Thursday, May 30, 2013

Taking the Python Plunge

I'm finally doing it. I'm learning Python through Codecademy's online tutorials.

So far, I really like Python. Everything is natively object oriented, so every variable is born as an object. Working within the paradigm has already shown me how much time I spend wrestling with IDL to make it work more like an object-oriented programming language, when it simply isn't. I've been known to call my style within IDL as Pseudo Object Oriented Programming, or POOP. It's often just as ugly as it sounds. So, philosophically, I'm finding that Python is well aligned with my programming style and needs.

Big ups to my student, Tim, for helping me with some of the nuances of Python arrays and syntax. He also pointed me to the Enthought Canopy programming environment, which is a great way to write code. Think of it as IDL's Development Environment, but less clunky.

I'm planning to start a new research project built solely on Python. It's a bit daunting, but I figure the only way to learn is to dive right on in. If you have any encouragement to lend, I could use it!

BTW, the Codecademy tutorials are a lot of fun, too. Here's a little easter egg that one of their "PSAs" alerted me to:

In [1]: import this
The Zen of Python, by Tim Peters

Beautiful is better than ugly.
Explicit is better than implicit.
Simple is better than complex.
Complex is better than complicated.
Flat is better than nested.
Sparse is better than dense.
Readability counts.
Special cases aren't special enough to break the rules.
Although practicality beats purity.
Errors should never pass silently.
Unless explicitly silenced.
In the face of ambiguity, refuse the temptation to guess.
There should be one-- and preferably only one --obvious way to do it.
Although that way may not be obvious at first unless you're Dutch.
Now is better than never.
Although never is often better than *right* now.
If the implementation is hard to explain, it's a bad idea.
If the implementation is easy to explain, it may be a good idea.
Namespaces are one honking great idea -- let's do more of those!

Wednesday, May 29, 2013

Minerva update

Minerva Telescope 1, by day.
And by night. My astro-instrumentation student, Mike, sorting out the rat's nest of cables feeding our Apogee camera, autoguider and dual filter wheels.
Last night we tried to get the autoguider working, but were thwarted by a loose USB power connector (we, meaning Mike working with me looking over his shoulder :). Fortunately, Mike stuck with it and fixed the problem earlier today. While Mike was working on the cables, I observed Saturn through the eyepiece on the other Naysmyth instrument port. Here's a picture of Saturn I snapped by putting my iPhone camera up against the eyepiece.

Trust me, this is not nearly as impressive as seeing it with your eye! Even from Pasadena, with 3.5 arcsecond seeing, I was able to clearly make out Titan and three other, smaller moons. 

Tonight we'll continue working on the autoguider and perhaps start testing the photometric precision of our imaging system. 

Thursday, May 23, 2013

Friday Afternoon Music Break

A little stellar astrophysical hip hop by Kilo Kish:

Wednesday, May 22, 2013

Chinese Proverb Apropos to Time Domain Science

Via Jay Anderson (STScI): "The best time to plant a tree is twenty years ago. The second best time is now."

This is extremely valuable advice for anyone doing astrometry, Doppler-based planet searches, or any other science that gets better with longer time baselines. When's a good time to start your monitoring? 20 years ago? The next best time to get your survey up and running? Right now!

Tuesday, May 21, 2013

Liveblogging LathamFest Part 2

We're all caffinated now!

Stéphane Udry and The Role of Radial Velocities in the Quest for New Earths

Stability is key. Many aspects to stability! Two methods of precise RVs: absorption cell and simultaneous emission lamp. These calibrate the wavelength scale and instrumental profile.

Using a Fabry-Perot with HARPS shows that there are irregular pixel spacing on the CCD, which may be responsible for various 1-year signals.

Controlling the instrumental profile requires fiber optic cables, which scramble the input light. Octogonal fibers are much better than circulular (cross-section) fibers. Big success at SOPHIE.

HARPS-N is the latest PRV spectrometer in La Palma, Canarie Islands. David Phillips (CfA) installed a laser frequency comb on HARPS-N, tuned to work in 500-600nm bandpass (author note: previous LFCs only work in NIR). Difficult to faint stars with HARPS-N, so portion of 80-night GTO time is used for high cadence on bright stars. Shows RV time series of stable stars (author note: this is a first for the Geneva team, to my knowledge)!! rms = 1.2 m/s for two stars. "Most of this comes from stellar noise and undetected planets."

Future: ESPRESSO on VLT. Ultra-stable, R=120,000, laser-comb calibrator, dual-arm (red and blue side). 100 nights of GTO time already allocated (whoa!). Looks like a mega-HARPS.

From technical aspect to stellar stability. Stars are not really stable even if our instruments are. Asteroseismology campaigns give us expectation for amplitude and frequencies of oscillation noise. Both modes and granulation. Bin RVs over night or nights to beat down this noise. 10 cm/s can be reached for binning over several days. Simulations of effects of spots for different levels of stellar activity. 10 cm/s reached after binning over 10s of days . Longer-period planets, you can use larger bins, and simple binning can allow RV searches to reach down to 2 Mearth in ~200 day orbit. Shows cool simulation results.

Binning also works with real data and matches simulations. (author note: Very cool!)

Also need activity indicators since RV variability tracks activity cycle of the star (think butterfly diagram of the Sun). 30% of HARPS high-recision sample show activity-correlated RV signals.

Also need to watch out for aliasing (Dawson & Fabrycky 2010).

All of these lessons used for alpha Cen Bb (Dumusque et al. 2012).

Looking ahead: CHEOPs, which is a 33 cm space telescope dedicated to photometric transit followup. Launches in 2017, same time as TESS!

Enjoyed working with Dave over the years. Lots of pictures of Dave with wine. Beautiful merger of fine wine and fine science. Share one sentence to share with audience, taken from letter of reference for Stephane: "...he will always bet welcome in my office, my house and my cellar."

Suzanne Aigrian Combining Multiple Activity Diagnostics to Improve Radial Velocity Precision

Dave says he wants masses...we will get masses. Or, at least, try very hard!

Effects of spots on stars is similar to the Rossiter effect. Spots also surpress convection. Normally, hot material rises, cool material falls due to convection, averages to convective blue-shifted line profile. Spots change this averaging (Meunier et al. 2010).

Sunspots (big black smudges the size of the Earth (Earth image added for effect). Granulation from convection are smaller inhomogeneities around spots. Bright, hot rising material is yellowish, while cooler falling material is orangish. Spots surpress rising material and suppresses convective blueshift
"Note that I'm not talking about pulsations." Pulsation is short-time-scale and mitigated well by binning.

For activity-induced RV noise, use:

  • Bisector span
  • Ca H&K index
  • Sine-fitting at harmonics of the rotation period
  • Simultaneous light curves

There is a clear link between these effects and RV, but no clear, predictable correlation.

Aigrain, Pont & Zucker (2012) for FF' method. Spot-modulated RV variation looks like (and is, mathematically  the first derivative of the photometric modulation. So F = flux, F' = flux derivative. Used the example of the Sun's disk-integrated light curve, and (a form of) the sun's disk-integrated RV measurement. Simulated RVs match observations to 1 m/s. See the development of a spot region on the Sun. Currently testing using Sun RVs from sunlight reflected off of asteroid Vesta.

Working with Geoff Marcy to do the same types of tests using Kepler light curves and contemporaneous Keck/HIRES RVs. Problem is that KOIs have planets that cause RV variations! Fit expected activity RVs and planet RVs (know planet periods and phases, but not amplitudes). In case of KOI-82, the planet's dominate the RV variability. Would love to do this for more planets.

Active stars are not all bad targets. It may be easier to correct the active RVs in active stars! alpha Cen B is a good example. See also Meunier & Lagrange (2013) for planet detectability of the Sun as a star. "...we estimate that the prob larger than 50% to detect a 1 Mearth at 1. AU requires more than 1000 well-sampled observations and a Ca H&K S/N larger than 120."

What about M dwarfs? Rotation periods are very slow for M dwarfs. Really cool spots have very different spectrum than that of star. (not just convective blueshift) Actually have different, often molecular species in the center of spots! (Audience sais whoa!) Scharmer et al. (2011; Science)

Andrew Szentgyorgyi (Pronounced: Saint-Yeorge-ee) Improving Exoplanet Mass Measurements in the ELT Era

[Can't load movie] "Well, that get's me back on schedule!"

Giant Magellan Telescope (GMT) is 7x8.4m primary mirror segments for 25.4-meter effective aperture. The first primary mirror is complete, next one being cast in October.

Developing a precision RV spectrometer for the GMT called G-CLEF, which is only accepted PRV instrument for an ELT. Anna Frebel is the Science lead.

G-CLEF is very HARPS-like and mounted in a room within a room. 35% peak efficiency! Big advantage of large aperture is large S/N, which is crucial for 10 cm/s.

Going to use a laser frequency comb for wavelength calibration (astro-comb). Shows results of astro-comb test results on HARPS-N, shows plot demonstrating what Stephane said earlier about nonuniformity of CCD pixel sizes.

Shows test results from HIRES show that the spectrometer instrumental profile changes on timescales of minutes. Temperature and pressure are not changing on such a timescale. Must be guiding and/or seeing. Problem for Keck/HIRES, but GMT will have a visible-light AO system in the mirrors. Debra Fischer has been a voice in the wilderness saying that AO in the visible would stabilize the stellar PSF on the fiber/slit. This out-of-the box thinking is pretty important. You also get better fiber coupling, reducing slit losses and boosts system throughput. MagAO tests by Laird Close show that this will work really well with GMT (MagAO has a fiber port!).

You can buy a G-CLEF t-shirt. Also invited to the Oct 22-23 G-CLEF meeting at Carnegie Pasadena. Shows GMT video and warns us to watch out for lasers when you fly into Santiago!

Liveblogging LathamCon

Jason Wright liveblogged the Dave Latham celebration yesterday, here, herehere and here. I'll continue for day 2.

The first session is The Legacy of Multi-Planet Systems and Photo-dynamics.

Dan Fabrycky is first up with It's All About Time. Before Kepler, Gl876 and the Pulsar planetary system were only systems that required dynamical fits to explain data. RVs and timing are 1-D measurements. Hard to get 3D info from 1D measurements.

Transit timing variations (TTV) technique first published circa 2005, could be used to discover planets, but initially only really used to place limits on additional planets in known transiting systems. More and more data led to more and more upper limits. Frustrating. But because companions to hot jupiters are rare. TTVs took 5 years of toil to little avail.

Kepler brought an embarrassment of riches for TTVs, seeing variations up to 1 day. What's embarrassing is that there are so many we haven't analyzed them all! Dan shows lots of TTV O-C plots.

Huge number of compact, low-mass, multi-planet systems were unexpected. Intriguing lack of pileups at 2:1 and 3:2 resonances, but are many planets just longward. TESS with only 30-day dwell time will see many multis but will have limited TTV opportunities.

"I can't give a talk without showing this." Shows Kepler Orrery II with Flight of the Valkaries in background. If he didn't include the music, Dave Latham would complain "Where's the music?!" So here's the music.

Shows example of Kepler-30 system with 1.5 day TTVs. This is a completely parameterized system, with masses, radii, inclinations and spin-orbit angle. This is a system with large planets. With small planets we're stuck with degeneracies.

There are a lot of forward-looking words in this conference: Emerging, opportunity, Push toward. Dan's session has "Legacy" in title, which is kiss of death fo ran assistant professor :)

Looking ahead the ground-based programs should capitalize on Kepler's legacy. Three projects:

1) With Kepler no longer looking at the sky, so we have to figure out masses of of potentially habitable-zone planets. RV amplitude falls off at larger P, but TTVs increase in precision at long P. In Dan's opinion, this was one of best motivations for the extended mission. How to do this now? From space since transits are shallow so it'll be hard [even with HST]. We did this for Kepler-62 to get weak mass measurements. Lisa Kaltenegger asks "What did you find?" Dan: "You can read the paper. I think you're on it :)" Lots of laughs.

2) Predict forward for future transits that can be observed from the ground later. Spend an extended TESS mission on the Kepler field with a long (~1 year) dwell time.

3) Search for secular effects, namely transit duration variations. Doing this with Sean Mills at Chicago for KOIs in Kepler database. There are some mutually-inclined systems, but they are much rarer than aligned (flat) systems. These secular effects are worth following up.

Next up is Bekki Dawson with The Legacy of Giant Planet Dynamical Histories.

(Like Dan Fabrycky, Bekki is a good friend and frequent scientific collaborator of

Thanks Dave for inspiration while she was at the CfA. Wonderful mix of kindness and high expectations of young people.

Imagine that you find two RV planets at 0.11 and 0.24 AU. How'd they get there? Now imagine that you find a Super-earth transiting a bright nearby star. Tempting to think this is pre- and post-Kepler. But this is the same system (55 Cnc)!

We used to think that giant planets form far away and somehow move inward to form diversity we see today. What migration process does this? Two classes: gentle disk migration, violent multi-body interactions. Gentle mech. should circularize and leave planets aligned with star. Violent should pump ecc and tilt orbits. Both can lead to hot Jupiters, but how to distinguish.

Socrates et al. (2012) predict that violent mechanism should have planets following a constant constant angular momentum tracks, with many super-eccentric proto-hot jupiters. Let's look for them. But traditionally need RVs to measure eccentricities. But Bekki developed a method called the "photoeccentric effect" to measure ecc from the light curve (Dawson & Johnson 2012). Used the method to search for super-ecc planets in Kepler sample. Didn't find as many as Socrates et al. predicted.

But did find KOI-1474 with e = 0.8 and huge TTVs (~1 hour) from non-transiting planet (Dawson et al. 2012). Photo-dynamical analysis of huge number of follow-up measurements and Kepler data give us a unique solution for perturbing body: mutually aligned with inner planet and P = 660 days. HIRES data collected by CPS team at great expense. RVs match photoeccentric ecc very well!

Back to missing super-eccentric proto-hot-jupiters. Upper limit of 33% on contribution of super-eccentric migration mechanism of Socrates et al. Consistent with Noaz et al.

Jupiters around metal-rich host stars have much higher eccentricities, while metal-poor stars have more circular orbits (Dawson & Murray-Clay 2013). Also, the period distribution is very different between high- and low-metallicity stars. The hot-jupiter "pile up" only found around metal rich stars, which may explain deficit of hot Jupiters in Kepler sample if Kepler field stars are slightly metal-poor compared to Solar neighborhood. Planets orbiting metal-rich stars show signatures of planet-planet interactions.

Lack of giant planets around metal-poor stars, but common small planets around metal-poor argues for looking for bio-signatures in transiting planets around metal-poor stars since those Earths will be less harassed by large planets in the system. Conclusion: look for small habitable-zone planets, but do not ignore the effect of large planets!

Tsevi Mazeh is up next with BEER analysis of Kepler and CoRoT light curves: Kepler-76b: A Hot Jupiter with Evidence for Superrotation.

BEER = BEaming, Ellipsoidal & Reflection effects

Story of how Tsevi met Dave. The poster of Dave on his Ossa motorcycle used to be on Dave's door. Tsevi needed to find a collaborator who could do precise RVs. He wrote to Dave and agreed to stop by the CfA on his way from Lick to Tel Aviv and saw the poster on the door. At the time Dave was the Associate Director of the OIR division, he was very busy that day due to an emergency and couldn't meet with Tsevi. Wasn't sure if he wanted to work with such a busy person. Later decided to collaborate after all and they did an RV survey.

They focused on M dwarfs. Doh! M dwarfs don't have hot Jupiters. But did publish (Mazeh & Latham). Not many citations, but led to a great friendship. Also published a bunch of binaries. Always announced in units of 40's.

BEER algorithm: First source of effect is relativistic beaming. Star is brighter when coming toward you, fainter as it recedes. Second, tidal deformation of the star. Third, reflection of starlight off of the planet. All are periodic, but different signals that can be decomposed from a light curve. The amplitudes of effects have very different scalings with period ($P^\alpha$ where $\alpha = {-1/3, -4/3, -2}$).

BEER observed in KOI-74, but the planet was already known from transits. Want to use BEER to find new, non-transiting/non-eclipsing planets/binaries. Then confirm with RVs from collaborators such as Dave. Found seven new binaries (Faigler et al.).

Found a planet and announced last week: Kepler-76 (Faigler et al. 2013). Turns out there is a transit, a big one, too! Even an occultation. But V-shaped, so it was cataloged as an EB and wasn't a KOI. "We saved it from that horrible fate! :)"

However, for another system, the BEER-predicted Doppler amplitude is a factor of 4 larger than measured RV amplitude. "Factor of 2 you can handle. Factor of 4? Too much!" Turns out that the planet has "super rotation" with a hot-spot offset from the sub-stellar point due to atmospheric winds in planet (see Knutson et al. 2009, Showman & Guillot 2002). See this in four other cases!

Illustration of super-rotation for a hot Jupiter: surface map.

Future prospects of this new discovery method are very good! Especially for massive, close-in planets and brown dwarfs.

Instead of raising a wine glass, I toast Dave with a glass of BEER!

During Q&A, Andrea Dupree remarked that this technique doesn't require the same strict alignment as transits, so BEER light curves can be used to detect many more planets than transit light curves!

Dave Latham's need for speed

I'm in Cambridge, MA for the celebration of Dave Latham's 50 years in astronomy. Dave is an astronomer at the Harvard Center for Astrophysics (my soon-to-be home) and one of the pioneers in the field of precision radial velocity measurements. In fact, in 1989 he found the first planet, HD114762b, which is now known as Latham's Planet (read here or see Latham et al. 1989).

At the celebratory dinner I learned that, among many other interesting things, Dave was once a world-class endurance motorcycle racer. Seriously. He was once the best in the US and ranked #14 worldwide. He was even sponsored by Kawasaki Motorcycles who provided him with a competition bike. Here's an ad I found after a bit of Googling:

The caption reads
Dave Latham is an astronomer for the Smithsonian Astrophysical Observatory, a profession which demands technical precision. He is also a motorcycle competitor---a gold medal winner at the Isle of Man I.S.D.T. Dave chose the Ossa bike for his competition bike. His Ossa bike has successfully completed the following events: 
1970 International Six Day Trials El Escorial, Spain  
350 Mile Lonesome Pine 
National Enduro, Virginia 
500 Mile Greenhorn National Enduro, California 
Trask Mountain Two-Day International Trials, Oregon 
The Corduroy 500 Mile Canadian National 
…plus hundreds of miles of riding, and winning local events. 1972 Dave Latham Ossa competition bike – Reliability above all
How appropriate it is that the add highlighted Dave's precise velocity!

Monday, May 20, 2013

How a differential works

No, not a differential equation. But a differential. Here's a wonderfully clear tutorial that was brought to my attention by fellow astronomers, Cullen and Tom, over dinner tonight:

Where are all the women professors? Among the recently hired!

I recently wrote a series of posts entitled "Where are all the women professors?" here and here. I began with a simple premise: "men and women are equally capable of succeeding as professional astronomers. There is no inherent (intrinsic) difference in mental capacity, creativity, ability to learn, or any other factor that plays into the success of an astronomer." From there I examined the role of unconscious bias as one of the factors in a "leaky pipeline" that has resulted in an underrepresentation of women among astronomy professors.

The post was picked up by the Women In Astronomy Blog, and a commenter wondered, "What is the fraction of women hired on tenure track during the same time period as the statistics of the graduating students?" While the present representation of women among various astronomy faculty hovers somewhere around 15%, is there evidence that there have been improvements in recent years? The question stuck with me, but I wasn't sure how to assess it. However, the method recently became obvious: the Astronomy Rumor Mill!

The Rumor Mill is a fairly accurate accounting of the results from each hiring season (roughly Feb-Apr each year). While the short-lists don't always reflect reality, the final decisions, denoted in bold text, usually are. I went combing through the last three years of the rumor mill (previous years, before it was hosted by AstroBetter, are unavailable) and found the outcomes of 67 faculty searches at US institutions that advertised junior faculty appointments in either astronomy or astronomy-oriented physics positions (this was somewhat subjective in a small number of cases). Here are the statistics:

Number of men hired:          42
Number of women hired:        25
Percentage female (2011-2013) 37.3%

That's pretty impressive! As I noted previously, roughly 30% of recent PhDs over the past decade were awarded to women. These numbers are evidence the field has been hiring at a commensurate (or higher) rate over the past three years.

So while the current percentage of women on the faculty of various astronomy departments is much lower than the graduation rate of women, there is evidence over the past three years that the systematic biases in past hiring are being corrected. This is very good news for the field. What will be interesting to see is whether these newly hired women professors earn tenure at the same rate as their male colleagues.

There were many other interesting features in the statistics of recent hires. I recorded the PhD institution of each hired individual. Harvard grads where hired at the highest rate, making up 8 (12%) of the 67 hires. Next was the Princeton cohort, which made up 6 (9%) of all recent hires, followed by Berkeley (7.5%), and OSU (6%). Then came Caltech, Hawaii, UCLA , Washington and Arizona (3 [4.5%] each). Graduates from these 9 institutions made up 57% of all hires in the past three years. (Keep in mind that comparisons among these institutions is difficult since the sizes of their grad programs vary quite a bit. Harvard has the largest number of hired graduates, but it also has the largest graduate program.)

The gender breakdown of hiring among these nine institutions is pretty remarkable: women make up 8 out 17 recent hires at these institutions. To my eye, this is a sign of significant progress. But there's also a way to go still. The faculty of those nine institutions comprise a total of 195 professors, of which only 35 (18%) are women, and a third of those women were hired in the past 6 years.

So I'd conclude that there's a lot of ground to make up, but that based on the statistics of the past three years, the field is making significant progress. And I suspect that that progress will be accelerated as more people come to value diversity as a integral part of overall excellence.

Friday, May 17, 2013

This is water

Here's an incredibly inspirational speech and video (via Trevor's Google status). From the Youtube video description:
In 2005, author David Foster Wallace was asked to give the commencement address to the 2005 graduating class of Kenyon College. However, the resulting speech didn't become widely known until 3 years later, after his tragic death. It is, without a doubt, some of the best life advice we've ever come across, and perhaps the most simple and elegant explanation of the real value of education. 
We made this video, built around an abridged version of the original audio recording, with the hopes that the core message of the speech could reach a wider audience who might not have otherwise been interested. However, we encourage everyone to seek out the full speech (because, in this case, the book is definitely better than the movie).

The Kepler Mission ain't over till it's over

From Joseph, who works at JPL:
Don't be so quick to bury Kepler. There is a famous quote from Mark Twain: "The reports of my death have been greatly exaggerated." Kepler is now in Point Rest Mode, with very low fuel consumption, giving the reaction wheel jockeys a chance to revive reaction wheels #2 and #4. It ain't over 'till it's over.
C'mon Kepler, you can do it!

Thursday, May 16, 2013

Kepler: RIP

Kepler's reaction wheel #4 has failed, leaving the telescope unable to point precisely. It has been an amazing ride. We now know of 2740 transiting exoplanet candidates, many which are still awaiting confirmation and characterization. We know of circumbinary, Tatooine-like planets. We've found eclipsing degenerate objects tidally deforming their companion stars. We've found compact systems of tiny planets parked right next to their stars. We now know that small planets abound throughout the Galaxy. The field of exoplanetary science, as well as astrophysics in general has been forever revolutionized by NASA's highly successful Kepler Mission.

But now it's time to say goodbye.

I did a phone interview yesterday with a reporter from the LA Times. She asked if I was mourning. I hadn't thought about it until then. But yes. Yes, I was mourning. My group about to have 5000 hand-picked red dwarf targets added to the Kepler observing list. Those stars and their myriad small, potentially habitable planets will go unobserved. So personally, yes, I'm extremely sad.

I'm also sad for all of the hard-working scientists and engineers at NASA Ames. They have gone above and beyond on a shoe-string budget to generate the exquisite Kepler data sets that my group has made a living on for the past three years. Thank you Kepler science team! We mourn with you the loss of this amazing scientific instrument, this previously unblinking eye into the Cosmos.

Geoff Marcy sums it up well on Facebook:

Stop all the clocks, cut off the internet,
Prevent the dog from barking with a juicy bone,
Silence the pianos and with muffled drum
Bring out the coffin, let the mourners come.

Let jet airplanes circle at night overhead
Sky-writing over Cygnus: Kepler is dead.
Put crepe bows round the white necks of doves,
Let the traffic officers wear black cotton gloves.

Kepler was my North, my South, my East and West,
My working week, no weekend rest,
My noon, my midnight, my talks, my song;
I thought Kepler would last forever: I was wrong.

The stars are still wanted now; let's honor every one,
Pack up the moon and dismantle the sun,
Pour away the ocean and sweep up the woods;
For nothing will ever be this good.

With thanks to W.H.Auden.

Tuesday, May 14, 2013

Exoplanets Explained

Here's a continuation of my group's conversation with Jorge Cham about stars, planets and everything.