Lab 6

For LHC, the introduction slide is available here: Lab6 LHC bkg suppression.pptx Download Lab6 LHC bkg suppression.pptx. The lab descriptions are here: lab6_LHC_instruction.html. Download lab6_LHC_instruction.html.

For HERA, the general idea of our analysis is to identify frequencies that correspond to constant (with time) signals from distant galaxies. An introductory slide is Lab6 HERA bkg removal.pptx Download Lab6 HERA bkg removal.pptx

To do so, we will proceed in two steps:
1. Eliminate frequencies that have been contaminated by human sources (e.g., TV, etc.).
2. Determine using statistical methods the minimal amplitude that is significant, given the noise inherent in our detection, and identify frequencies that have amplitude above that threshold.

Step 2 will be covered in Lab 7/8. For Lab 6, we will focus on step 1. There are two options: analyze the data and determine which frequencies seem to be contaminated or consult a list of commonly used frequencies and eliminate those. For this lab, we will do the first method (in practice, you would do both and more carefully study the results). 

Now, since the 'signal' in this study is expected to be relatively constant, what we should look for are unexpectedly large variations in amplitude over a short period of time; these can be identified as likely due to human contamination and be discarded. So, to identify such variations, what we are interested in is the amplitude of the difference in visibility between two adjacent time slices. Do the following:

1. For some baseline pair, compute the amplitude of the difference in visibility between each pair of adjacent time slices (there should be 19 such pairs, in total).
2. Make some plots investigating this data and determine a reasonable amplitude threshold that will eliminate outliers. 
3. Apply this threshold to each pair of adjacent time slices to get a list of all frequencies that you will remove. To be conservative, if the amplitude for some frequency is above your threshold in any of the pairs, then that frequency should be removed. 
4. Make plots of your data (e.g., waterfall plots) before/after filtering to convince yourself that you have removed frequencies that exhibit large variation over time.
5. Remake the average amplitude vs frequency plot that you did in the previous lab before/after filtering. Are there still any frequencies that appear to have large amplitude? If so, why were they not removed?
6. To identify contamination in this lab, we transformed our data, looking at the difference of adjacent time slices to find large variation vs time. This is, in general, a key concept: transforming our data in some fashion to isolate a particular feature of the data and analyze it more closely. For the next lab, we will look for signals that are constant with time. What is a similar transformation that is likely to increase our sensitivity to such signals, compared to transient signals?