae might not be as strong as previously claimed.

CHALLENGING ISOTROPY WITH HIGH-REDSHIFT RADIO GALAXIES

Further research investigated the assumption of isotropy in the universe, which is fundamental to the standard cosmological model. By analyzing high-redshift radio galaxies, astronomers looked for kinematic dipoles caused by motion relative to a uniform distribution of sources. Initial observations by another astronomer were dismissed due to catalog limitations. However, a more rigorous cross-correlation with nearby infrared galaxies helped to eliminate the influence of local radio sources and confirmed a significant dipole. This observed velocity was found to be four times larger than expected and in the same direction as the CMB dipole, challenging the notion of a fully isotropic universe on large scales.

THE QUESTIONABLE ASSUMPTION OF LARGE-SCALE ISOTROPY

The standard cosmological model assumes that the universe becomes homogeneous and isotropic when averaged over large scales, typically beyond 100 megaparsecs. This assumption allows for a simplified mathematical framework. However, the findings from the radio galaxy analysis, along with subsequent investigations using tomography of the local Hubble flow and supernovae data, suggest that peculiar velocities and bulk flows extend much further than 100 megaparsecs. This indicates that the universe may not converge to an isotropic state at the scales assumed, implying a potential flaw in a core tenet of the standard model.

THE RELIANCE ON THE STANDARD MODEL AND POTENTIAL CONFIRMATION BIAS

Much of the evidence cited in support of dark energy, such as baryon acoustic oscillations (BAO), is analyzed under the presupposition of the standard cosmological model (Lambda-CDM). Measuring subtle effects like the BAO peak requires millions of precisely measured galaxy redshifts, which are not yet available. Current analyses often rely on existing templates, leading to confirmation bias where the data is interpreted to fit the expected results. Some studies show that BAO data alone is consistent with models lacking acceleration, suggesting that the perceived concordance might be manufactured rather than a robust finding from independent measurements.

DISCREPANCIES IN DATASETS AND THE HUBBLE TENSION

Significant concerns have been raised regarding the quality and consistency of publicly available cosmological datasets, particularly supernova catalogs like JLA and Pantheon. Discrepancies have been found in redshift measurements between these catalogs, exceeding stated uncertainties. Furthermore, the peculiar velocity corrections applied to supernova data have been criticized as inconsistent and potentially unphysical. These issues impact the calculation of cosmological parameters, including the Hubble constant. The reported 'Hubble tension' between local measurements and early universe inferences might be an artifact of these systemic uncertainties in the supernova data and analysis methods.

MANUFACTURED CONCORDANCE AND THE NEED FOR BLIND ANALYSIS

The author argues that the "concordance" of evidence for the standard cosmological model is partly manufactured. This stems from confirmation bias, the reliance on a 100-year-old model that is mathematically convenient but potentially detached from reality, and the difficulty in deviating from this model due to mathematical complexity. The lack of satisfactory explanations for data discrepancies exacerbates this issue. The author emphasizes the need for blind analysis in cosmology to mitigate confirmation bias, highlighting that current data, while appearing to support the standard model, may not be sufficiently robust to draw definitive conclusions about dark energy.

Mentioned in This Episode

●Products

●Software & Apps

●Organizations

●Studies Cited

●Concepts

●People Referenced

Common Questions

The primary argument is that the statistical analysis of supernova data used to claim dark energy's existence was not principled and assumed the model it was meant to test. Re-analysis suggests the evidence for acceleration is marginal, less than three standard deviations, not the five sigma expected for a discovery.

Topics

Technology & Innovation Science & Mathematics Data Analysis Hubble Tension Statistical Analysis

Mentioned in this video

People

University of Oxford

The location where the interview/discussion took place.

Dominic Schwartz Benefield

Involved with the SKA and aware of the radio source dipole result, advocating for it as a flagship project.

Sue Berserker

Professor of theoretical physics at the University of Oxford, interviewed about his research on dark energy evidence.

Yep a Nissen

Master's student at Copenhagen who worked on the statistical analysis of supernova data.

Albert Ponty

Statistical expert and co-author on the paper re-analyzing supernova data.

Jacques Colin

Astronomer at the Institute of Astrophysics in Paris involved in the anisotropy study.

Mohamed Ramiz

Postdoc who worked with the speaker on anisotropy studies and later analyzed supernova catalog discrepancies.

Saul Perlmutter

Led the Nearby Supernova Factory, which analyzed data to show extended flow.

Adam Riess

One of the Nobel laureates awarded for dark energy discovery, who publicly criticized the re-analysis findings.

Nir Shaviv

Cosmologist who pointed out the Late Integrated Sachs-Wolfe effect as a potential test for lambda.

Wendy Freedman

Led the Hubble Key Project collaboration that measured the Hubble constant using the Hubble Space Telescope.

McClure and Dyer

Researchers who pointed out variations in the Hubble constant measurements on the sky.

Software & Apps

Very Large Array

Radio telescope facility in New Mexico used to catalogue radio sources.

Maximum Likelihood Estimator

A principled statistical method used in the re-analysis of supernova data.

2MASS Survey

Survey used to cross-correlate with radio source catalogs to identify distant objects.

Square Kilometre Array

Future radio telescope project expected to provide much larger catalogs of radio sources.

Pan-STARRS survey

Survey from Hawaii that contributed supernovae data to the Pantheon catalog.

Dark Energy Spectroscopic Instrument

Recently commissioned instrument that will measure millions of redshifts, expected to provide more data on dark energy.

Studies & Research

Union 2

An earlier supernova catalog used to establish a dipole in supernova distribution.

6dF Galaxy Redshift Survey

A survey from Australia that measured peculiar velocities of 11,000 galaxies, supporting the extended flow hypothesis.

JLA catalog

A supernova catalog mentioned in relation to peculiar velocities and later contrasted with the Pantheon catalog.

Dark Energy Survey

Survey that contributed supernova data.

Products

Planck

Space observatory whose measurements of the CMB are referenced in the context of the Hubble tension.

WMAP

Space observatory that provided precise measurements used in cosmological analysis; its data is referenced regarding the consistency of Lambda measurements.

Hubble Space Telescope

Mission that was used for the Hubble Key Project to initially measure the Hubble constant.

Concepts

dark energy

The core topic of the video, with a re-examination of the evidence supporting its existence.

Supernova Data

The primary data set analyzed, with questions raised about the statistical methods used to interpret it for dark energy evidence.

Cosmic microwave background

Ancient light from the universe used to study anisotropy, specifically noting a pronounced dipole.

Friedman Robertson Walker (FRW) cosmology

Standard cosmological model equations drafted nearly 100 years ago, which assumes isotropy.

Tully-Fisher relation

An empirical correlation between a spiral galaxy's rotation speed and its luminosity, used for distance measurement.

Fundamental Plane

A correlation for elliptical galaxies used in distance measurement.

Standard Model of particle physics

A highly confirmed model in particle physics, contrasted with the perceived status quo in cosmology.

Anomalous magnetic moment of the muon

An example of a potential anomaly in particle physics experiments that could lead to progress beyond the Standard Model.

B decays

Another example of anomalies in particle physics that are being investigated for progress beyond the Standard Model.

Baryon Acoustic Oscillations

Features in the Cosmic Microwave Background anisotropy, used as evidence for the standard model of cosmology.

Sakharov oscillations

An earlier name for the phenomenon now known as baryon acoustic oscillations.

Lambda-CDM model

The standard cosmological model that the speaker suggests is overly relied upon and may be based on questionable assumptions.

Cosmic Microwave Background (CMB) fluctuations

Fluctuations that are stated to be insensitive to dark energy by themselves and primarily used to infer energy through the cosmic sum rule.

Cosmic sum rule

A rule in cosmology where matter, curvature, and the cosmological constant sum to one, used to infer dark energy from CMB data.

Late Integrated Sachs-Wolfe effect

An exception where CMB photons passing through inhomogenities after lambda dominates can show a correlation with large-scale structure, acting as a test for lambda.

Einstein's equations

Mentioned as being difficult to solve in non-symmetric situations, which complicates deviating from the standard cosmological model.

Lemaître–Bondi model

A cosmological model that drops the assumption of homogeneity but preserves isotropy, potentially removing evidence for acceleration from supernovae.

Hubble rate

The rate of expansion of the universe, with a claimed 'tension' between local and distant measurements that the speaker suggests might be premature due to data uncertainties.

Organizations

Joint Likelihood Analysis Collaboration

Collaboration that made supernova data public in 2014, enabling independent re-analysis.

Institute of Astrophysics in Paris

Institution where Rui and Jacques Colin worked.

Andromeda

Used as a reference for the distance of a megaparsec.

Nearby Supernova Factory

Professional astronomers who performed analysis on their data, showing extended flow.

Locations

Shapley Supercluster

A large-scale structure at approximately 260 megaparsecs, noted as being within the extended bulk flow.