These are several different fits, each one gaining another pair of variables in the fourier transformation. The fit is about the same from 4 parameters on and actually gets worse when getting to large numbers of parameters. I've also included the chi value and the BIC value on every single graph. The BIC value as of right now increases fairly linearly as I add more parameters.







Here are some additional graphs I've been working on that have been fit with a fourier fit of 4 different terms
2MASS-J16291840+0335371








SDSS-J075840.33+324723.4

Comments

Post a Comment

Popular posts from this blog

flux percentage graphs

Image
Today I've been working on several different items. First I have figured out why the flux percentage was failing. I was summing the edges of the data and that for whatever reason was causing the whole flux to drop significantly. By only summing the flux from 10600 to 17000 I found that it greatly stabilized the result and I got exactly what I was expected to get. I added to the upper left when less than a percent changes. It is pictured below. Interestingly enough, I found that the extraction width is fairly close to the optimal noise for the waterband. This can show that using the noise in the waterband as a guide to choosing an extraction width agrees with using the amount of total flux captured.
Read more

working on the waterband graphs

Image
Today I found something interesting playing around with the test data. I found that when graphing just the raw data, I got a very noisy looking graph with only a sort of dip in the waterband. I thought this was strange, especially since in the paper they have a very clear dip in the waterband. Through some experimenting I found that using a rolling mean to smooth out the data, I produced nearly the exact same graph as was produced in the paper, with the same type of features in the waterband. I then did this same thing on the data that we have and produced a much cleaner and clear signal than what we had before. I included the test data and the data we are currently working with as well. The smoothing I think brings out a better waterband feature for both test targets that we have been working with. SIMP0136 2MASS-J16291840 SDSS-J075
Read more

waterband depth and graphing variables seperatly

Image
Today I worked on producing a waterband depth graph, which turned out to be more complex than I first thought, especially when propagating the error through to the end. I have left an example of one of the waterband depth plots at the bottom of this post. I've also been working on graphing the variables separately for the curve_fit function, which I did successfully. Interestingly enough, each sin cosine pair came out smooth. However, when fit for the function as a whole, the graph comes out bumpy. I also asked the internet about any ideas of what it could be, and I got the answer that it was an expected behavior of the curve_fit function. I've also included pictures below.
Read more