HNHacker News
TopNewBestAskShowJobs

KolenCh

350 karma · joined September 18, 2021

submissionscomments
KolenCh··on Yadm: Yet Another Dotfiles Manager
Doesn’t that limit you to deploy dot files to nix available system only? That’s the primary reason I don’t use home manager.
KolenCh··on Researchers spot black hole feeding at 40x its theoretical limit
As the black hole getting larger, it is more difficult to notice this difference (of crossing the event horizon or “observe the geometry around you”.) and as we are talking about the whole visible universe being inside a black hole, we are in this extreme large scale.

Also, I’m not sure why you’re arguing about the radial coordinate being time-like. You can only measure in your own local reference frame. You wouldn’t necessarily be able to transform between your own local reference frame to the blackhole’s if you don’t know you’re in one.

The universe as a black hole is actually a very old idea: https://en.wikipedia.org/wiki/Black_hole_cosmology

I’m not saying we are in one, but I’m saying it is not as obviously false as you might be arguing.

KolenCh··on It might be possible to detect gravitons after all
Photon-photon interaction is photon self interaction. Gravity/graviton self interaction then means graviton-graviton interaction. In general relativity, all form of energy would have an effect of gravity, and also react to gravity. Since all matter, including photon and graviton, has energy, then they should self interact.

In QED, photon and photon do interacts too and you can calculate its effect to be small. In GR, you can expect self interaction is small if the space time curvature is small.

(Photon is the particle that mediates QED, and graviton is the hypothetical particle that mediates gravity.)

KolenCh··on Yes, we did discover the Higgs
Having been in his lectures in the past, he is the kind of person who teach you to question what you are told/taught. You should know he basically does the same thing to the field of ML (as this does to HEP).

You know, when I first read this thread and the 3 posts involved, I find the original post Ben wrote arrogant and hard to swallow. But once I searched who he is, and recognized him, knowing his character I immediately “get his point”. While not an expert in the fields, I have both graduate level educations in HEP and ML. My point is that my conclusion is unlikely due to the lack of understanding of these fields, but more because of my understanding of who he is…

Admittedly, he should not assume people read it as he intended how people would perceive. It took a lot of contextualizations, including the expectation from the title he explained in the later posts, to really take his posts seriously.

KolenCh··on NumPy QuadDType: Quadruple Precision for Everyone
Thanks! That’s very interesting! Now I need to find my next project to try this…
KolenCh··on NumPy QuadDType: Quadruple Precision for Everyone
Thanks, it is very helpful. 2 questions:

1. What’s the implications when parallelization is considered? I.e. is those accurate sum algorithms parallelizable, and how would it compares to traditional parallelized sum? 2. What is the performance penalty I should be expecting?

KolenCh··on UK will give sovereignty of Chagos Islands to Mauritius
I’m sure the context is totally different. And yours is right as you are the citizen there.

But being a Hong Kong citizen, I have a totally different reaction to this news. (Projected to our own context.)

KolenCh··on Xkcd 1425 (Tasks) turns ten years old today
IIRC Jack “Wow” Davis (a photoshop expert) mentioned this is what his clients like to say to him: “Can you just remove the thing in the photo?” The “thing” can be the eyeglass they are wearing, etc. And he said, “come on, I’m going to come to your house and take it again.”

“Can you just…” sums it all.

KolenCh··on Optimizing your career for happiness
Books/articles like this is presenting a world view. I don’t think they think this is the only way we should view the world. But I guess they are preaching a way to view the world, and as any preacher does they try to convince you why this is a better way.

(Somewhere they did mention not all people have the luxury to practice this, but she did documented examples where people in less-than-ideal situation have it turned around and do what they love.)

Regarding self-help, she said

“There are other ways in which 168 Hours does not aim to be like many self-help or time-management books. I approach this not as a productivity guru, but as a journalist who is interested in how successful, happy people build their lives. I am particularly interested in how people who are not household names achieve the lives they want, and what we can learn from their best practices. There are plenty of books out there on Fortune 500 CEOs’ or celebrities’ tips for success. I’m more interested in the woman down the street who—without benefit of fame, outsized fortune, or a slew of personal assistants—is running a successful small business, marathons, and a large and happy household.”

Excerpt From 168 Hours: You Have More Time Than You Think Laura Vanderkam

KolenCh··on Ask HN: Former gifted children with hard lives, how did you turn out?
My ACE score is 7, but I don't feel like that for 2 reasons:

1. there are some questions that technically I should answer "Yes", but is of milder scale of the spectrum. 2. I occasionally shared my upbringing to other people, and they are all very surprised. Once I shared with a group of "trainee chaplains", the teachers are commenting something like "why I'm describing my experience without the emotion", that is, when I talked about something very sad, I am smiling when saying it out. And they also pointed out that I often are asking "why" when facing different situations. I.e. I tends to become "philosophical".

May be because of my upbringing then, I loved mathematics since I was a kid. I remember myself once said when I was very young, "I love mathematics because it concerns truth, that is independent of our existence, nor the existence of the universe for that matter." In the hindsight, I think it might be some sort of protective mechanism for me to escape from the reality.

But perhaps because of my upbringing, I've taken many detours in life. When I was in secondary school, I have no motivation to study anything, and so my academic results were very bad. Fortunately, I like to read books, and got interested in popular books on math and physics, and also read a lot of computer magazines. That helps me having good grades in those subjects. Eventually I was discovered to be gifted because of physics olympiad. And that changed the course of my life. Had that didn't happen, I'm not sure if I would even got into university, let alone having a PhD.

So in this sense the "gifted" part of your question becomes my ticket to get out of my "hard life".

Unfortunately, life is complicated. First, having a "hard life", I don't have anyone to guide me making good decisions. Like the sort of decisions about making life choices, how to pursue goal in life (or resist the temptation to pursue too many goals), etc. Or work ethics, self-disipline, etc. Raw talent can only help so far, and as I ages I found that while my will can be strong, my body is getting weaker (e.g. can't get too much all nighters anymore.) Finally, I made a fatal mistakes to want to help everyone, including my parents. That puts me into a financial disaster and almost ruined my career and life.

How satisfied am I with my life now? "Thinking" helps. As a christian studying the bible theologically, it does shape my world view in a way I can "let go" and don't dwell in the hardship I had. As a researcher, my "thinking" is my primary value to get a job. A bonus of being a researcher is that I'm far away from home, so that I get much less connected with those negativity back home than in the past, and the change in my wellbeing is noticeable to my wife. Loving "thinking" means I love reading and listening to podcasts, enriching my perspectives in life.

The most fulfilling part of my life is my family, I have a lovely wife, a lovely baby.

Financially we aren't brilliant. But somehow, even as both my wife and I grew up from very poor families, we don't value money too much. I just don't find it motivating to make a ton of money, nor measures my success based on that. (It doesn't mean we don't have financial stress, as supporting a family of 3 including a baby in the UK is quite harsh.)

I feel that your life is getting better, but not to the point you felt satisfied yet. I hope you'll find your satisfaction, and meanwhile, don't worry too much. You have seen many hardships. You're tough and you can handle it when it comes. Don't dwell too much on those that may come though. Good luck!

KolenCh··on Optimizing your career for happiness
> Don’t take it wrongly but "you should do what you love or your life is meaningless" is a very bourgeois idea.

I agree with you but it is not what Laura said or mean. She said,

> “Harsh as this sounds, if you’re not in the right job—a job that is moving you toward where you want to be in life—then you’re wasting almost all the time you’re spending at work.”

I'd like to point out 2 things:

- Perhaps it is understood without saying: there's no "the right way" here, as different people have different utility function. The corollary is that the alternative of what Laura is proposing can be equally fulfilling or more.

- Sticking to Laura's proposition then, is that you only have 168 hours a week. Work constitute to 30-50 or more hours, a significant chunk of your total amount. So, that's what "you’re wasting almost all the time you’re spending at work" means. Note that she's not saying this defines the meaning of your life. She's saying plainly that majority of your quota (hours) is spent towards work, if those amount of time is not spent towards "moving you toward where you want to be", then those time is counted as wasted, in terms of "moving you toward where you want to be". In order words, you need to be moving faster in the remaining amount of time you have elsewhere to compensate, which is harder. (Compare to another hypothetical you that can "moving you toward where you want to be" in both portions of your time, which would almost surely make more progress.)

In your case, you mentioned "actively disagreeing with most (but not all) of what the company I work for does". Having read Laura's book, I don't think in her definition yours must be "wasting almost all the time you’re spending at work". Because as far as I understand her book, if what you're spending time doing at work is, for example, building up a skill you want to build, or solving problems you like to solve, then this constitute towards "moving you toward where you want to be".

The more dangerous situation Laura warned against is something like the majority of the time spent at work, say 20+ hours, are spent towards meaningless things that don't advances you, such as attending many meetings you are not needed there, performing a duty that is not your duty and not something you want to develop a skill in, etc. I.e. what "advances you" is defined by you, but the author is actively guiding the readers to seek these out and defines it clearly.

KolenCh··on Optimizing your career for happiness
Laura anticipated that response:

“Ferriss downplays the idea of dream jobs. As he writes, “For most people . . . the perfect job is the one that takes the least time. The vast majority of people will never find a job that can be an unending source of fulfillment, so that is not the goal here; to free time and automate income is.”

And then she further elaborate from there, discussing how to create the right job, bridging the gap between the ideal and the reality. Here's the 3 levels of changes to do that:

“... If the answer is “no” to any of these four questions, what can I change? In the next week? In the next year? Can I create the right job within my organization? Another organization? Or will I need to go out on my own?”

I know we're not in a book club, and I, even the author, does not have to ultimate answer. But as she argued, there are things we can have more meaningful/fulfilling life from our 168 hours a week and it is not as pessimistic as it seems. I invite you to give the book a chance.

Excerpt From 168 Hours: You Have More Time Than You Think Laura Vanderkam

KolenCh··on Optimizing your career for happiness
Not an expert here, so I'll quote one from the highlights I have reading the following book:

“Harsh as this sounds, if you’re not in the right job—a job that is moving you toward where you want to be in life—then you’re wasting almost all the time you’re spending at work.” “First, “you should do what you love” means finding work that “matches well with your expertise, your creative thinking skills, and your strongest intrinsic motivations.” Intrinsic motivation means liking the substance of the work for its own sake. The second part, “you should love what you do,” means “finding a work environment that will allow you to retain that intrinsic motivational focus, while supporting your exploration of new ideas.”

Excerpt From 168 Hours: You Have More Time Than You Think Laura Vanderkam

KolenCh··on Kids who use ChatGPT as a study assistant do worse on tests
To paraphrase, it seems that it is suggesting that leaning is harmed because assistant helps them to get better results, but hurt them in the process of doing so, where less understanding and thinking is involved.

It is interesting that it likens that to autopilot, where FAA is recommending people to minimize for a similar reason.

Since Math is known to be not a forte of Gen AI, it is a bit questionable if that contributes to the effects they are seeing. But as they claimed, it seems the effects is related to the reliance and “short cutting the thinking process” but not the ability of gen AI in that subject.

That makes me think about parents helping kids to do homework (it’s a thing in Hong Kong.) The immediate effects would enhance their results in homework grades. But now I wonder if they are inadvertently hurting their exam grades.

In order words, I wonder how this result would generalize to human tutor as well. I suppose it would too, and depending on their pedagogy (or the lack of) it would harms students in a similar way.

KolenCh··on Kids who use ChatGPT as a study assistant do worse on tests
Link to paper: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4895486
KolenCh··on How we made Jupyter notebooks load faster
In an ideal world that we control everything and/or don’t need to collaborate with others, then whatever tooling one use is actually not that important (and each can choose the best fitting their needs.) So Jupyter+Jupytext is useful in the context of collaboration, where you can’t control your collaborators but want something from them.

While in an ideal world scientists who write softwares should write professionally, the same goes for anything they do, including math and stats used in their research, writing and typesetting and generates publishing quality visualization… That rarely happens because of how the academic world is financed, and the incentives associated with it. I can certainly complain about that all days, but in short a researcher hired by a research university, especially with a tenured track position in the US, will not be successful to get such position, let alone getting tenured, if they had not focused their scarce resource of time to maximize their “research output” (publications, grant, etc.), where software engineering is not part of. (Sorry, sentence too complicated.)

KolenCh··on How we made Jupyter notebooks load faster
I don’t disagree anything you said. Jupytext can be a good tool to bridge some gap, where you pair ipynb to a py script and can then commit the py only (git-ignore all ipynb for your collaborators.)

Also, while many practices out there is questionable, in alternative scenarios where ipynb doesn’t exist, they might have been using something like matlab for example. Eg, in my field (physics), often time there are experimentalists doing some coding. Ipynb can be very enabling for them.

I think a piece of research should be broken down and worked by multiple people to improve the state of the project. Some scientists might be passing you the initial prototype in the form of a notebook, and some others should be refactoring to something more suitable for deployment and archival purpose. Properly funding these roles is important, and is lacking but improving (eg hiring RSE.)

In my field, the most prominent way when ipynb is shared a lot is for training. It’s a great application as that becomes literate programming. In this sense notebook is highly underused as literate programming still hasn’t got mainstream.

KolenCh··on Show HN: Dump entire Git repos into a single file for LLM prompts
I tried code2prompt but the interface has a quirk—even if you specified an output file, it will always put the output to the paste buffer.

Which one of these you find the best? It’s quite tempting to write one myself for something as simple as this.

KolenCh··on What's functional programming all about? (2017)
You are talking past each other on what formula means. In FP, a recursive function is a valid “formula”. P.S. I’m not arguing for that characterization.
KolenCh··on String Theorists Accidentally Find a New Formula for Pi
You’re not wrong that your formula computes to 2 regardless what λ is. The difference is, this formula is trivial. Another difference is, the formula over there is an infinite series.

What it means is, first, try prove that formula does not depends on λ. It is not obvious on first glance it is. (You may try to d/dλ and proof that equals 0.)

Second, change λ, it gives you a different infinite series, while each converge to π, they converge at a different rate.

In fact, part of the game of computing digits of π is to use different formula, where each converges to π eventually, but they have different converging “speed”, in the sense that giving n=100 terms, some gives you much more accurate digits. Also, since each term has a different computing “speed” too, ie using a computer to compute a term might be much slower in one series, that even if it has better converging “speed”, would be very slow to compute.

The art of breaking world record then is to find a balance between the 2, such that overall it is fastest.

Regardless of that though, discovering a whole family of series converging to π is in itself interesting. A special case reducing it to something we already know is not too surprising, but a good entry point for a story.

KolenCh··on Black hole in orbit around planet could be sign of advanced civilization
This is speculation, sure. You may say all theorists do are speculations. They propose ideas, show that it is falsifiable, and it is not their duty to then write proposals and come up with experiments and then take observations and only having done that in a decade to publish one single citable thing. Not to mention this signature should be looked for in an experiment that has much larger scope and can be one of the hundreds of things a single analysis has included, and might at most constitute a few sentences in a dozens-page paper.

Your expectation is infeasible. Setting that expectation is irresponsible.

KolenCh··on Compressing data with sample points and polynomial interpolation
In that sense, extrapolation should never be used in “production”. At best it is for exploration.

One characteristics of ML is that this distinction often is not clear. (Hallucination, generalization, etc.)

KolenCh··on Compressing data with sample points and polynomial interpolation
New readers of John Cook may not know this: his posts are pedagogical, often are illustrating the gist of the idea or message, not including all the details or the optimal analysis. This post mentioned the motivation too.

Also, he is a consultant, and this blog generates interests of that. He often works in clients’ data and perform EDA and provides solutions, such as statistical reasoning or methods. Ie while he write codes, it’s often not for production. He points people in the right direction so to speak.

KolenCh··on Show HN: Python Compression Suite for Pandas DataFrames, CSV and Excel Files
I don’t think you understand what this project is doing. I skim through the doc and it seems it’s neatly summarized by the first paragraph:

> The library's main goal is to compress data frames, excel and csv files so that they consume less space to overcome memory errors. Also to enable dealing with large files that can cause memory errors when reading them in python or that cause slow operations. With lzhw, we can read compressed files and do operations column by column and on specific rows only on chunks that we are interesred in.

If I understand it correctly, it means that the goal is to keep a losslessly compressed copy of data in memory, and provide ways you could work with the data column by column or even in chunk, to reduce the amount of memory needed to complete an operation. And it deals with it generally (not necessarily categorical data) and losslessly (you cannot impose lossiness arbitrarily).

But this library seems to be designed for a very niche purpose. It mentions laptop here and there in the doc. And the use case is the kind of datasets with size just above what your laptop memory has, whereas the losslessly compressed data still fits. That makes it hard to write production code with, as the advantage of compression is unpredictable. Even if it is just for explorative data analysis, it puts a burden on the mental model to reason with, as you really need to be just in the right spot for this to be useful. (There are techniques that can stream data from file to handle data bigger than available memory.)

KolenCh··on Ask HN: How to manage a PhD project?
Some random suggestions:

- use Zotero to organize anything that can be referenced: papers obviously, but also any books with ISBN, even websites with snapshots (use the Zotero web extension) - learn about the CSL style and use Zotero to automatically export citations. Better BibLaTeX supports output to BibLaTeX. But if you can adapt your workflow to CSL YAML/JSON then it would be better. - use Notion with your academic email and then you can have free pro features. It has its own pros and cons, but you can upload anything there with virtually unlimited storage. (But obviously be careful to become over reliant on this.) - Keep logs of everything you do in a retrievable way, eg git. Take the lab notebook analogy but apply it to everything else, especially when code is involved. - try to make any code reproducible-ish. This is hard, but starts with manually logging things. You will want to be able to retrace your steps. - write a lot, write like you’re explaining to your future self (that may forget some of the details you know like the back of your hand right now). When you need to explain things to others, jot them down. When writing presentations for example, try write it like an outline of a mini paper (using speaker notes for things you don’t want to show on slide for example). The things you wrote could become handy when you later want to gather materials for a paper or dissertation. - because of this, when preparing presentation, with reasonable amount of time spent, try to generate images or equations that is publishing quality if you think it is going to be useful in the future. That kind of related to reproducibility. Going back to regenerate a graph differently is going to be hard. So better abuse the storage space and have a few extra formats than needed at the moment (matplotlib with pgf output for example.) - think about how you organize projects across different ecosystems with the same structure. The book Second Brain promotes using 4 categories Projects, Area, Resource, Archive and routinely tidying them up. It’s worth a read but it has its limitations.

Start writing your CV and drafting your dissertation in the beginning. The things you mentioned probably all fit into them and knowing how they fit into a CV/dissertation is a way of “managing” them.

P.S. using pandoc for all the writings is quite useful to me. Having a single source authoring presentation, LaTeX, documentation, etc. is quite useful. That allows me to copy and paste something I wrote for a presentation to my dissertation for example. YMMV however, as I’m a minority in that regard.

KolenCh··on I put a toaster in the dishwasher (2012)
It looked like the water going down splashed some milk back to there repeatedly and dried and formed some sort of crystallized milk powder that looks quite disgusting. I would recommend a wipe.
KolenCh··on I put a toaster in the dishwasher (2012)
I have a similar but less extreme experience.

I have a Tommee Tippee prep machine for baby formula milk. The part it has water coming out got so dirty after months of use. Due to lack of sleep like any father of newborn does, I flip it upside down and clean it with running water, and then I found out all the water finds its way into the machine.

Then I completely disassemble it, I use a dehumidifier to blow dry air directly to the screen that water made its way between the glasses.

Then after it is completely dried, I put it back, it continued to work.

So happens I’m a physicist too.

Now I should find an electronics to throw into the dish washer. Unfortunately I already get rid of my toaster because it got so dirty, I should have kept it.

KolenCh··on Workers are stuck in place because everyone is too afraid of a recession to quit
Read more carefully:

> I'm now ignoring recruiters who are begging me to interview for insanely well-paying openings. I have a decently well-paying job with good work-life balance, and colleagues who respect me.

So their utility function has terms other than $. So define "successful", or define the utility function more objectively. It is very clear that yours vs theirs are quite different.

KolenCh··on Workers are stuck in place because everyone is too afraid of a recession to quit
Define successful. I think you two had an opposite meaning of what successful is.
KolenCh··on GIL Become Optional in Python 3.13
To be a bit pedantic, I guess may be it is fairer to say python is the main glue language for scientific computing where DS and ML are part of. The “big bang” seems to have started from Numpy/Scipy and friends and cascading from there.

Companies like Facebook also have huge interests in speeding up python as they have a lot of stacks written in python.

← PreviousPage 3 of 13Next →