README.md

Documentation

In this document we will describe all use cases of the RTModel package and provide ready-to-use examples that you can copy/paste to your code.

Quick start

Provided you have prepared a folder for your microlensing event (e.g. /event001) with all datasets in the subfolder /event001/Data as detailed in the section about data preparation, in order to analyze the microlensing event with the default settings, just type

import RTModel
rtm = RTModel.RTModel()
rtm.run('/event001')

Then the modeling starts and you may sit back and watch the progress bars tracking the status of modeling. You will notice that all fits are nearly instantaneous except for the Binary-lens-single-source fit that takes 1-2 hours. At the end of all fits, you will see the final assessment of RTModel for the event, along with a summary of the best chi squares found for each category of models.

The final model(s) proposed by RTModel are collected as separate files in the directory /event001/FinalModels. You may plot them using the RTModel.plotmodel package:

import RTModel.plotmodel as plm
import glob

event = '/event001'
models = glob.glob(event +'/FinalModels/*')
model = models[0]

plm.plotmodel(eventname = event, modelfile = model)

Summary

In this documentation, we describe all functionalities of RTModel in detail. A novel user should read the following pages at least.

• Data preparation

• Full modeling run

• Model Categories

• Plotting models and results

Additional useful functionalities are discussed in the following pages.

• Archiving and updating models

• Satellite datasets

• Limb darkening

• Constrained fits

Advanced users may attempt a deeper understanding of the modeling steps and optimize RTModel by numerous options.

• Data pre-processing

• Initial conditions

• Fitting

• Model selection

• Final assessment and results

Astrometric information on a microlensing event can also be incorporated. The following pages discuss possible cases.

• Astrophotometric datasets

• Astrometric plots

• High-resolution imaging observations

Some more miscellaneous functionalities can be useful for diagnostics and customization

• Preliminary models

• Animation of the fit process

• Template library customization

Success rate

The success rate of RTModel has been evaluated on the simulated events created for the WFIRST data challenge by Matthew Penny. Here we report the results from the current version.

Planetary regime

In this challenge there were 51 binary events with q<0.03 (planetary regime). The results obtained by the current version of RTModel with the default options were the following:

• 47 full successes;
• 3 cases in which an s<1 solution was found instead of the s>1 or viceversa;
• 1 case in which the anomaly was not detected or ignored (too much noise)

Binary regime

In the stellar binary regime q>0.03 there were 75 events simulated in the data challenge. We note that orbital motion was simulated by assigning the two transverse components, which leads to non-physical trajectories not reproducible by circular orbital motion. Anyway, the results obtained are the following:

• 64 full successes;
• 6 cases in which an s<1 solution was found instead of the s>1;
• 5 cases in which a different binary model was found;
• 1 case in which the anomaly was too weak

The success rate is observed to decrease for events with high orbital motion. This should be partly due to the different way this effect is taken into account in the simulation and in our fitting code.

In addition, we note that such events with strong higher order effects would have been better fit with the nostatic = True option (see Initial conditions).