The Microlensing Data Challenge is now officially closed - thank you to all the teams who took part!

What happens now? Evaluating the Entries

All entries will be collated, and the fitted parameters cross-checked against the event parameters used to simulate the data. Each entry will be anonymized to minimize unconscious biases as much as possible, before it is evaluated by a panel of judges between December and January. The intention is to announce the results at the Microlensing 23 meeting in New York in January 2019. A paper summarizing the simulated dataset, the approach taken by each time, and the results of the challenge will be prepared and all participants will be invited to be co-authors. However, teams are encouraged to prepare independent papers describing any innovative aspects of their approach to tackling this challenge.

Goals of the Data Challenge

The analysis and modeling of microlensing events has always been a computationally-intensive and time-consuming task, requiring a powerful computer cluster as well as well sampled lightcurves. While the number of interesting events with adequate data remained fairly low, it has been practical to perform a careful interactive analysis of each one, often with the aid of a powerful computer cluster. Even so, a number of challenges remain, particularly concerning the analysis of triple lenses.

This is expected to change with next-generation surveys, especially with the launch of WFIRST. This mission is expected to detect thousands of microlensing events, including hundreds of planetary events. Clearly, our analysis techniques need an upgrade to fully exploit this dataset, and we encourage people from outside the current microlensing community to bring in diverse expertize.

  • To stimulate research effort into outstanding modeling issues
  • To stimulate development of algorithms to detect and classify microlensing events in WFIRST data
  • To stimulate development of software for modeling microlensing events, capable of conducting analyses of WFIRST-scale datasets
  • How the Challenges Work

    We will release a series of simulated datasets, each consisting of a set of lightcurves with durations and cadence representing those expected from the WFIRST microlensing survey. The datasets will be designed to test our analysis capabilities in different ways.

    Release date: late January/early February each year
    Submission deadline: October 31 of the same year, 23:59 UTC.

    Each dataset will be made available from this website coincidently will the annual microlensing conference, together with a link for the submission of entries, and documentation describing the data products and metrics required for the evaluation of all entries.
    The exact parameters of the simulated lightcurves will be known only to the simulator, who will keep them secret until the submission deadline.
    All entries will be judged by a panel of experts drawn from the microlensing community.

    Although entries will be assessed on the accuracy of their models, the emphasis of the challenge is on stimulating novel approaches to the challenge problems. For example, an entry that tests a new solution to a particular element of the modeling or classification process is encouraged, even if it doesn't provide a "complete" solution in every case. Full guidelines to the challenge can be found here

    Publicly Available Software Tools

    A number of software packages for microlensing modeling have been made publicly-available and we recommend that entrants explore them. Details of these can be found under the Software Tab

    Tiffany Meshkat from IPAC has kindly developed a Python notebook tutorial which provides an introduction to three public microlensing packages: PyLIMA, MulensModel and muLAn. This is a great way to get started if you're new to the field! Click here to download a tarball of the notebook, data and instructions, or visit the IPAC website

    Want to learn more?

    To hear updates about the challenge, please subscribe to our data challenge mailing list
    To access the challenge datasets, to ask questions and to form or join a team, please join our data challenge github organization by contacting rstreet (at) lco.global.