Parallelize computation of first million zeroes of H_t for reasonable t

[sujitnair]

Goal: Optimize the existing framework (root solver approach) and target to have the first million zeros of {H_t} (for a reasonable range of {t}) and the corresponding plots.

I will start a branch for this shortly and update this issue thread.

[km-git-acc]

Although I have commented on this at Terry's blog, just wanted to mention I did not use the root solver approach. I evaluated H_0 using the AFE at small intervals of z and then counted the sign changes (the classical approach).

Because I knew the ground truth for H_0, I then compared it with the estimated roots and found the ones which went undetected.

[sujitnair]

@km-git-acc

Here is my proposition to parallelize without building a full-fledged SETI type system. We can keep it mostly manual for now but the following will minimize human errors.

1. Maintain a Google sheet which keeps tab on who is computing what range
2. The sheet should contain the range of t and the number of zeros targetted
3. To make sure all the numbers are computed with the same code/logic, we should store the git commit hash (output of "git log")
4. The sheet should contain other metadata like the command which is run and the status (progress, done, etc.)

Let me know how this sounds to you.

[km-git-acc]

Sorry I seem to have missed this. This looks like a good plan. I will create a Google sheet and fill it with some of my runs. How do I invite the collaborators here to that sheet?