Introducing myself (by myself)

[DavidMartin]

How did you find PF?

AI Search

Hello,


I’m an independent researcher with an interest in theoretical physics and information-based approaches to spacetime.
I’ve previously worked on data compression and information theory concepts inspired by Shannon entropy.


I’m here to learn, read discussions, and ask focused technical questions.


Looking forward to participating.

(link removed to unpublished paper o data compression for IoT devices)

 

[.Scott]

I’m an independent researcher with an interest in theoretical physics and information-based approaches to spacetime.

Welcome to these forums.
Information and space-time have one well-explored connection. You may already be familiar with the Bekenstein Bound.

 

[jedishrfu]

Hi @DavidMartin,

I read your paper and was surprised by its topic and how well you laid out your reasoning. I hope you're trying to get it published. Sadly, at PF we cannot review any unpublished papers or personal theories. Consequently, I've had to remove the link to your paper.

Please take some time to read our site's global guidelines. PF is a great site, but it is also highly moderated. Our global guidelines cover most of the dos and don'ts here. We limit our discussions to peer-reviewed papers published in reputable journals. All others we remove from the site.

---

I mentioned earlier that I had read your paper briefly and was surprised to see it was on data compression. I'm a retired software engineer who worked in the related field of data communications. However, more recently and relevantly, I'm in a PhD program at a local university, and just last semester (my first semester), I did a rotation project on Polynomial Regression of scientific data. Some of my work was gleaned from a paper by Bello et al., whose team developed a linear regression data compression algorithm for single-board computer hubs as part of a project to collect utility data from 780 million households in England for near-realtime analysis of the grid.

My project was to compare polynomial regression on scientific F32 data with several well-known compressors, including gzip, SZ(3), ZFP, and PFPL. My program was no match for any of them. In general, when the data is smooth and slowly changing, you can get impressive results, though the well-known compressors still excel. But when the data is more chaotic, it becomes very hard to find the right algorithm for compression, and polynomial regression was not the right choice.

I did a project pivot by suggesting that carefully applied polynomial regression could reveal data structure, giving it an edge in data analysis. I thought of it as the DNA of the data: apply quadratic regression to extrema, linear regression to monotonically changing data, and higher-degree polynomials to more chaotic regions.

Here's the Bello paper:

https://www.sciencedirect.com/science/article/abs/pii/S2352467722001540

You mentioned you are an independent researcher. What does that mean? Are you a retired engineer? Or a former graduate student?

It was refreshing to see your paper.

Take care,
Jedi

 

[DavidMartin]

Dear Jedi,


Thank you for taking the time to read my paper and for your kind words — it truly means a lot coming from someone with your background in data communications.

To answer your question honestly: I have no academic degrees or formal credentials. I'm entirely self-taught. "Independent researcher" is my way of saying I followed a problem that fascinated me and kept pulling the thread until something coherent emerged. I understand that may raise eyebrows in academic circles, but I believe the work should speak for itself.

(Mentor note: Removed article references)

Your rotation project sounds fascinating, and I see strong parallels with my approach. You've identified exactly the core challenge: smooth, slowly changing data compresses well with almost anything, but chaotic regions are where most compressors fall apart. My project addresses this through an evolving shared context that adapts to the data's behavior over time — essentially learning the "personality" of each sensor stream. Your idea of polynomial regression as a way to reveal data structure (what you beautifully called "the DNA of the data") resonates deeply with the complexity monitoring layer we're building with engineers like you and your background.

Thank you for the Bello et al. reference — compressing utility data from 780 million households for near-realtime grid analysis is precisely the kind of use case my project was designed for.

I'd welcome the chance to continue this conversation if you're interested.


Take care,David

 

[jedishrfu]

Hi David,

Sadly Zenodo is not a peer-reviewed reputable journal. I encourage you to read our global guidelines (url is in my signature below) to learn more about PF.

We have this stipulation for a reason. Too many times, we get posters wanting to push their personal theory of everything and we have to shut them down because it attracts trolls and others who think they understand physics better than the physicists.

I think this came about because of the myth that Einstein flunked math in school but then became a great physicist through self study and his thought experiments.

So please don’t post any more links to your work or websites. PF is not an SEO advertising billboard.

You can however interact with the community learning, posting and helping others with your experience and knowledge.

Jedi

Closing the thread.