lena-lin / cta_transient_search Goto Github PK

Evaluation fails sometimes, because transient start slice is not correct! We take it as true position for evaluation.

The noise added to the templates is fixed for all cubes at the moment. All transients based on one template will have exactly the same shape.

Plot axes should be in ra/dec.

Difference between simulated start and first trigger is often 1, while first trigger is earlier than simulated start. Possible reason in np.where ?
evaluation.py / count_tp_fp_fn():
rt, = np.where(np.diff(ts==1)), with ts beeing cube[trigger_index], a list of False and true trigger values.

Test: a = [false,false,false,true,false]
np.where(np.diff(a==1)) results in 2 but first trigger should be at the index 3

Add +1 option in branch Find_Position

If you use this gitignore file you will have no issues with the __pycache__ stuff.

https://github.com/github/gitignore/blob/master/Python.gitignore