Hacker News new | past | comments | ask | show | jobs | submit login
Granular convection (wikipedia.org)
46 points by glcheetham on Dec 10, 2019 | hide | past | favorite | 22 comments

I remember a journal club where we read a paper in which they demonstrated the inverse brazil nut effect (http://large.stanford.edu/courses/2007/ph210/spector2/). The scientists built a physical nut shaker- like, a huge tube that they bounced. that's when I realized I had picked the wrong field (theoretical biology instead of experimental physics).

What do you do in theoretical biology?

Nothing anymore- I became disillusioned with modern scientific research, especially biology, because I believe that most of what is published is simply false or wrong (it's much harder to publish false crap in experimental physics because everybody knows more math, is better at analyzing experiment design, and the systems are "simpler"). Now I work on software engineering related to neural network performance (focusing on things that have worked well for years).

That said, things I've worked on include: classifiers to identify genes in DNA, molecular dynamics simulations of proteins and nucleic acids, methods to discover new drugs, analysis of protein function from an evolutionary perspective, protein design.

> I became disillusioned with modern scientific research, especially biology, because I believe that most of what is published is simply false or wrong

I’d love to hear more about that. Do you have any good examples?

Sure, there was a paper about yeast. They created a collection of 6000 clones- each one had a "single gene deletion" and nothing else. THey were able to study each clone and explain whether the gene deletion was fatal (IE, if you delete this gene that is essential for yeast to grow, the clones all die).

I was tasked to analyze the paper results, specifically a table that listed "all the genes we newly discovered that, when removed, are fatal".

Since I'm interested in overlapping genes, I took the list of new genes and intersected it with all the other known genes, and noticed that for each gene they newly discovered had a fatal outcome, it intersected a gene that was already known to be necessary for yeast to live (housekeeping genes, essential enzymes, dna checkpoint repair, etc).

The most likely conclusion from this is that when they deleted the gene, they also truncated or deleted the other gene, which abolished the function of the known-necessary gene. Therefore, they didn't discover anything new- they just disabled things that we already knew broke yeast, by accident, by ignoring the fact that genes overlapped.

I showed this to my advisor, who said "Good catch, why don't you write them a letter?" So I sent them a letter showing my results, never got a response back. They published a paper a year later, totally unapologetically, saying how they had "discovered interesting interactions between overlapping genes"...

You can also look back at the original human genome sequencing papers, they were massively overestimating the accuracy of the assembly and the representation of the genome (compared to the wider population). It took decades of additional large-scale sequencing to verify this, but it still bothers me just how overly egotistical and self-confident the folks who worked on the human genome project were.

Molecular biology and biochemistry has a pretty major replication crisis of its own. (It's not just social psychology - even classically "hard" fields like organic chemistry have reached the point where practioners are accustomed to never being able to get a reaction in a new paper to work as well as they say it does)

Farmers know the rocks come to the suface over the winter.

interesting, I suppose if the top 6 inches of a field freezes then it slightly expands in volume and when the soil thaws the smaller soil particles would then have an opportunity to fall beneath the rock... not sure if this is dependent on the top or bottom thawing first though...

In many places, the frost line is more like 3 to 6 feet, than 6 inches.

This is a recurring problem in the animal feed industry. Easiest explanation for the phenomena: the small particles fall through the gaps gaps between the large particles.

My first instinct would be that this is due to entropy rather than minimizing energy, but the article doesn't mention that hypothesis. Is it easy to rule this out?

Isn't "minimizing energy" a definition for entropy?

My idea is that, when shaking, all pieces try to move. Big pieces can only go upwards, while small pieces can go sideways (and take the space that was previously taken by a big piece). This ends up with the effect of big pieces floating to the top

> Isn't "minimizing energy" a definition for entropy?

No. If you have a system with a well-defined energy/entropy exchange rate (temperature) you can use it to put entropy in units of energy, add it to your energy term, and model the combined system behavior on the basis of this "free energy." The free energy is not a fundamental quantity, though, and it doesn't define entropy. That's why the first two laws of thermodynamics are "Energy is Conserved" and "Entropy Increases," not "Free Energy decreases."

> Isn't "minimizing energy" a definition for entropy?

No. I’m not really sure how you’re trying to define entropy, but you’ll probably have quite a bit of difficulty with endothermic processes if you try to think about entropy like this. For example, if you sprinkle sodium acetate in water, the sodium acetate dissolves, and the solution absorbs heat in the process, thus increasing the energy of the system.

In general, entropy increases but energy just moves around.

I use this to sort out favorite flavors and types of snack mixes all the time. Irritates my wife a little when I've eaten all of one type and left the rest.

It's about maximizing density, which corresponds roughly to 1) minimizing energy by making the aggregate have as low of a center-of-gravity as possible, and 2) minimizing entropy of the arrangements of the components. Often energy and entropy minimization coincide quite closely but don't exactly share the same minimum nor optimization path.

the Brazil nut effect results in less randomness in the system as larger items rise to the top and smaller ones descend

The system isn’t closed - you’re exchanging energy and entropy with the shaker.

That makes sense, thanks for resolving it for me.

Here I was this morning just thinking about this - weird how that happens.

My wife got this gift recently, that at first looks like a snowglobe, but actually has fine sand and seashells inside. You can kind of shake it or rotate it about, and the seashells that "rise" to the top on the sand change as you do it.

For some reason, that got me thinking about this topic. Then I see it on HN...

Maybe OP bought pack of assorted nuts for Christmas and wondered the same.

this also has application in mixed spice grinders. you can shake it right side up or upside down to get the desired effect

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact