Ups and Downs and Ups And

Visitors to oklo.org may have noticed that the site was down for most of Sunday. I’d been neglecting to update my WordPress installation, which lead to a problem with the database, and a huge load spike for the server. Everything seems stable now, and I’m now flossin’ 2.8.4 inch rims.

In the relatively near future, I will be modernizing some aspects of the look and feel of the site, which will make it more discussion-friendly, and more smoothly slotted into the hum of the outside world. No need to worry, though. We’ll continue to roll ad-free.

I’ve updated the second systemic console tutorial which guides the user through the remarkable Upsilon Andromedae radial velocity data set. The back-end database is getting closer to its relaunch, and the systemic console (version 1.0.97) is freely available for download here.

Read on to work through the tutorial.

Continue reading

the pause that refreshes

The systemic backend will be offline for a period of time starting on Monday Aug. 03. We’re pulling our server from its current rack space. When it comes back on line, it be on the UCSC network. The database has been fully backed up, so despite the temporary unavailability, there’ll be no loss of data. The oklo.org web log will continue uninterrupted.

When we return, we have several goals in mind for the backend. First, there will be support. Several UCSC physics and computer engineering undergrads will be joining the systemic team, and will be focused on improving the backend and keeping it running smoothly. Due to time constraints, and despite best efforts, we just weren’t able to keep up with this ourselves. Second, the backend will maintain improved integration with the console as the console develops, and will be more focused on scientific tools rather than the web 2.0 social network aspect. Third, we’ll be introducing features geared toward the use of the console as an instructional tool in astronomy, physics and astrobiology classes.

CoRoT-exo-2 c?

Image Source.

The CoRoT mission announced their second transiting planet today, and it’s a weird one. The new planet has a mass of 3.53 Jupiter masses, a fleeting 1.7429964 day orbit, and a colossal radius. It’s fully 1.43 times larger than Jupiter.

The surface temperature on this planet is likely well above 1500K. Our baseline theoretical models predict that the radius of the planet should be ~1.13 Jupiter radii, which is much smaller than observed. Interestingly, however, if one assumes that a bit more than 1% of the stellar flux is deposited deep in the atmosphere, then the models suggest that the planet could easily be swollen to its observed size.

The surest way to heat up a planet is via forcing from tidal interactions with other, as-yet unknown planets in the system. If that’s what’s going on with CoRoT-exo-2 b, then it’s possible that the perturber can be detected via transit timing. The downloadable systemic console is capable of fitting to transit timing variations in conjunction with the radial velocity data. All that’s needed is a long string of accurate central transit times.

The parent star for CoRoT-exo-2-b is relatively small (0.94 solar radii) which means that the transit is very deep, of order 2.3%. That means good signal to noise. At V=12.6, the star should be optimally suited for differential photometry by observers with small telescopes. With a fresh transit occurring every 41 and a half hours, data will build up quickly. As soon as the coordinates are announced, observers should start bagging transits of this star and submitting their results to Bruce Gary’s Amateur Exoplanet Archive. (See here for a tutorial on using the console to do transit timing analyses.)

The latest on 55 Cancri

Image Source.

Here’s a development that systemic regulars will find interesting! In a press release today, came announcement of the detection of a fifth planet in the 55 Cancri system (paper here). The new planet has an Msin(i) of 0.144 Jupiter masses, a 260-day orbital period and a low eccentricity. The detection is based on a really amazing set of additions to the Lick and Keck radial velocities:

For background on the 55 Cancri system, check out this oklo.org post from December 2005.

The outer four planets in the 55 Cancri system all have fairly low eccentricities in the new five-planet model. This leads to a diminished importance for planet-planet interactions, but nevertheless, the system does require a fully integrated fit. Deviations between the Keplerian and integrated models arise primarily from the orbital precessions of planets b, c, and e that occur during the long time frame spanned by the radial velocity observations.

Eugenio has added the velocities onto a fully updated version of the downloadable systemic console. The new version of the console adds a wide variety of new features (including dynamical transit timing) that were formerly available only on the unstable distribution. Check it out, and see the latest news on the console change log and the backend discussion forum. Over the next month, we’ll be talking in detail about the new features on the updated console.

Very shortly, a new entries corresponding to the updated 55 Cancri data sets will be added to the “Real Stars” catalog on the systemic backend. I’ll then upload my baseline integrated 5-planet fit to the joint Keck-Lick data set. I’m almost certain that with some computational work, this baseline model can be improved. Such a task is not for the squeamish, however. Obtaining self-consistent 6-body models to the 55 Cancri data set is a formidable computational task for the console. There are 29 parameters to vary (if the Lick, Keck, ELODIE and HET radial velocity data sets are all included). The inner planet orbits every 2.79 days, and the data spans nearly two decades. Fortunately, Hermite integration is now available on the console. Hermite integration speeds things up by roughly a factor of ten in comparison to Runge Kutta integration.

There have been hints of the 260-day planet for a number of years now because it presents a clear peak in the residuals periodogram. After the 2004 announcement of planet “e” in its short-period 2.8 day orbit, Jack Wisdom of MIT circulated a paper that argued against the existence of planet “e”, and simultaneously argued that there was evidence for a 260-day planet in the data available at that time. More recently, a number of very nice fully self consistent fits to the available data have been submitted to the backend (by, e.g., users thiessen, EricFDiaz, and flanker). Their fits all contain both the 2.8 day and the 260-day planets, and happily, are fully consistent with the new system configuration based on the updated velocities. Congratulations, guys!

Interestingly, the best available self-consistent fits to the system indicate that planets b and c do not have any of the 3:1 resonant arguments in libration. It will be interesting to see whether this continues to be the case as the new fits roll into the systemic backend.

Systemic in the Classroom

Image Source.

In our development of the systemic console and the systemic backend, we’ve strived to build a professional-quality tool that can be used by the general public. There’s no better way to get a sense of them planetary discovery process than to participate yourself, and so we’d like to encourage astronomy instructors to fold the systemic console into their curricula.

This link points to a Word format document of a sample homework assignment that makes use of both the console and the systemic backend. We’ve had good success with this particular problem set at UCSC, and it’s currently being implemented at MIT as well. The level has been found to be appropriate to an astrobiology class for science majors. There’s no math prerequisite, so it can also be fully useful for a non-major survey course.

If you’re an astronomy instructor and you’d like to incorporate hands-on planet finding into your course, let me know, and we can set up a fit submission aggregator for your students on the systemic backend.

fit to be timed

Image Source.

One reason why extrasolar planets are so exciting is because they are accessible. You don’t need a Ph.D. or a large laboratory or a space-borne telescope to make an important discovery. There are very few areas in science where such a wide pool of workers can contribute in a fully meaningful way.

On the systemic backend, the focus is largely on planet characterization through the analysis of radial velocity data. At Transitsearch.org, the goal is to provide the information that will allow small telescope observers to discover transiting planets. Transitsearch, however, is mainly a repository for transit predictions. We maintain information about when and where to look, but we fall short when it comes to explaining how to obtain high-precision photometry. There has long been a need for a good end-to-end manual on the art and science of photometric transit detection.

Bruce Gary is an experienced observer of transiting extrasolar planets, and is a member of the XO network, which has had made several discoveries over the past year and a half (see e.g. here). Bruce has written a book, Exoplanet Observing for Amateurs which he’s made available for free in .pdf form.

Bruce has also launched the Amateur Exoplanet Archive (AXA), which is a repository for light curves obtained for known transiting planets. If you get a photometric transit time series of one of the planets, then make sure that you submit it to Bruce’s archive. With all the data in one place, everyone will have easy access for analysis projects.

Transit midpoint times can be measured from individual light curves, and a sequence of midpoint times can be used to improve the characterization of a particular planetary system. To this end, Stefano has extend the .sys file format used by the systemic console to include “transits” data files (which take a .tds suffix, and which are separate from the .vels files that the console has used all along). If you have transit data, it’s simple to implement one of these files for yourself.

To see how it works, consider the recently discovered transiting planet XO-2. The published radial velocity data for this planet is already bundled with the console. On the AXA site, a total of five transits have already been archived for XO-2. Each of these transits has a measured Heliocentric Julian Date (HJD) for the time of transit midpoint, along with an associated uncertainty. I copied these data into a newly created “X0-2.tds” file in my console’s datafiles folder:

I then added the following lines to the .sys file for the XO-2 system:

Having done that, I launched the latest (“unstable” Aug. 21, 2007 version) of the systemic console. Stefano has been steadily improving the console’s algorithms, user interface, and performance. If you’ve been working with the standard stable downloadable console, you’ll immediately notice that there’s a lot of new functionality. We’ll be getting a manual out as soon as the much-anticipated Systemic Jr write-up is completed, but in the meantime, there’s a wide variety of resources on the backend that can help you navigate the latest console features.

With the .tds file linked in, the observed transit midpoint times appear as vertical red lines in the radial velocity timeline window. If the “fit transits” option is unchecked, then the console considers only the radial velocity data. If the “fit transits” option is checked, however, then the observed transit times are included as data to be fit. The uncertainties in the transit midpoints can be very small, and so this provides a very strong constraint on the period of the orbit and the time at which the planet crosses the plane containing the line of sight to the Earth. Note that the transit fitting can be done in a fully self-consistent N-body fashion if integration is enabled.

Try it for yourself!

As more transit data is accumulated, it will become possible to do some increasingly sophisticated analyses. Transit timing is potentially a very powerful method for detecting additional, as-yet unseen perturbing bodies in a given system. Objects like Gl 436 b are especially good candidates for this type of approach, and quite a bit of photometric data is being accumulated during the Gl 436 transits.

ƒr3$h R4Ð14£ V3£0(1713$

Image Source.

Eugenio has finished combing through this summer’s literature, and has added twenty newly published radial velocity data sets to both the systemic backend and to the current version of the downloadable systemic console. As a result of his efforts, new or augmented data is now available for the following stars: Cha Ha 8, GJ 317, HD3651, HD5319, HD11506, HD17156, HD37605, HD43691, HD75898, HD80606, HD89744, HD125612, HD132406, HD170469, HD171028, HD231701, NGC2423, NGC4349, HAT-P-3, and TrES-4. As always, the published literature citations for the velocities are contained in the “vels_list.txt” file that comes bundled with the systemic console download. The vels_list.txt file can be indispensible if you want to publish results that use the systemic package as a research tool — indeed, we’re quite excited that researchers are starting to adopt the console in the course of carrying out state-of-the-art research (see, e.g. here.)

There’s quite a bit to explore with these new data sets. Eugenio has had a first look, and included in his recommendations are:

GJ 317: This system (discovered by John Johnson and the California-Carnegie planet search team, preprint here) is only the third red dwarf that’s been found to harbor a Jovian-mass companion. The data shows clear evidence for one planet “b”, with at least 1.2 Jupiter masses and a 693-day orbit, and there’s a strong hint of a second planet in the radial velocity variations. Check it out with the console!

HD 17156: This data comes from a recent paper by the California-Carnegie team. There are radial velocities from both the Keck and the Subaru telescopes, and the signal-to-noise of the orbit is very high.

The data show a ~3 Jupiter-mass planet on a 21.2 day orbit. The orbit is remarkably eccentric for a planet on such a short period, leading to a 25-fold variation in the amount of light received during each trip around the star.

It’ll be interesting to get a weather forecast for this world, and it’s also important to point out that the orientation of the orbit is very well suited for the possibility of observing transits. Periastron is reasonably close to being aligned with the line of sight to Earth, leading to an a-priori transit probability of more than 10%. In the discovery paper, a preliminary transit search is reported, but only about 1/4th of the transit window was ruled out. With a Dec of +71 degrees and a nice situation in the winter sky, this is definitely one for Transitesearch.org’s Finland contingent.

transitsearch dot org

Image Source.

Gl 436 b was the first planet to be detected in transit after the radial velocity detection of the planet itself was publicly announced. Gillon et al.’s discovery shows that the basic strategy of checking known Doppler wobble stars for transits can pay off dramatically, and indeed it’s recharged my interest in keeping transitsearch.org up and running.

Successful transit predictions depend on having accurate ephemerides, which in turn depend on fits to the most recent radial velocities available. The period error in an old fit builds up to the point where the predicted transit window is longer than the orbital period itself. Indeed, relying on a published fit that’s five, six, or even eight years old, is akin to showing up at the 2007 Grammy Awards in a 2001 Escalade.

We’ve thus started the job of making sure that the transitsearch.org candidate tables are as up to date as possible. I’ve committed to spending a bit of time each day checking and updating the master orbit.data and star.data files that are used as input to the cron job that runs every night to update the prediction tables. In each case, we’ll use the most recent published orbital data for a given planet.

In addition, the eighteen known transiting planets have all had their ephemeris tables updated using the latest literature values for the orbital parameters. I got the most of these data from Frederic Pont’s useful summary table, and took the radial velocity half-amplitudes from exoplanet.eu and exoplanets.org. At the moment, the occultations are all treated as central transits by my code, which means that the predicted transit durations will in general be longer than the actual observed events. This discrepancy will be patched shortly, but in the meantime, the predicted transit midpoint times in the ephemeris tables should be extremely accurate for all 18 planets. (See the candidates faq for more information).

We’ve made the decision to base the main transitsearch.org candidates table only on published orbital fits that have appeared in the refereed literature. In many cases, however, one finds a need to go beyond predictions based on published fits. There are two main circumstances under which this can occur. (1) The systemic console provides the ability to obtain fits to all existing radial velocity data for any given system. For many systems, one thus has the opportunity to obtain orbital parameters for the planet that are more accurate than published values that are based on fewer data sets. (2) You may have used the console to locate a candidate planet that is not yet published. If this planet can be observed in transit, then you’ve got dramatic confirmation of your discovery.

Eugenio has written an extension to the bootstrap window of the most recent version of the console that allows anyone to make transit predictions for any planet produced by the console. In an upcoming post, we’ll look in detail at how this new feature works.

Systemic Jr. Fit Drive

Image Source.

A big thank-you to everyone who’s been participating in the drive to characterize and study the catalog of synthetic “Systemic Jr.” planetary systems on the Systemic Backend. There’s now enough data to indicate that the analysis is going to be very informative. We’re looking forward to revealing the properties of the underlying planetary systems that were used to generate the data. In the meantime, we need your help to adequately characterize all 520 systems. Data in need of better characterization are marked by flags:

Our backend server is now swarming with various hard-working software robots that Stefano has assembled. The 100-year stability bot is rousted out of bed and set to work whenever a new fit is submitted. It reports a quick initial assessment of orbital stability. Planetary systems that pass through the 100-year stability screen are then put in a queue to wait for the attentions of the 1000-year stability bots. Systems that make it through 1000 years with less than a 1% change in semi-major axis of their planets are awarded a snazzy green flag:

Occasionally, systems that are in mean-motion resonance can show periodic semi-major axis variations of more than 1% while still remaining indefinitely stable. A resonance bot that will go through the fits and check for these special cases is currently being readied.

Systems that pass the minimum stability requirement are handed to a bootstrap bot which uses the bootstrap method to estimate uncertainties on the planetary orbital parameters for each stable fit. We’re currently running the bootstrap bot under the assumption that the orbits are pure Keplerian ellipses, and so the calculations are usually quite rapid. Very shortly, the error estimates for the parameters in submitted fits to the real systems and the Systemic Jr. systems will be showing up on the back-end data pages.

Finally, an “F-bot” has been activated which performs successive F-tests on submitted multiple-planet systems. Using its results, we’ll have a better idea of when the addition of a planet to a system is warranted.

Bootstrap

Image Source.

Stefano and Eugenio have been making quite a bit of development progress on the downloadable systemic console. A new version of the console (available for beta testing on the systemic backend) is now capable of providing an estimate of the uncertainties on the orbital elements associated with a fit to a particular data set.

Radial velocity data don’t provide an exact determination of planetary orbits. The most obvious shortcoming is that Keplerian orbital fits can’t determine the inclination of the planetary orbits, and so for a given system, we’re only able to measure then mass of the planet multiplied by the sine of the unknown inclination angle. Furthermore, the stellar radial velocity signal created by a planetary system is corrupted by astrophysical noise introduced by the parent star, as well as by noise introduced during the measurement process here on Earth.

Determination of the true uncertainties in a planetary orbital model is a subtle problem (for more detail, see Eric Ford‘s recent work in this area). As a first straightforward step, we’ve implemented the so-called “bootstrap” method of error estimation into the console. The bootstrap works by taking the original data set, and then successively redrawing time + velocity + uncertainties triples from the data with replacement. This procedure creates alternate realizations of the original data set in which some of the original measurements appear more than once, and in which some don’t appear at all. The best-fit parameters obtained by the console are then used as a starting guess to fit the bootstraped data sets. The standard deviations measured from the distributions of orbital elements thus obtained give error estimates for the parameters of the original fit.

The bootstrap routine is menu-accessed, and is simple to use. First, create a fit to a dataset. In the example just below, I’ve fitted to the data for HD 80606:

Once the fit has been polished, the bootstrap can be run. In the default configuration it uses Keplerian fitting and does 100 trials.

HD 80606 has been observed for nearly 20 orbital periods, and velocities have been obtained at a wide variety of orbital phases. As a result, the orbit is very well constrained. The bootstrap indicates that the uncertainty on the e=0.932 eccentricity is only 0.003. For other systems, such as hd 20782, which also seems to have a high eccentricity:

the uncertainty on the parameters is much larger:

Give the routine a try! In upcoming posts, we’ll talk more about how uncertainty estimates will be incorporated into the planetary catalogs on the backend.