Astrobites RSS

M101

Editor’s note: Astrobites is a graduate-student-run organization that digests astrophysical literature for undergraduate students. As part of the partnership between the AAS and astrobites, we occasionally repost astrobites content here at AAS Nova. We hope you enjoy this post from astrobites; the original can be viewed at astrobites.org.

Title: Untangling the Galaxy I: Local Structure and Star Formation History of the Milky Way
Authors: Marina Kounkel, Kevin Covey
First Author’s Institution: Western Washington University
Status: Accepted to AJ

Despite our home planet being embedded in it, the Milky Way and its immediate environment remain an enigma to astronomy. Once thought to have few satellite neighbors, The Milky Way has been found to have many dwarf galaxies orbiting it. New stellar streams are being uncovered as well, likely remnants of past gravitational interactions with dwarf galaxies, in which the Milky Way pulled rivers of stars from its now-dissipated partners. This burst of discoveries of new nearby and entangled structures are thanks to advancements in technology allowing astronomers to observe dimmer objects and to track stars with greater precision.

Today’s paper utilizes one of these advancements, the much lauded Gaia mission, in tandem with machine learning methods to identify new substructures within the Milky Way and, in so doing, learn about its murky past.

Re-Clustering the Star Clusters

To begin, the authors are presented with the challenge of identifying stellar structures within the enormous Gaia dataset. In order to group stars together the authors use a clustering algorithm, which is a series of steps designed to isolate populations of objects with similar characteristics; the characteristics in question here are the stars’ coordinates within the Milky Way, their parallaxes, and their proper motions. A data sample of over 19 million stars are selected from the Gaia catalog, chosen to isolate stars for which the above characteristics are measured with high certainty. After much testing of the algorithmic parameters, the model returns over 1,900 star clusters, many of which have been independently identified in other studies. However, they also identify new structures that appear to have eluded other investigations (Figure 1).

star clusters on sky map

Figure 1: Map projection of the portion of the sky considered in today’s paper, with algorithm-identified star clusters marked in blue. Yellow markings indicate star clusters previously identified using different methods. Galactic coordinates are indicated with b and l. [Kounkel & Covey 2019]

In order to learn about our galaxy’s past, the authors must gain more information about these clusters to construct a star formation history. The star formation history of a galaxy is exactly what it sounds like: a combination of all star-forming events in a galaxy’s past that contribute to the current picture seen by astronomers. However, one can’t fully understand the history by only knowing the what and the where of star formation; also important is the when.

The authors determine the ages of their identified clusters testing two separate methods: analysis by a convolutional neural network (CNN) and isochrone fitting. Training the CNN using both known real clusters and a multitude of artificial ones, they only reproduce the accepted ages of clusters in 44% of cases. Similarly, using isochrones alone is only successful in a minority of cases. Using the CNN age estimate as an input to their isochrone model, however, increases the success rate to 77%, so this methodology is used to obtain ages for the remainder of the work.

Finding Loose Strings

While investigating the distribution of their identified star clusters, the authors noticed that they tended to be distributed in long, narrow structures. These strings, as the authors call them, are about 200 parsecs in length and lie parallel to the plane of the Milky Way. They appear similar to stellar streams, but are these simply new streams, or something new entirely? The answer lies in a peculiar trend noticed by the authors: although these strings act very similarly to normal clusters in terms of their motion, they are markedly younger than the population of clusters as a whole (Figure 2).

star cluster age distribution

Figure 2: Histogram of the age distribution of the star clusters (called “groups” here) compared to the strings. Notice how the distribution of string ages appears to have lower ages. [Kounkel & Covey 2019]

Now, one might intuitively think that the strings were formed by tidal stretching, i.e., that the stars formed in a roughly spherical cloud that was then stretched out by tidal interactions with other structures. However, many of the strings don’t show any evidence of a residual core of stars, leading the authors to conclude that they just formed this way. This interpretation is supported by previous observations of molecular filaments within the Milky Way, long string-like structures of the dense, molecular gas that is so crucial to forming stars. The authors suggest that the strings formed from these very same molecular filaments.

string subsample

Figure 3: 3D plot of a subsample of the strings, where the thick lines represent the “spine” of the string and the thin lines perpendicular to the spine indicate the velocities of the stellar components of the string. Color indicates age, and a redder string is a younger string. Check out an interactive version of this plot on the Dr. Kounkel’s website. [Kounkel & Covey 2019]

Further, analysis of the global distribution of strings (Figure 3) indicates that strings of different ages seem to lie close together, coagulating into four coherent streams of structure. Due to a correlation between the position of the youngest stream and the Local Arm of the Milky Way, the authors contend that these collections of strings may correspond to past star formation in old spiral arms within the Milky Way that have become less visible after losing their star-forming gas.

If so, deeper analysis of these strings might provide a way of studying the past structure and star formation history of our home galaxy.

About the author, Caitlin Doughty:

I am a fourth-year graduate student at New Mexico State University. I use cosmological simulations to study galaxy evolution during the epoch of reionization, with a focus on metal absorption in the circumgalactic medium.

Centaurus A

Editor’s note: Astrobites is a graduate-student-run organization that digests astrophysical literature for undergraduate students. As part of the partnership between the AAS and astrobites, we occasionally repost astrobites content here at AAS Nova. We hope you enjoy this post from astrobites; the original can be viewed at astrobites.org.

Title: Positive and Negative Feedback of AGN Outflows in NGC 5728
Authors: Jaejin Shin, Jong-Hak Woo, Aeree Chung, Junhyun Baek, Kyuhyoun Cho, Daeun Kang, Hyun-Jin Bae
First Author’s Institution: Seoul National University, Republic of Korea
Status: Accepted to ApJ

One of the many mysteries of galaxy evolution is how the formation of stars is affected by a process called feedback. Unlike comments coming from a teacher on an essay, in the galactic context, feedback is coming from powerful sources of energy such as active galactic nuclei (AGN). Star formation in galaxies requires a lot of dense gas (also called the interstellar medium, or ISM), so any feedback processes that disrupt the presence or the denseness of said gas can affect the ability of a galaxy to form stars. Simulations have shown that AGN are theoretically capable of providing negative feedback by heating up the ISM or blowing it away. However, they might also provide positive feedback by compressing the ISM with their winds, making it denser and triggering bursts of star formation.

Each of these options have theoretical merit and are observed in simulations, but it can be hard to observe the effects in the wild. Today’s paper takes advantage of a particularly well-situated Seyfert 2 galaxy, NGC 5728, to enhance our understanding of AGN feedback processes. The Seyfert 2 designation is used to describe galaxies containing AGN that are similar to quasars, but that have visible host galaxies while most quasars do not.

Positive Feedback from Outflows?

The authors use observations of NGC 5728 in the optical wavelength range, the range visible to humans, to target light from stars and several emission lines generated by ionized (i.e. heated) gas. Targeting emission from a molecular transition (the carbon monoxide CO (J=2–1) transition, to be specific) also helped them trace out the distribution of molecular gas within NGC 5728, which is useful for examining how much material is available to form stars.

NGC 5728 Halpha

Figure 1: Emission map of NGC 5728 in hydrogen alpha, where the color indicates the amount of flux in a given pixel. The position of a star-forming ring and spiral arms are noted with grey dashed lines, while a biconical outflow is traced in white dashed lines. Note the black square, region A, that indicates an intersection between the AGN outflow and the star-forming ring. [Shin et al. 2019]

From these observations, the authors noted a few prominent structures. First, there are two spiral arms (only faintly visible in Figure 1). Second, there is a ring of star formation about 1 kiloparsec from the center. Lastly, there are prominent biconical outflows made up of ionized gas and full of high-energy radiation, like X-rays. Most importantly, there is an apparent intersection (labeled “A” in the figure) between the star forming ring and the northwest (i.e. the upper right in Figure 1) cone of the AGN outflow. Luckily, this intersection provides an ideal scenario for testing whether the AGN is helping or hindering star formation.

The authors define three other regions in the star forming ring (B, C, and D in Figure 2) that are located well away from the northwest biconical outflow, and can therefore serve as controls when looking for peculiarities in the star-forming characteristics of region A. Using a BPT diagram, the authors were able to calculate the percentage of flux contributed by stars in each pixel of the image and from this calculate the star formation rate in their selected regions.

AGN fraction

Figure 2: Fraction of emission in each pixel from AGN contributions (take 1 minus the AGN fraction to find the stellar contribution). A very blue color corresponds to a low AGN fraction, and thus a pixel containing gas whose hydrogen-alpha emission is dominated by stellar light. [Shin et al. 2019]

While region A has more solar masses of stars formed per year than the combined average of regions A–D, it is not particularly unusual in this respect. However, the brightness of the emission from molecular gas at region A is quite low compared to the other regions, meaning that it has an unusually high star formation efficiency (Figure 3). Indeed, it is a factor of 3–5 higher in star formation efficiency than the control regions — a significant difference! This result seems to indicate that the presence of the AGN feedback had a positive influence on the star formation, boosting it significantly and consuming a large amount of molecular gas.

star formation efficiency

Figure 3: The star formation efficiency of the galaxy, which is the ratio between the star formation rate and the available mass of molecular gas to form stars. [Adapted from Shin et al. 2019]

This seems like a pretty good indicator that an AGN should always significantly alter a galaxy’s star forming ability, right? Not quite! While this factor of 3–5 appears pretty large, when comparing to the calculated star formation rate in the entire galaxy, region A only accounts for a miniscule <10% of the total.

The jury is still out, then, on whether AGN will invariably cause galaxies to have significantly different star formation rates. This scenario is even further complicated by the fact that the biconical outflows may be simultaneously expelling accreted molecular gas from the spiral arms of the galaxy, preventing star formation from occurring there. Regardless, NGC 5728 is providing a rich test case to examine theories of AGN feedback and star formation.

About the author, Caitlin Doughty:

I am a fourth-year graduate student at New Mexico State University. I use cosmological simulations to study galaxy evolution during the epoch of reionization, with a focus on metal absorption in the circumgalactic medium.

gas-giant transit

Editor’s note: Astrobites is a graduate-student-run organization that digests astrophysical literature for undergraduate students. As part of the partnership between the AAS and astrobites, we occasionally repost astrobites content here at AAS Nova. We hope you enjoy this post from astrobites; the original can be viewed at astrobites.org.

Title: The Intrinsic Temperature and Radiative-Convective Boundary Depth in the Atmospheres of Hot Jupiters
Authors: Daniel P. Thorngren, Peter Gao, Jonathan J. Fortney
First Author’s Institution: University of California, Santa Cruz
Status: Submitted to ApJL

HAT-P-7b

Artist’s impression of HAT-P-7b, an inflated hot Jupiter. [NASA, ESA, and G. Bacon (STScI)]

Jupiter-sized gas-giant exoplanets in close orbits around their stars, commonly referred to as hot Jupiters, have been the prime targets for probing planetary atmospheres beyond our solar system. One of the many mystifying features of hot Jupiters — which, ironically, also makes them easier to detect and characterize — is their inflated radii. A good fraction of known hot Jupiters have sizes larger than those predicted by evolutionary models that take into account the properties of the system like temperature, age, and metallicity of the system. What could be causing these hot Jupiters to puff up?

A proposed mechanism to explain hot-Jupiter inflation is deposition of energy from stellar irradiation deep into the interiors of the planet. However, in addition to inflating the planet, energy from stellar flux heating up the planetary interiors can also radically alter the thermal structure (temperature variation with altitude) of its atmosphere which has direct consequences on its inferred atmospheric properties. Today’s paper attempts to draw a connection between the stellar irradiation of hot Jupiters and their intrinsic temperature, and how that ultimately affects the observations and our understanding of the atmospheres of these gas giants.

Structuring the Atmosphere of a Gas Giant

The vertical thermal structure — also referred to as the pressure–temperature profile — of a planetary atmosphere is directly related to change in the mode of heat transport (radiation or convection) within the atmosphere at different heights. You can think of this in the context of the Earth’s atmosphere: closer to the surface heat exchange occurs through convection, with hot parcels of air rising up and adiabatically expanding and cooling. This causes the temperature to steadily decrease as you go up until a certain altitude called the tropopause; above this you hit the stratosphere, where the air absorbs most of the heat from ultraviolet radiation from the Sun, causing the temperature to now increase with altitude. Even before this happens convection begins to weaken considerably and radiation takes over as the dominant mode of heat exchange. The altitude or the pressure level at which this happens is called the radiative-convective boundary (RCB; see Figure 1 for example). Such stratification of atmospheres is very commonly seen in planetary atmospheres in the solar system and has been studied extensively from measurements by probes like Galileo and Cassini-Huygens.

hot Jupiter profiles

Figure 1: Pressure-temperature profiles for hot Jupiters at different distances from a Sun-like star, and hence different equilibrium temperatures (Teq). Note that on y-axis, the pressure decreases as you go up, corresponding to going higher up in the atmosphere. The thick parts of the profiles mark the regions of the atmosphere that are convective, and you can see how the radiative–convective equilibrium boundary moves to lower pressures for hotter planets. [Thorngren et al. 2019]

Determining the height of the RCB for gas-giant atmospheres requires an understanding of the heat flux from the planetary interiors, which can be described by the planet’s effective intrinsic temperature (Tint). To give you an idea of the numbers, Jupiter with Tint ~ 100 K has an RCB around the height corresponding to the pressure of 0.2 bars (1 bar = pressure at sea level on Earth). In the case of hot Jupiters, on the other hand — which, given their proximity to the star, receive radiation of thousands of times that received by Jupiter — the atmosphere remains radiative to a much greater depth. Here the RCB can be expected to lie much deeper, at pressures of around 1 kilobar (remember pressure increases with depth). However, this is a good estimate only if you assume Tint ~ 100 K for hot Jupiters as well. As mentioned before, observed radii inflation of hot Jupiters points toward possible heating of their interiors by stellar irradiation (the strength of which is reflected by the equilibrium temperature of the planet Teq). This implies that hot Jupiters can have much higher Tint, which would push the region of convection and hence the RCB to larger altitudes (lower pressures). Since Tint and Teq both affect the height of the RCB, and Tint also depends on Teq, at what height should we expect the RCB for a hot Jupiter with a given Teq?

To answer this question, the authors calculate temperature–pressure profiles from thermal equilibrium atmospheric models of archetypal hot Jupiters with a range of Teq, and they then investigate how the height of the RCB changes with respect to different levels of stellar irradiation (see Figure 1 and 2).

Marking the Boundary

As is evident from Figure 1, the RCB moves to lower pressures (larger altitudes) with higher Teq, similar to how Tint increases with Teq. The surface gravity and metallicity of the planet also affect the RCB height, as seen in Figure 2.

RCB pressure vs Teq

Figure 2: The RCB pressure level with respect to the Teq of the planet, as calculated for different surface gravities and metallicities of the planet. Note that the RCB ends up at higher pressures for higher surface gravity and lower pressures for higher metallicity. [Thorngren et al. 2019]

The variation of RCB height and Tint with respect to Teq of the planet has several significant implications for models and observations of hot Jupiters. When the RCB lies at lower pressures (larger altitudes), this implies that more heat can now be deposited into the convective region of the atmosphere from stellar irradiation, allowing the mechanism of Ohmic dissipation to be even more efficient at inflating the planet. A higher Tint (of the order of few 100 K) would also affect predictions of day–night energy transport and atmospheric circulation predicted by global circulation models. It would also mean that phase curve observations of some hot Jupiters might be able to probe flux from this intrinsic heat of the planet. Moreover, higher Tint means that cloud condensation will occur much higher up in the atmosphere, affecting the observed emission from the day side of the planet.

With more exoplanet discoveries from TESS and exoplanet characterization opportunities from JWST on the horizon, we can hope to obtain a stronger constraint on atmospheric boundary conditions such as these, which would be important for accurate interpretations of exoplanet atmosphere observations.

About the author, Vatsal Panwar:

I am a PhD student at the Anton Pannekoek Institute for Astronomy, University of Amsterdam. I work on characterization of exoplanet atmospheres to understand the diversity and origins of planetary systems. I also enjoy yoga, Lindyhop, and pushing my culinary boundaries every weekend.

galaxy and CGM simulation

Editor’s note: Astrobites is a graduate-student-run organization that digests astrophysical literature for undergraduate students. As part of the partnership between the AAS and astrobites, we occasionally repost astrobites content here at AAS Nova. We hope you enjoy this post from astrobites; the original can be viewed at astrobites.org.

Title: The Impact of Enhanced Halo Resolution on the Simulated Circumgalactic Medium
Authors: Cameron B. Hummels, Britton D. Smith, Philip F. Hopkins, et al.
First Author’s Institution: TAPIR, California Institute of Technology
Status: Submitted to ApJ

It can be easy to think of galaxies as islands in the universe, floating around in isolation. However, a galaxy is actually surrounded by a huge sea of low-density gas that extends out to its virial radius and beyond. This gas is known as the circumgalactic medium (CGM), and more and more research is showing that the CGM has a crucial role to play in galaxy evolution. Observing the CGM has proven difficult due to its extremely low density, though, so simulations have played a large role in understanding the physics of this region. In today’s paper, the authors detail the effects of running a CGM simulation with significantly increased resolution, capable of resolving cool gas that precipitates in the CGM and rains down on the galaxy.

What Do We Know About the CGM?

Residing just outside of the galaxy, the CGM is home to large-scale flows of gas that drive galaxy evolution. These gas flows provide fuel for star formation, regulate the interactions between dark matter halos and the intergalactic medium, and contain the energy, mass, and metals of large outflows from a galaxy. In fact, the CGM is predicted to hold at least as many baryons and heavy elements as galaxies themselves, and most of the metals in the universe are found in the CGM. These metals (meaning anything heavier than hydrogen or helium in astronomy terms), deposited by galactic outflows, serve as the dominant coolant for the CGM. They are capable of radiating energy away more easily than elements like hydrogen, so an increased abundance of metals can lead to cooler gas. Consequently, this influx of metals helps to create two phases of gas: “cool” (10,000 Kelvin) gas composed of neutral hydrogen and other elements in low-energy ionization states, and “hot” (300,000–1,000,000 Kelvin) gas that contains oxygen, nitrogen, and neon in high-energy ionization states.

Unfortunately, computational work has chronically underproduced the observed abundances of these ions across redshifts by orders of magnitude. Recent work has shown that AGN feedback can increase the abundances of oxygen and other ions in the hot gas, but the discrepancy remains for hydrogen and other ions in the cool gas. In today’s paper, the authors discuss the effect of increased simulation resolution on these discrepancies.

Resolving the Resolution Issue

Perhaps one reason that simulations struggle to reproduce observations of the CGM lies in their resolution limits. Similar to how using more pixels in a television or computer screen gives a better image, increasing the resolution in a simulation means using more cells or particles to obtain a better physical picture of what is going on. However, each increase in resolution increases the computational cost of the simulation. This means your simulation that took a few days to run could instead take a few months.

Consequently, most simulations of galaxies apply their highest resolution to regions of high density where most of the matter is. This is great for figuring out what happens in the dense disk of a galaxy, but not ideal for studying the low-density CGM. Today’s paper runs simulations that force high resolution upon the CGM, reaching resolutions that are comparable to those normally obtained in the disk of the galaxy. This technique is appropriately named Enhanced Halo Resolution (EHR). Figure 1 shows the resolutions obtained by both a normal cosmological simulation and an EHR one for a region encompassing a galaxy and its surrounding filaments.

resolution plots

Figure 1: Plots of resolution for a traditional (AMR — adaptive mesh refinement) and EHR simulation. Each of these grids is made up of many cells, and spatial resolution refers to the physical length (in kiloparsecs) of the smallest cell that is present in a region. In the left panel, many galaxies are present and a particularly massive galaxy lies at the center. Its virial radius is shown by the dotted white line. Resolution in the CGM is roughly 16 times worse than in the disk of the galaxy. On the right, the EHR simulation enforces high resolution approximately to the virial radius, ensuring that interactions within the CGM are given much more computational attention. [Hummels et al. 2019]

What Does this Computational Cost Buy You?

By better resolving the gas in the CGM, the authors note that a number of physical effects present themselves. Firstly, the balance of cool and hot gas is shifted, leaving more cool gas and less hot gas than in simulations with lower resolution. The clouds of cool gas that form are also greater in number and smaller in size. Finally, the amount of neutral hydrogen and other low-energy ions found in the cool gas increases, while the abundances of oxygen, nitrogen, and neon in high-energy ionization states fall due to the decrease in hot gas. Coupled with the aforementioned work on AGN feedback, this can bring simulations closer to the observed abundances for these ions.

simulated galaxy and CGM

Figure 2: A galaxy and the CGM in an AMR simulation and an EHR one. A significant increase in HI (neutral hydrogen) can be seen in the EHR simulation. Recall that neutral hydrogen tracks the cool gas, which condenses into many clumps on the right that weren’t resolved in a traditional AMR simulation. Many of these clumps fall back into the galaxy because they no longer have enough thermal energy to resist the gravitational pull of the galaxy. [Hummels et al. 2019]

In other words, EHR causes more gas in the CGM to cool, condense into clouds, and potentially fall back into the galaxy. This is completely analogous to water vapor in our own atmosphere, which often cools, forms clouds, and rains back down to Earth. In this way, the CGM can be conceptualized as the atmosphere of a galaxy. Figure 2 shows cool gas condensing into these clouds, some of which fall into the galaxy.

Why does an increase in resolution result in more cool gas? The answer lies in how gas mixes in simulations. With lower resolution, clouds of cool gas are typically resolved only by a few cells, inducing artificial mixing between the hot and cool gas. The authors perform a test simulation demonstrating this, shown in Figure 3.

cloud test problem

Figure 3: In this test problem, a 4-kiloparsec-wide cloud of cool gas sits in a flow of hot gas for 260 million years. In the low-resolution test, the boundary of the cloud is only resolved by a few cells. This artificially thick boundary means that much of the cool gas quickly mixes with the hot gas and eliminates the HI (neutral hydrogen). In the high-resolution case, the boundary becomes much thinner, allowing the interior cool gas to survive much longer. [Hummels et al. 2019]

Resolution clearly makes a big difference in understanding the physics of the CGM and galaxies. For example, just like plants on Earth sprout after a rain, cool gas that condenses in the CGM and falls into a galaxy can trigger star formation. Understanding the ecology and geology of Earth requires a detailed picture of the atmosphere, and perhaps unlocking the mysteries of galaxy evolution may depend just as strongly on our understanding of the CGM. 

About the author, Michael Foley:

I’m a graduate student studying Astrophysics at Harvard University. My research focuses on using simulations and observations to study stellar feedback — the effects of the light and matter ejected by stars into their surroundings. I’m interested in learning how these effects can influence further star and galaxy formation and evolution. Outside of research, I’m really passionate about education, music, and free food.

cosmic distance ladder

Editor’s note: Astrobites is a graduate-student-run organization that digests astrophysical literature for undergraduate students. As part of the partnership between the AAS and astrobites, we occasionally repost astrobites content here at AAS Nova. We hope you enjoy this post from astrobites; the original can be viewed at astrobites.org.

Title: The Carnegie-Chicago Hubble Program. VIII. An Independent Determination of the Hubble Constant Based on the Tip of the Red Giant Branch
Authors: Wendy L. Freedman, Barry F. Madore, Dylan Hatt, Taylor J. Hoyt, In Sung Jang, et al.
First Author’s Institution: University of Chicago
Status: Accepted to ApJ

Author’s note: Credit for “H0tTake” goes to the conference “Tensions between the Early and the Late Universe” hosted at the Kavli Institute for Theoretical Physics!

The value of the Hubble Constant (H0) is a beast to pin down. However, it’s integral to our understanding of how the universe evolved and will continue to evolve. H0 relates the speed with which distant objects are moving away from us — due to the universe’s expansion — to how far away they are (see this Astrobite for a detailed explanation of how H0 assumed its place of importance). Measurements of H0 can be made using the early universe, from the cosmic microwave background (CMB), and the late universe, from distance measurements for stars, galaxies and other objects.

Under our current understanding of the universe, these two sorts of measurements ought to yield similar values of H0. Instead, we’ve witnessed a growing divergence between them that’s only gotten worse (or more interesting?) with time (see Figure 4, though it does come with a spoiler). Currently, early universe measurements of H0 rely on CMB observations made by the Planck satellite, while late universe measurements rely on Cepheid variables and Type Ia supernovae (Sne Ia). The discrepancy between these early and late measurements of H0 could be chalked up to new physics in the early universe that is outside our current models. But before claiming that, we’d want to rule out any hidden issues in how these measurements are being made.

On the side of the late universe, this requires using other astronomical objects to make measurements of H0  and to calibrate the distances to standard candles (objects whose brightness we understand very well), like Sne Ia. Very recently, a new measurement of H0 was announced, which used strong gravitational lens systems for distance calibration (see this Astrobite for a good summary). The paper being discussed in today’s Astrobite comes out of the Carnegie-Chicago Hubble Program, which was established to calibrate Sne Ia through alternate methods. Here, the authors use something called the Tip of the Red Giant Branch (TRGB).

The TLDR on the TRGB

color-magnitude diagram

Figure 1. A color-magnitude diagram of globular cluster Messier 55 (M55). The TRGB can be seen at the upper-right. [B.J. Mochejska, J. Kaluzny (CAMK), 1m Swope Telescope]

The TRGB consists of stars that are at a pivot point in their evolution. Red Giant Branch (RGB) stars are stars that have nearly exhausted the hydrogen in their cores. The next stage of their life is triggered when they start fusing helium in their cores instead. TRGB stars have just begun this stage of helium burning, and they can be distinguished by their characteristic redness and brightness (see Figure 1). These standard features of the TRGB make it highly suitable for measuring distances, since we know how bright it ought to appear at a certain distance.

The authors use the TRGB in lieu of Cepheids to calibrate the distances to galaxies that have hosted Sne Ia. TRGB stars have some advantages over Cepheids: they are much more common and can be found in uncrowded regions of their host galaxies, making them easier to identify. They also don’t need multiple observations to be recognized. Another useful quirk of TRGB stars is that their brightness in the I-band does not vary greatly with metallicity (the composition of the star), so the TRGBs in different galaxies shouldn’t look terribly different.

Could it (TRG)be?

In the near future, parallax measurements of Milky Way TRGB stars taken by the Gaia satellite will be available to anchor TRGB calibrations. For now, the authors use I-band observations of the Large Magellanic Cloud’s TRGB as well as parallax measurements for their analysis. The authors analyze the TRGB of 18 Sne Ia hosts, ranging from 7 to nearly 20 Mpc away, to calibrate the distances to those galaxies (see Figure 2). Their sample consists of galaxies that were not obscured by dust and had observations of their halos, where the TRGB could be cleanly measured. The TRGB calibrations were then used with a larger sample of Sne Ia to measure the distances to those Sne.

Sne Ia host galaxies

Figure 2. Nine of the eighteen Sne Ia host galaxies whose TRGB were studied. The squares represent the areas of the halo that were targeted. The hatched areas show the regions that were analyzed. [Freedman et al. 2019]

Finally, *drumroll* the authors present their measurement of the Hubble constant — 69.9 ± 0.8 ± 1.7 km s-1 Mpc-1 (the two errors are statistical and systematic respectively). This new result is shown clearly in a Hubble diagram showing their 18 TRGB calibrators and 99 Sne Ia from the Carnegie Supernova Project (see Figure 3). A Hubble diagram is a plot of distance versus speed, and the slope of the plot gives us a value of H0.

Hubble diagram

Figure 3. The Hubble diagram produced from the TRGB calibrators and Sne Ia from the Carnegie Supernova Project. The slope of the line is where the measurement of H0 comes from. The y-axis of the upper plot is the distance modulus (a measurement of distance using the relation between the absolute magnitude and apparent magnitude for an object). The y-axis of the lower plot is the difference between the points and the fit to the data. The x-axis of both plots is a quantity relating the distance modulus with redshift (see Section 7.1 of the paper). [Freedman et al. 2019]

This number falls squarely between the CMB and Cepheid-Sne Ia measurements of H0 (see Figure 4). The authors are careful to note that their result does not resolve the discrepancy in H0 values, but reiterate that additional, independent late universe measurements of H0 could change that. And the future is teeming with possibilities: aside from Gaia, the James Webb Space Telescope and LIGO and Virgo offer other avenues for measuring distances across large swaths of space, not to mention better measurements of strong lensing systems and tried-and-tested Cepheids. All in all, this is a very exciting time for cosmology!

Hubble constant over time

Figure 4. Measured values of H0 over time, showing where the TRGB measurement lands relative to the CMB and Cepheid measurements. The red star is the measurement from the paper being discussed. [Freedman et al. 2019]

About the author, Tarini Konchady:

I’m a graduate student at Texas A&M University. Currently I’m looking for Mira variables to better calibrate the distance ladder. I’m also looking for somewhere to hide my excess yarn (I’m told I may have a problem).

binary supermassive black holes

Editor’s note: Astrobites is a graduate-student-run organization that digests astrophysical literature for undergraduate students. As part of the partnership between the AAS and astrobites, we occasionally repost astrobites content here at AAS Nova. We hope you enjoy this post from astrobites; the original can be viewed at astrobites.org.

Title: Discovery of a close-separation binary quasar at the heart of a z ∼ 0.2 merging galaxy and its implications for low-frequency gravitational waves
Authors: Andy D. Goulding, Kris Pardo, et al.
First Author’s Institution: Princeton University
Status: Published in ApJL

With the announcement of the experimental confirmation of gravitational waves by LIGO in 2016 in tandem with additional electromagnetic follow-up of a neutron-star merger, astronomy was quickly ushered into an era of truly multi-messenger science. Although the number of gravitational-wave events observed by LIGO since is already substantial, the sheer number of black holes (and neutron stars) predicted to exist within our universe vastly outweighs this cumulative yield. One reason why LIGO is not constantly finding strong gravitational waves from all of these black holes (the gravitational wave background or GWB) is that not every black hole exists in a pair, which is a necessary condition to spiral inwards, merge, and set off a gravitational-wave event. Predictions suggest that the timescales required for some of these events to occur are a sizable fraction of the age of our universe! 

LIGO has, however, shown direct evidence for the merging of solar-mass black holes with atypically large masses between 10–40 M☉, and although easier to observe due to their strong gravitational signal, their existence has continued to challenge theoretical explanations. Scaling up to even greater masses does not reduce the pressure on theory either, as the predicted dominant mass contribution to the yet undetected GWB is from supermassive black holes (SMBHs) on the order of 108–109 M☉.

SDSS J1010+1413

Figure 1: Hubble Space Telescope images of SDSS J1010+1413, showing wide-field galaxy morphology and a zoom-in view of the central SMBH pair with the F621M, F689M, [OIII], and F160W bands. The galaxy shows evidence of a past merger: a disturbed shape and stripped gas streams. The [OIII] extent is seen to be coincident with the continuum F689M light. [Goulding et al. 2019]

It has long been hypothesized that SMBHs inhabit the central-most regions of almost all galaxies, and they accumulate mass through the slow accretion of gas and stellar material. When galaxies undergo the often violent processes of a wholesale merger, these SMBHs are predicted to collect within the central region of a galaxy and become gravitationally bound on short Myr timescales, accelerated by dynamical friction. However, the black-hole pair can only bleed off so much energy through interacting with nearby material, and at some point within the final parsec, the merger is predicted to stall out. This so-called “final parsec problem” has yet to be resolved. For the sample of intermediate-mass-black-hole mergers suggested by the LIGO observations, it appears that nature has ways around this problem. Whether this is true for SMBH mergers has yet to be seen.

Putting the final parsec problem aside, the authors of today’s astrobite provide definitive evidence for a precursor system that may one day produce a low-frequency gravitational-wave event consistent with a strong contribution to the GWB.

As shown in Figure 1, observations with the WFC3 instrument aboard the Hubble Space Telescope revealed a pair of tightly bound SMBH candidates in a highly luminous post-merger galaxy, poetically named “SDSS J1010+1413”. Accounting for the cosmological distance, the separation between the SMBHs is found to be ~430 pc (1,400 light-years). Previous studies examining the velocities and dynamics of the galaxy confirm the outward telltale signs of a trainwreck galaxy and provide the context for the apparently resolved pair of SMBHs in its core. Special imaging with an appropriate [OIII] narrowband filter corroborates this picture by defining the extent of the extremely luminous [OIII] emission — characteristic ionized gas associated with powerfully accreting SMBHs (AKA quasars). However, coincident X-ray observations with the Chandra Space Telescope showed very little X-ray light, a fact that, when compared to infrared estimates, suggests an obscuring cloud of thick gas along the line of sight.

Despite the awesome resolution of the Hubble Space Telescope (~0.04”), such an observation of a supposed SMBH pair may be ambiguous. To increase confidence in this interpretation, the authors modeled each of the sources with a coincident point-like model and an extended Gaussian model. Even so, a single SMBH with extended [OIII] and bisecting obscuration due to a dust lane could mimic this scenario, albeit with a significantly worse fit. Given the former scenario, each SMBH is estimated to have a minimum mass of 4 x 108 M☉ based on the Eddington luminosity limit, putting them in the sweet spot of GWB contribution.

merger stages

Figure 2: Merger stages with timescales shown. From the right, dynamical friction accelerates the first stage of coalescence, followed by stellar hardening. Gas infall may help prevent stall-out and overcome the “last parsec problem”. Then, in the final stages of coalescence at < 1 pc, Pulsar Timing Arrays should be sensitive to the predicted nanohertz gravitational-wave signal prior to the merger. [Goulding et al. 2019]

Lastly, the authors make tentative predictions for a future low-frequency high-mass gravitational-wave event, as shown in Figure 2. By considering carefully motivated arguments for dynamical friction, they surmise a coalescence timescale of 0.1–2 Gyr, where the lower limit is argued from the seemingly large gas reservoir near the SMBHs, which may help dissipate energy and rapidly close their orbital separation. Ignoring the “final parsec problem”, they argue that once the system reaches < 0.1 pc separation, the gravitational wave emission will enable the pair to finally merge within ~700 Myr. Given that the lookback time to this galaxy at z ~ 0.2 is on the order of the predicted merger timescale, this discovery provides strong evidence that such galaxies could be contributing to the GWB right now.

The low-frequency nanohertz GWB signal will not be detectable by current observatories such as LIGO. However, these increasingly powerful facilities will soon be complemented by hyper-sensitive Pulsar Timing Arrays which should be able to detect a nanohertz GWB signal, as may be produced by such a pair of quasars in this and other trainwreck galaxies.

About the author, John Weaver:

I am a PhD student at the Cosmic Dawn Center at the University of Copenhagen, where I study the formation and evolution of galaxies across cosmic time with incredibly deep observations in the optical and infrared. I got my start at a little planetarium, and I’ve been doing lots of public outreach and citizen science ever since.

SOHO image of solar chromosphere

Editor’s note: Astrobites is a graduate-student-run organization that digests astrophysical literature for undergraduate students. As part of the partnership between the AAS and astrobites, we occasionally repost astrobites content here at AAS Nova. We hope you enjoy this post from astrobites; the original can be viewed at astrobites.org.

Title: Chromospheric Cannonballs on the Sun
Authors: Shuhong Yang, Jun Zhang, et al.
First Author’s Institution: Chinese Academy of Sciences, China
Status: Accepted to ApJL

We’ve all been there. You’re enjoying a nice float in the pool on a hot summer’s day when suddenly you hear that dreaded word… “CANNONBALL!” The next thing you know, your uncle soars overhead and you brace yourself for the ensuing tsunami. But you might not have known that as your uncle was cannonballing, an analogous process may also have been happening on the surface of the Sun. Today’s paper reports the discovery of a phenomenon in the Sun’s atmosphere that the authors name “cannonballs” due to their circular appearance and arc-like trajectory, much like your uncle and his pooltime performance.

Our Dynamic Sun

the Sun

Figure 1: The layers of the Sun. Click to enlarge. [NSF]

Our star is an active place, despite being relatively calm in comparison to many other stars. Similarly to Earth, the Sun has an atmosphere with several layers and complex processes occurring within each one of them (Figure 1). Today’s authors specifically consider the chromosphere, the middle layer of the Sun’s three main atmospheric layers.

Situated between the cooler photosphere below and the scorchingly hot corona above, the chromosphere is responsible for transferring heat between these two layers. As a consequence, the chromosphere’s temperature increases the closer it gets to the corona. The heat transfer occurs through a number of processes, which result in a myriad of dynamical features with exceedingly pleasing names (see Figure 2): 

  • Spicules (Figure 2A): fast-moving, short-lived (~15 minutes) jets of hot plasma that can shoot tens of thousands of kilometers up into the chromosphere before collapsing and disappearing. They are typically associated with areas of high magnetic flux.
  • Surges (Figure 2B): small scale, short-lived (~2–10 minutes), upside-down-Y-shaped jets of plasma created when small magnetic field lines touch and connect. Also called chromospheric anemone jets.
  • Ellerman bombs (Figure 2C): tiny, short-lived (~5 minutes) solar flares that occur near the edges of sunspots where the magnetic field is breaking through the photosphere. Also known as Severny moustaches.

Today’s paper adds another pleasing name to the list of chromospheric phenomena in the form of cannonballs.

spicules, surges, bombs

Figure 2: From left to right: A) spicules, B) surges, and C) an Ellerman bomb. [NASA; Nishizuka et al 2011; David Darling]

How does one go about finding cannonballs in the Sun’s atmosphere in the first place? Yang et al. studied images of the Sun taken by both the New Vacuum Solar Telescope (NVST) in China and NASA’s Solar Dynamics Observatory (SDO). Specifically, they looked at sequences of images that used an H-alpha filter, a deep-red filter that measures the light emitted when the electrons in hydrogen atoms fall from the third-lowest to the second-lowest energy level. In one sequence spanning about ten minutes, the authors noticed a round, dark structure moving along a curved trajectory: cannonball! (Figure 3). They further identified similarly moving structures in two other image sequences that appeared bright as opposed to dark.

cannonball trajectory

Figure 3: Image sequence using NVST that shows the movement of a cannonball on the Sun on 28 October 2017. [Yang et al. 2019]

From the images, the authors calculated a variety of properties of the cannonball structures. They found that the three cannonballs traveled at an average speed of 55.9 km/s, which is roughly five times Earth’s escape velocity, or almost two times its orbital speed around the Sun. They also found that the structures encompassed an average volume of 1.53 billion cubic kilometers, roughly the volume of all of Earth’s oceans or 600 trillion swimming pools. Assuming the density of the cannonball is the same as the chromosphere, Yang et al. then calculated an average cannonball mass of almost 170,000 US tons—about 25,000 elephants, or roughly 1.3 million uncles.

So what are these structures? Cannonballs on the Sun are a bit more complicated than your uncle jumping into the pool or an actual ball shot from a cannon. To gather more information, Yang et al. used observations from SDO at ultraviolet (UV) and extreme ultraviolet (EUV) wavelengths, together with measurements of the solar magnetic field, that were simultaneous with the H-alpha images from NVST. The extra measurements revealed evolution in the magnetic field, as well as heightened emission in both the UV and EUV measurements near the locations of the cannonballs. This suggested that the solar magnetic field was intimately involved in the creation of these structures.

The authors propose that a solar cannonball forms when magnetic reconnection occurs within the chromosphere (Figure 4). In this process, small-scale magnetic field loops emerge and detach from stronger large-scale loops. The small loops rise up toward the large loops and reconnect, flinging chromospheric plasma along the large loops. Magnetic reconnection results in the conversion of magnetic energy into kinetic, potential, and thermal energy; this explains why the cannonballs move so quickly, and it also provides an efficient heating mechanism for the chromosphere.

cannonball formation schematic

Figure 4: Schematic of cannonball formation. A small-scale magnetic loop emerges and recombines with a large-scale loop, converting magnetic energy into kinetic, potential, and thermal energy and flinging a cannonball away in the process. [Yang et al. 2019]

As with any new discovery, more questions are raised than are answered in today’s paper. Why are the cannonballs shaped like blobs rather than jets? Why are some cannonballs dark while others are bright? Future observations will answer these questions and almost certainly raise several more. In the meantime, we can enjoy the Sun’s glow from our pool here on Earth, hopefully cannonball-free. ☀️

About the author, Stephanie Hamilton:

Stephanie is a physics graduate student and NSF graduate fellow at the University of Michigan. For her research, she studies the orbits of the small bodies beyond Neptune in order learn more about our solar system’s formation and evolution. As an additional perk, she gets to discover many more of these small bodies using a fancy new camera developed by the Dark Energy Survey Collaboration. When she gets a spare minute in the midst of hectic grad school life, she likes to read sci-fi books, binge TV shows, write about her travels or new science results, or force her cat to cuddle with her.

gas-giant formation

Editor’s note: Astrobites is a graduate-student-run organization that digests astrophysical literature for undergraduate students. As part of the partnership between the AAS and astrobites, we occasionally repost astrobites content here at AAS Nova. We hope you enjoy this post from astrobites; the original can be viewed at astrobites.org.

Title: The hot Jupiter period-mass distribution as a signature of in situ formation
Authors: Elizabeth Bailey, Konstantin Batygin
First Author’s Institution: California Institute of Technology
Status: Published in ApJL

To fully understand how and where planets can form, astronomers must look to the extremes. One of the most exotic discoveries in exoplanet research has been of a class of planets known as hot Jupiters. These are gaseous worlds, hundreds of times the mass of the Earth, that orbit their host stars in mere days. Given the major role that Jupiter had in shaping our solar system, it is crucial to understand how gas-giant planets form in a variety of environments.

How to Build a Jupiter

The formation of a Jupiter-sized world is thought to be a two-step process. First, material in the protoplanetary disk conglomerates to form a solid core. If this core grows larger than about 10x the mass of the Earth, its gravitational pull becomes strong enough for the planet to accumulate a gaseous envelope. As this envelope grows, the gravitational pull gets stronger, allowing the planet to attain a huge mass fairly quickly. Eventually, the gaseous envelope becomes too hot for material to continue to condense and the growth is throttled.

For intermediate-sized worlds, radiation from the star can blast away the atmosphere if the planet is too close. This results in a dearth of close-in planets around 1/10 the mass of Jupiter. For larger worlds, however, this evaporation is ineffective. Even very highly irradiated Jupiter-sized planets only ever lose about 1% of their mass. There appears to be a very sharp cutoff, below which hot Jupiters that are too small and close to their host stars simply don’t exist. The authors of today’s paper explain this cutoff with a wonderfully simple and succinct model and use it to argue that most hot Jupiters formed at their current location, rather than having been built further out and subsequently migrating inwards.

stellar magnetic field

Figure 1: A diagram showing the structure of a star’s magnetic field (thin black lines) alongside a protoplanetary disk (thick black lines). Close to the star, the magnetic field is strong enough to disrupt the protoplanetary disk, preventing planet formation within a distance known as the “magnetic truncation radius”. [Camenzind 1990]

It turns out that there is a limit on how close to a star planets can form. Young stars have strong magnetic fields that interact with the surrounding protoplanetary disk. As the disk loses angular momentum due to its inherent viscosity, material continually falls inward onto the star. Close to the star, the magnetic field can be strong enough to force material up out of the disk and along the field lines. The distance at which this occurs is known as the magnetic truncation radius (shown in Figure 1). Interior to the truncation radius, the protoplanetary disk becomes too disrupted for planet formation to occur. If the protoplanetary disk material is vigorously falling towards the star, the disk can work its way far inward before being torn apart by the magnetic forces.

An Inner Limit for Gas Accretion

Next, the authors use this battle between the disruptive magnetic field of the star and the inwardly streaming protoplanetary disk material to explain the observed lack of close-in, less massive hot Jupiters. They make the assumption that the final mass of a hot Jupiter is set by how quickly the protoplanetary disk material is streaming inwards, or accreting. Because this also implies that the magnetic truncation radius is smaller, one should expect larger hot Jupiters to lie slightly closer to the star. This is all, of course, assuming that these worlds formed in place, rather than being constructed further from the star and then migrating inwards.

exoplanet orbital distance v. mass

Figure 2: Orbital distance vs mass for all known exoplanets. Planets fall into three distinct groups: hot Jupiters (top left), cold Jupiters (top right) and sub-Jovian worlds (bottom center). For the hot Jupiter population, there is an absence of planets below and to the left of the solid black line, which the authors argue is set by the magnetic truncation radius. [Bailey & Batygin 2018]

Figure 2 shows the distribution of known exoplanets as a function of semi-major axis (distance from the host star) and mass. The hot Jupiters are the cluster of points towards the top left of the diagram. The straight black line shows the predicted cutoff due to the magnetic truncation radius. The vast majority of hot Jupiters lie above and to the right of this line. The authors argue that the sharp cutoff is evidence that worlds are being constructed in place right up to the magnetic truncation boundary. Had these bodies formed elsewhere in the disk and moved around, the distribution would not follow this cutoff so closely.

What About Tides?

Above about 1 Jupiter mass, there are a handful of planets that do not seem to follow the cutoff denoted by the solid line. The authors explain this discrepancy as a result of tidal evolution. If a planet is massive enough and close enough to the star, its gravitational pull will distort the star slightly, similar to the way that the Moon invokes tides on the Earth. Strong tidal interactions between a star and a nearby planet can actually remove a significant amount of orbital energy. The result of this is that the planet’s orbit will shrink, possibly below the cutoff described in the previous paragraph. This should result in planets being found right up to the curved black line shown in Figure 2, below which there are indeed no observed hot Jupiters.

All of the features described in Figure 2 are consistent with the idea that the final mass and position of most hot Jupiters are set by the availability of planet-forming material at the inner edge of the disk. This is a strong indication that the gaseous envelopes of these worlds, which make up most of their mass, were constructed at or near their present locations. With that being said, it is not clear where and how the cores that seeded the gas accretion formed.

Finally, it is worth noting that there exists a small but significant population of hot Jupiters that have highly eccentric orbits. These worlds most certainly formed further out and lost orbital angular momentum to a companion planet and do not fit into the framework described here. The fact that the majority of known hot Jupiters lie above the cutoff described by the model in this paper suggests that most hot Jupiters do not undergo orbital migration. This is an important clue on the path to understanding why many exoplanetary systems appear so vastly different than our own solar system.

About the author, Spencer Wallace:

I’m a member of the UW Astronomy N-body shop working with Thomas Quinn to study simulations of planet formation. In particular, I’m interested in how this process plays out around M stars, which put out huge amounts of radiation during the pre main-sequence phase and are known to host extremely short-period planets. When I’m not thinking about planet formation, I’m an avid hiker/backpacker and play bass for the band Night Lunch.

Hubble extreme deep field

Editor’s note: Astrobites is a graduate-student-run organization that digests astrophysical literature for undergraduate students. As part of the partnership between the AAS and astrobites, we occasionally repost astrobites content here at AAS Nova. We hope you enjoy this post from astrobites; the original can be viewed at astrobites.org.

Title: Morpheus: A Deep Learning Framework For Pixel-Level Analysis of Astronomical Image Data
Authors: Ryan Hausen & Brant Robertson
First Author’s Institution: UC Santa Cruz
Status: Submitted to ApJS

Dreaming of a Better Way to Classify Galaxies

In Greek mythology, Morpheus is the God of Dreams, who shaped and formed the dreams of mortals. It is fitting, then, that Morpheus is now dabbling in classifying galaxies based on their shape, to help us mortals with our astronomy. Born of Tensorflow and Python 3, the 21st-century Morpheus is a new neural network dreamed up by the authors of today’s paper to perform galaxy classification.

Stephan's Quintet

Hubble’s view of Stephan’s Quintet, a group of five galaxies with a variety of morphologies. [NASA/ESA/Hubble SM4 ERO Team]

The shape, or morphology, of galaxies is critical to understanding their formation and evolution. As it is such an important characteristic, astronomers must have found a robust algorithm or quantitative model that determines morphology, right? Not quite — it turns out that the most accurate way to classify galaxies morphologically is to round up a pack of trained astronomers and have them look through pictures of galaxies by eye.

Unfortunately, galaxies far outnumber astronomers. The most well-known method that addresses this challenge is Galaxy Zoo, which enlists interested internet users to classify galaxies. While very successful, this approach is still limited by accuracy and scalability. To address these issues, researchers have begun to use machine-learning techniques to push morphological classification forward.

Today’s paper introduces Morpheus, a new deep-learning network to classify astronomical images. The network determines the morphological type of each pixel in an astronomical image, an approach that increases its capabilities beyond existing methods.

How to Train Your Neural Net

Neural networks like Morpheus work by learning how inputs, often images, are associated with their desired outputs, often called labels. For example, you could train a network by feeding it images of cats and dogs, labeled with the appropriate word “cat” or “dog.” Then, when you input new images of furry friends it hasn’t seen before, it should be able to assign each the appropriate label. Check out this astrobite for a great explanation.

In the case of Morpheus, the inputs are images of galaxies through multiple color filters. (This is already an improvement over previous methods, which use composite images.) The labels are the morphologies of the galaxies: disk, spheroid, and irregular, as well as point source/compact to account for unresolved sources.

The authors trained Morpheus on images of 7,629 galaxies in the CANDELS survey, in the GOODS South region. To label these training images, we still need that pack of trained astronomers: multiple experts voted on the classification of each galaxy. Morpheus goes beyond previous works by using not just the winning classification, but all of the expert votes as labels. This allows the network to learn the uncertainties in morphology, for example knowing when a certain source looks similar to both disks and spheroids. Further, it learns which pixels in the images are most relevant to the experts’ votes.

Morpheus then outputs a “classification image,” which labels each pixel with the probability that it corresponds to each classification. This allows for not only the classification of objects, but also spatially resolved morphological information and source detection.

example field classified by Morpheus

Figure 1: An example field classified by Morpheus. Left panel: A composite image of the input data. Middle panels: The dominant classification of each pixel. Right panel: The output Morpheus “classification image” color-coded by dominant morphology. The brightness of the color indicates the dominance of the most dominant morphology of each pixel, with white meaning indeterminate classification. [Adapted from Hausen & Robertson 2019]

Figure 1 shows a field region classified by Morpheus. The left panel shows a composite image of the input data, with many galaxies and other objects visible. The four panels to the right show the dominant label of each pixel for the types: spheroid (red), disk (blue), irregular (green), and point source/compact (yellow). The Morpheus classification image on the right again shows the dominant morphology of each pixel, now with the brightness corresponding to the difference between the dominant class and the second-most dominant class, so that white pixels mean similar results for multiple classes. The brightest objects in the image are well-classified into their visually apparent galaxy morphologies, while the fainter objects are mostly classified as point sources.

pixel-level classification of the GOODS South region.

Figure 2: Morpheus’s pixel-level classification of the GOODS South region. The colors correspond to the dominant classification of each pixel, with white meaning comparable classifications for the pixel. [Hausen & Robertson 2019]

Morpheus classifies the entire GOODS South field in this way. Figure 2 shows the result, with the colors again corresponding to the dominant type, with more certain classifications in brighter colors. To see Morpheus at work, check out the mesmerizing video below.

Evaluating Galaxy Classification: Morpheus vs. Astronomers

If Morpheus is classifying pixels and the astronomers classified objects, how can we compare the two to measure Morpheus’s performance? The authors do this by computing the brightness-weighted average of the pixels in the object and selecting the dominant classification. But we still expect some uncertainty in the classification, because for many sources even the “truth” (astronomer-determined labels) was unclear. As Morpheus was trained not just on the majority-voted classification but on all of the votes, Morpheus’s assignments should match the distribution of astronomer votes. This can be evaluated by looking at the confusion matrix, shown in Figure 3.

confusion matrices

Figure 3: Confusion matrices that show the distribution of morphology classifications. The left matrix shows the degeneracies in visual assignment by astronomers, and the right matrix shows Morpheus’s replication of those degeneracies in its assignments. [Hausen & Robertson 2019]

The matrix on the left shows the natural degeneracies in astronomer-classified objects, meaning how often astronomers confused two types of galaxies for each other. For example, for objects that the majority (80%) of astronomers agreed were disks (the “K15 Dominant Classification” axis), the remaining astronomers classified as spheroids 9% of the time and irregulars 11% of the time (the “Classification Distribution” axis). The matrix on the right shows the Morpheus vs. astronomer degeneracies. Continuing the above example, for objects that the majority of astronomers labeled as disks, Morpheus agreed for 76% of the objects but thought that 8% were spheroids and 16% were irregulars, close to the astronomer distribution. The two matrices clearly agree quite well overall, showing that Morpheus succeeds at reproducing the intrinsic uncertainty (represented by astronomer disagreement) in the object classifications.

The authors use many other metrics to evaluate how Morpheus performs, including inserting simulated sources to test for false negatives and completeness. These couldn’t all fit in an astrobite, so check out the paper to learn more!

The authors anticipate that Morpheus will be useful for upcoming large-scale imaging surveys, and can also be expanded to learn other information like galaxy redshift. Keep an eye open for what the Morpheus team will dream up next.

About the author, Kate Storey-Fisher:

Kate is a PhD student in the Center for Cosmology and Particle Physics at New York University. She studies the large-scale structure of the universe using cosmological simulations and galaxy surveys. She is still waiting for the galaxies to respond to the SurveyMonkey she beamed to them.

LMC

Editor’s note: Astrobites is a graduate-student-run organization that digests astrophysical literature for undergraduate students. As part of the partnership between the AAS and astrobites, we occasionally repost astrobites content here at AAS Nova. We hope you enjoy this post from astrobites; the original will be viewable at astrobites.org once the site has been fully restored.

Title: Large Magellanic Cloud Cepheid Standards Provide a 1% Foundation for the Determination of the Hubble Constant and Stronger Evidence for Physics Beyond ΛCDM
Authors: Adam G. Riess, Stefano Casertano, Wenlong Yuan, Lucas M. Macri, Dan Scolnic
First Author’s Institution: Space Telescope Science Institute and Johns Hopkins University
Status: Published in ApJ

Hubble’s law tells us that all galaxies, stars and planets are moving away from each other, and the more distant the object, the faster it is moving away. We quantify this expansion as a speed per distance, which gives us a unit like km/s (speed) per megaparsec (distance). This value is known as the Hubble constant, or H0.

The Hubble constant has been determined using various methods. However, two of these titan measurements disagree with each other in a way that astronomers deem significant.

The first of the measurements comes from studying the oldest electromagnetic radiation in the universe — the cosmic microwave background (CMB). See this Astrobite for a detailed explanation of how we are able to do this. The most recent results from the CMB give us a Hubble constant of roughly 67 km/s/Mpc.

The second measurement comes from using Type Ia supernovae as standard candles to calibrate distances to them (see this Astrobite for more). Essentially, by looking at these stars at various distances, we can correlate their distance with their apparent brightness. By assuming supernovae are dimmer proportional to their distance from us, we can measure the gradient of this correlation. Recent results put H0 at 73 km/s/Mpc.

So, one of the most prominent problems in cosmology boils down to a 6 km/s/Mpc difference. Certainly, each of these measurements have their own subtleties but there are two main things to note:

  • The Hubble-constant measurements using the CMB and Type Ia supernovae are independent. They do not rely on the same measurement technique, and therefore do not have any source of error in common. This makes it harder to dismiss the tension as something which comes from a shared, inaccurate measurement.
  • The Hubble-constant measurement from the CMB uses data from the early universe, while the value obtained from supernovae is a late-time or local measurement. This could potentially be an interesting explanation for the tension.

A New Addition

Today’s authors stir the Hubble cauldron a bit more with 70 space-based observations of Cepheid variables in the Large Magellanic Cloud (LMC) from the Hubble Space Telescope.

A Cepheid variable is a type of star that pulsates over some period of time. Astronomer Henrietta Swan Leavitt deduced that the rate of pulsation for these stars is correlated strongly with their luminosity (see this Astrobite for more on her work and legacy). Therefore, one can know the brightness of these stars simply by observing their pulsation rate (Figure 1). Consequently, one can determine the distance to these stars just by comparing their known luminosity to the apparent brightness. Much like supernovae, this makes Cepheid variables powerful probes of the local Hubble constant. Furthermore, by studying galaxies containing both Cepheid variables and type Ia supernovae, the Cepheid-derived distances can be used to calibrate the accuracy of supernovae-derived distances, creating a robust distance ladder, which gets us to H0.

Period-luminosity relation

Figure 1: Period-luminosity relation for the 70 Cepheid variable stars. The colours in the figure indicate the different wavelengths used for observing these Cepheids. The agreement in the slope tells us the P–L relation is not dependent on any particular wavelength. [Riess et al. 2019]

To ensure an accurate Hubble-constant measurement with Cepheid variables, various sources of uncertainty are considered by the authors. Among these are the differences in the telescope sensitivity to fainter, distant Cepheids compared to nearer ones, which can affect the measured brightness. Another source of error is the inclination of the LMC itself, which results in some Cepheids appearing closer or farther than average by a very small degree. After taking all sources into account, the total uncertainty in the distance measurement, and hence the Hubble constant, is 1.28%, which is the smallest error for any Cepheid-variable Hubble-constant measurement to date.

So What’s the Tension Now?

Combining the LMC distances with two other distance calibrators for better constraints, the authors quote a Hubble constant of 74.03 km/s/Mpc, which is in a staggering 4.4-σ tension with the CMB Hubble-constant measurement. This effectively means that the probability that the new measurement is genuine rather than a statistical fluke is above 99.999%, and therefore so is the discrepancy.

Hubble constant

Figure 2: Various measurements of the Hubble constant colour-coded by whether they use data from the early universe (blue) or the late universe (red). At the top are potential modifications to our current cosmological model which could resolve the current tension. [Riess et al. 2019]

Much has been said on the nature of the Hubble disagreement already, both on its nature and from pacifists looking to ease the tension (see examples here and here). More recently, gravitational waves have burst onto the scene with another independent measurement (though it is not statistically significant enough to fuel the flames just yet). New physics could hold the key to breaking this Hubble stalemate. For example, our universe could have a non-zero curvature, a time-dependent dark energy, or interacting dark matter. Today’s paper shows that the tension is as strong as ever, so we wait for more precise, independent measurements to help clarify the nature of our expanding universe.

About the author, Sunayana Bhargava:

I’m a third year PhD student in the Astronomy Centre at the University of Sussex. I study galaxy clusters with X-ray and optical data to learn about cosmology and the properties of dark matter.

1 … 29 30 31 32 33 … 46