I am currently on the last part of data challenge 4. So far if we use the sampler, in the tutorials, we used on objects we know posterior values to beforehand. In this question, estimating the values of masses using bilby leaves me with the issue of not knowing any prior values of the signals I am finding. I can get the geocent time range from the q-transform and note the range of chirp mass from component mass range of 10-50 solar masses. And spins and alignments are known. But there is no mention of luminosity distance or, RA & Dec etc. Phase I can still give a range, but what am I supposed to set as priors for these values. Or do I not require this information and there is an alternate way to use bilby sampler to get the required estimates for mass.
Another question I had was that the before bilby it asks us to plot SNR and find estimates for component masses. Does it imply we need to estimate using SNR plots, or use bilby sampler here as well.
Please keep in mind, this is a little tricky, and most students will not run a successful
bilby job on the first attempt.
@Saksham You are welcome to estimate the SNR and masses using any method you like.
Thank you so much for the clarification. @jonah I have a humble request to extend the deadline for submission and checking of data challenge answers till the weekend. I am currently on the last data challenge but given the computational requirements, it is more hassle to deal with the data and hence takes time. I am sure I will be able to finish the challenge if more time is provided. If not till the weekend, I request at least a day’s extension if possible.
@Saksham We’ve set a target date for the data challenge to be completed by June 1. However, the online course will remain available, and you will receive a certificate of completion whenever you complete the course.
I would also like to be a part of evaluation of the Data challenge questions. Since I am mostly done with them only last challenge is left which will also be done by today.
In the last question, I have been facing this issue while running the sampler. I am not sure which value is complex since in prior I entered all values as either Uniform prior or fixed floats. I have tried other methods, but due to my lack of experience with other forms of samplers I am not able to figure out what to do next. If anybody understands what is going wrong please let me know.
maybe I’m not seeing from where the problem comes, but what I can suppose is that sometimes it happens that some functions return complex numbers by default without us knowing it, so I would not be surprised if something similar happened. If you already tried e.g. to print few elements of the likelihood or of the prior, could you please report the code which gives you these two quantities?
@virtuoso Thanks for helping out, but I somehow managed to resolve the issue. I used a different approximant, the same one I think the data challenge setters used to form the waveform they embedded in the challenge3.gwf file. So the code somehow worked even though it took about 3-4 hours to run. I don’t know how to deal with the time issue, maybe set more accurate priors.
@jonah, I have submitted the data challenge questions now since it took so long to do the bilby parameter estimation. I submitted the first 3 challenges and the data challenge 4 questions, except the last question related to the bilby part (which took a significant amount of time), before the 1st June deadline. I am not sure if we have to submit all answers for you to get access to them or if you got each answer as soon as I submitted the survey form of questions.
I did 7/8 questions before the deadline. And even the last question took a lot of time due to system and coding issues, but I still submitted it. I hope you can include my answers in the evaluation process. Kindly check my answers and let me know my score as well. It will help me a great deal.
Thank you to all tutors for this wonderful workshop and all the help. It was a great experience!
@Saksham Thank you! Glad you were able to complete the challenge!
Hey @Saksham which approximant did you use that worked? I had the exact same ‘can’t convert complex to float’ error as you.