Data Visualizations
A crucial element in any scientific research is the ability to visualize data. Most of my research focused on fitting parton distribution functions and using them to make physics predictions. Below is an assortment of different visualizations reflecting some of the work I have done. This sample is certainly not a complete collection and is merely a representative sample.
Most of these plots were created in Python or in Mathematica through the ManeParse package. Information on ManeParse can be found here and a sample Python plotting script that was designed to interface with nCTEQ++ can be found here.
Comparisons between experimental data and theoretical predictions are crucial for evaluating the quality of a PDF fit. I use a statistical technique known as χ2 fitting to do this. This involves calculating how far each predicted point is from the data point while accounting for the error estimate for that data point.
PDF fits come from collaborations all over the world. Here I used ManeParse to compare a collection of different fits.
It takes data from a lot of experiments to make a PDF fit. Here multiple data sets from three different types of experiments are shown for a sample fit. The small numbers above each bar refer to the number of data points in each set and the height of the bar is the χ2 contribution to the fit per degree of freedom.
Code validation is incredibly important in physics. I have done a lot of work confirming different programs operate in the ways they claim to. Here I show the precision of two such codes.
Being able to make predictions with a PDF is crucial to what a theoretical physicist does. This is where all the work analyzing data, studying theory, and coding software combine into something that benefits the physics community. This prediction here is for an experiment that has not even been built and was done partially as part of a presentation to funding agencies on the need for such an experiment.
Science without error bars isn't really science. Here I show two PDFs fit with error bars calculated in ManeParse.
PDF fitting is truly an exercise in connect-the-dots, except there are hundreds of dots and you have to connect them in 20+ dimensions. To do this I work with functions that have lots of parameters and then try to minimize that function by changing those parameters.
This plot may seem simple but it actually represents a critical discovery in the field of PDF fitting. Here I demonstrated that two completely different techniques could both be used to make predictions that could be compared to data. For more information on this, click here!
Parton Distribution Functions once all is said and done represent a probability and just like every other probability they have to add to 100%. Here I show how the different quarks and the gluon share the total momentum of a proton at different energies.