Lag Likelihood Overhaul

Algorithmic Work

• Find a mapping between true_lag and estimated_lags
• @pep3 wants to see a plot set that contains: noise mag. vs loc. error mag. (5b6d66da, 5b4757d8, 044d6f55, acc2f641, b1a1a7f2, dfe82e39, 0ec61b1d)
• @murata wants to see a plot set that contains: gdop vs loc. error mag. (5b6d66da, 5b4757d8, 044d6f55, acc2f641, b1a1a7f2, dfe82e39, 0ec61b1d)
• Investigate why ideal lags are always positive.
• Find out the range of empirical lags.
• Solve #13 (closed)

CI Work

• Better CI turnaround with CI Datasets.

CI Datasets

• Random sample (n=100) from the task 1 and recording 1: Call this smaller dataset: single-static
• Random sample (n=100) from the task 2 and recording 1: Call this smaller dataset: multiple-static
• Random sample (n=100) from the task 3 and recording 1: Call this smaller dataset: single-dynamic
• Random sample (n=100) from the task 4 and recording 1: Call this smaller dataset: multiple-dynamic
• Run make_dataset script for tasks={1,2,4} and recording={1}
• Add the new .h5 files as tar.gz archives into the repo
• Add the pipeline scripts just like single-dynamic.gitlab-ci.yml file

Small Fixes

• Solves #1, #7 (closed), #9 (closed), #10 (closed), #11 (closed), #12 (closed), #13 (closed),
• Typo in locata_sigmoid_temporal.py causing the main event loop to stop at iteration number 100.

Jokes

• Of course, I found the hardest thing to pronounce to call this thing. negative log-lag-likelihood or negative lag-log-likelihood?!