Dark Bites

18h

We disguise our data before looking at the final results of our analysis🤡We indeed want to avoid adjusting our anal… https://t.co/nFXWatoI3k

reply retweet favorite

24h

In case you missed the first one, tomorrow there is another 'virtual launch' event for the DES book! Several DEScie… https://t.co/jvPUvOVHV8

reply retweet favorite

Nov 23

Get into statistics. It'll change your life. #ThisIsDES full post at https://t.co/RxWjncU259 📸: Carles Sanchez https://t.co/qe6NH0d310

reply retweet favorite

Nov 21

Galaxies are like the visible tip of gigantic icebergs of dark matter 🧊 We measure galaxy clustering, not dark matt… https://t.co/D5T5uyfqeR

reply retweet favorite

Nov 19

Galaxies...it's complicated. In cosmology, we like to pretend we're above it all, working on the largest scales,… https://t.co/PMSKXTGyGb

reply retweet favorite

Facebook

18 hours ago

Day 32. Unbiased science 🎯

Let’s face it, humans are pretty biased creatures. This graph above shows the measurement of the neutron’s (one of the atom’s component) lifetime from different experiments at different time. See how in the 60's people independently measured it to be around 1010 seconds, and then in the 70's someone measured it to be around 940 seconds and people measured the same for years, although it is actually around 890 seconds! ❌This is because people knew about earlier results and when they got something not coherent with previous measurements, they would think they did something wrong and tried to obtain the same value as the previous result! 🧮 This is called confirmation bias.

In our analysis, we’d like this bias not to creep into our scientific experiments. 🔭Between our data and the model, there are many knobs to switch and dials to turn in our analysis choices. 🎛 To do this in the most scientifically resilient manner, we make our measurements and before anything is done with them - before they’re even accessible to the team, we disguise them. 🤡 That way, we can do our entire analysis based on reason, without knowing the results until the very end - ta da, Science! 👌🏾📏
#ThisIsDES

📸 : from K.A. Olive et al., 2014 ... See MoreSee Less

Share on Facebook Share on Twitter Share on Linked In Share by Email

24 hours ago

In case you missed the first one, tomorrow there is another 'virtual launch' event for the DES book! Several DEScientists will be on the panel.Join the editors and contributors of “The Dark Energy Survey – The Story of a Cosmological Experiment” to celebrate the release of this title which discusses the results and processes behind this project.

Find out how our Universe functions, why it is expanding and why this expansion is accelerating.

Register here: bit.ly/3n8lvhc ... See MoreSee Less

Share on Facebook Share on Twitter Share on Linked In Share by Email

2 days ago

Day 31: Bayesian statistical analysis

Over the past 30 days we’ve described the various complex measurements that we make on the images of the sky taken by our telescope. We’ve also talked about the cosmological models that we think describe our Universe; in some sense these are beautifully simple, with a handful of simple ingredients 🍲 like the densities of dark matter and dark energy. We use Bayesian statistical analysis 📈📉to connect these two - to “constrain” i.e. estimate the values of these simple ingredients, which we call “parameters”, by using the information in our measurements.

In Bayesian analysis, we update whatever we knew before about the model parameters (this is called the prior), with whatever new information our data tells us. This lets us assign probabilities to possible ranges of parameters, for example, we can make probabilistic statements such as “there is an X% probability that the density of dark energy is not zero”. Here X is a number close to 100 - we are very confident that there is dark energy in the Universe, or something that acts like it. While we’ll never measure it’s density *exactly*, we can make statements about our confidence 👍👎 that it is in some range.

One crucial tool for Bayesian analysis is a sampler. A sampler is a Monte Carlo algorithm that explores the space of allowed parameters , which can then be used to generate a model to compare to the data. It’s super important to make a sampler very efficient, because the space of allowed parameters can be high dimensional, and the model can be slow to generate. The movie shows a sampler exploring a two-dimensional parameter space. As we get enough samples, we can see the region where there is a high probability or confidence that the true parameters reside. 🐛🐛

Thanks to Carles Sanchez for the cool video! ... See MoreSee Less

Video

Share on Facebook Share on Twitter Share on Linked In Share by Email

5 days ago

Day 30. Galaxy bias ➕

In the universe, galaxies are like the visible tip of gigantic icebergs of dark matter. 🧊 Imagine galaxies living in gigantic haloes of dark matter! ⚫️ The correspondence between these haloes of invisible dark matter and galaxies is called the galaxy bias. Why do we call it “bias”❓Because our measurements of how clustered galaxies are are “biased” compared to what one would get if we could measure how clustered dark matter is instead. Because galaxies form in regions where there’s a lot of dark matter, they actually are more clustered than the underlying dark matter. ➕The galaxy bias is just how much more clustered they are. By using galaxy-galaxy lensing (see few posts ago!), we can learn about this bias, and extract useful cosmological information from the galaxy clustering signal 💥

#ThisIsDES

📸: Andreas Weith ... See MoreSee Less

Share on Facebook Share on Twitter Share on Linked In Share by Email

6 days ago

Day 29. Galaxies...it's complicated 🤯

It turns out galaxies are complicated! Galaxies aren’t just fuzzy blobs that move along dark matter filaments minding their own business. In cosmology, we usually work with such large scales that gravity is the only important force to consider.

But this assumption breaks down at galactic scales, where there are many complex interactions between the gas, stars and black holes that make up galaxies 💨💥🥘. Particularly spectacular are the powerful jets sometimes produced by the supermassive black holes at the center of galaxies, where huge amounts of gas and energy are spewed out of the galaxy at speeds close to the speed of light. ✈️✈️✈️🙀

This kind of behavior can even affect the distribution of dark matter in which the galaxy resides, which mean our gravitational lensing measurements are sensitive to it. So we need to be careful that we are modeling these effects correctly, or excluding the parts of our data that are affected.

The images today are from a study led by Vaidehi S. Paliya (ApJ 892, 133). They observed a jet which they believe resulted from the collision of two spiral galaxies; the first image is an artist’s impression, and the second is the actual data used, observed using the 8.2m Subaru telescope.

image credit: Vaidehi S. Paliya, NASA / ESA / Hubble / L. Calçada, ESO ... See MoreSee Less