Then he said that he wished his son, Tracy, had an interest in real science like me. When I asked what Tracy was doing, Henry described a silly project of hanging out with computer engineers.

I enjoyed "The Soul of a New Machine" when it came out. I always hoped that I could meet Tracy someday. I guess not. He did great work, I hope his father learned to be proud.

There are three things to show:

1. A pdf of the draft: https://www.fraserphysics.com/book.pdf

2. The hmm project which provides code for state-space models: https://gitlab.com/fraserphysics/hmm with documentation at https://fraserphysics.gitlab.io/hmm/

3. The hmmds project which provides code for using state-space models to address some examples: https://gitlab.com/fraserphysics/hmmds with doumentation at https://fraserphysics.gitlab.io/hmmds/ The default target for hmmds is to build the book. On my home system it takes about 10 hours to build the book.

I would be grateful for comments or suggestions on any aspect of any of the three things.

I am particularly interested in help with the following:

1. I use NixOS for developing. I've documented progress on making the code useful in other enviroments at https://fraserphysics.gitlab.io/hmmds/getting_started.html Please let me know if you are able to use the code elsewhere.

2. In the draft of the book, I have an analysis of the convergence properties of the EM algorithm spread between Section 2.5 "The EM algorithm" on page 43 and Appendix B "EM Convergence Rate" on page 137. The analysis is different from anything I've seen published. I would like feedback with corrections and or citations.

Comments URL: https://news.ycombinator.com/item?id=46006083

Points: 4

# Comments: 0

The code and text are at https://gitlab.com/fraserphysics/hmmds. From a Nix command line, "make book" builds the book in about 10 hours.

I'd be grateful for any feedback on the book or the software.

While a cup of coffee often helped me feel better in years past, I've found eliminating coffee from my diet has reduced my headache frequency. Triptans help when I get the occasional headache now.

I was disappointed that the article didn't talk about the variety of migraine manifestations. Some of the manifestations can be confused with transient ischemic attacks.

I am also not a doctor. I have an appointment with a specialist MD. My PCP was initially alarmed, but after some research and consultations believes that I am not in immediate peril.

Unfortunately the devices that those links point to would not provide 24 hours of continuous data. The data that Phillips won't give me is a continuous record of about 720 hours with a few small breaks. The interesting event was one 2 second interval in the middle of one night.

Do you have advice on how to get the actual data?

My second question for non-US readers:

Does $9,000 seem excessive?

Background: For 10 minutes in April I was seeing double. Fearing that it was a transient ischemic attack, MDs ordered lots of tests. The heart monitor was to check for arrhythmias that could cause a clot. I was surprised that the study found ventricular tachycardia; see: https://www.fraserphysics.com/page2_ecg_report.pdf

Coincidentally, I have been thinking about ECGs and have written code for analyzing them; see Section 6.3.1 of my book at https://www.fraserphysics.com/book.pdf. My third question is for anyone who knows about ECG analysis technology:

Is the analysis technique I describe in Section 6.3.1 interesting/novel?

PS. None of the test results are very disturbing. I believe that my event was related to the migraines that I have all too often.

Comments URL: https://news.ycombinator.com/item?id=44294050

Points: 1

# Comments: 3

1. Disabling a terrorist weapon. When you find a mysterious box in NYC making ticking noises and emitting radiation, who you gonna call?

2. Forensics and attribution. When 1 fails how do you figure out what happened and who is responsible?