HNHacker News
TopNewBestAskShowJobs

szvsw

898 karma · joined May 24, 2024

email: $hn_handle [at] mit [dot] edu

meet.hn/city/us-Boston

submissionscomments
szvsw··on Satellite powered estimation of global solar potential
Overhangs are considered exterior shading in the industry/practice/academia. Any obstruction that prevents solar gains by blocking radiation from entering the window falls within the general category of external shading, whether that’s a fancy high tech actuated shading system, a grille, a simple awning, a structural overhang, vertical fins, etc.
szvsw··on Satellite powered estimation of global solar potential
Oh I’m totally with you! There is a long and storied history of passive design strategies, and exterior shading is one of the oldest ones out there!

But what I stated is plainly true, and many people simply don’t want exterior shades (or just don’t think about it).

The point I was trying to make was just that there are thermal implications to the orientation, and you should think those through (using thermal simulations can help detect these issues) and come up with appropriate strategies (thermal simulations can help validate them). Maybe you don’t want shades, but you would be okay with emissivity coatings for your windows. Or maybe you just want to position windows on both sides of the home with continuous air volumes connecting them to promote natural ventilation. Maybe you can take advantage of thermal mass. The list goes on…

szvsw··on Satellite powered estimation of global solar potential
Yeah we refer to this as the heat loss form factor of the building, which is determined largely by the surface area to volume ratio (so you have a square-cube relationship at work) as well as the the number of floors in conjunction with the roof area. With more floors, the heat transfer through the roof (which can be substantial, as mentioned by a sibling comment) is less significant for the same roof area (after normalizing for the gross floor area).

Same goes for the slab/foundations (which can also have substantial thermal transfer in many contexts).

szvsw··on Satellite powered estimation of global solar potential
Yeah, don’t over look the fact that the thermal demand from space conditioning homes is way higher on a per capita basis in a suburban context compared to an urban context with multi-family housing/apartments etc. There’s just way more air volume to condition per person, generally more inefficient systems, etc.
szvsw··on Satellite powered estimation of global solar potential
There’s not always a lot of freedom to control roof angles like that - it might eg be directly determined by the orientation of the street - and even if there is, it might come into conflict with other thermal considerations. For instance, perhaps orienting the building such that the roof midline is E/W and the surface is due south results in more windows pointed due south, which in turn drives much more solar gain on the interior and greater cooling loads as a result - maybe the increased solar output outweighs those gains, maybe it doesn’t. You have to run some thermal sims to check. On the other hand, you will have more solar gains in the winter, which will decrease your heating demand.

So it’s not universally applicable - but it is absolutely true that it will increase solar output!

szvsw··on Satellite powered estimation of global solar potential
One advantage of distributed solar is that it can at least come online right away and when installed with a battery, can get a home pretty close to being fully self-sufficient (depending on the climate/heating system), whereas the generally much more efficient solar pv power facilities have to contend with backlogs in connecting to the grid, insufficient grid capacity, etc.

But yes, distributed solar will not be the general solution to decarbonizing our energy systems as a whole. Does serve a meaningful role though and there is no reason to not do both.

szvsw··on Programmers want flow. When programming, light turns red
Part of being on autopilot and in flow for many people might be thinking about/solving/planning what you will do next while you are writing code for the last item…

I find this especially true while working on data science tasks - slicing and dicing data, visualizing, fitting initial models, etc - things where the larger more complex data generation pipelines are already taken care of and the data (or temp data/subsets of data) are readily available and you are free to rely on the improvisational skills you have developed working on similar data science tasks countless times over.

Also feel something similar when working on frontend wireframing/laying out and styling dashboards with no reference design.

Anyways, point is, it’s definitely feasible to enter into a flow state where you mind is a few steps ahead of your fingers and is actively solving problems that your fingers catch up to.

szvsw··on Arc Prize 2024 Winners and Technical Report
Thanks for the response! I was trying to allude to what you are describing with the bit (ha) I mentioned about higher order thinking but you obviously articulated it much more effectively.

I guess I’m not sure it’s obvious where the right line to draw the boundary for “intentional human thought” is? Surely there is a lot of cognition and representation going on at extraordinary speeds that exist in some hazy border region between instinct/reflex/subconscious and conscious thought. Still, having said that, I do see what you are saying about trying to compare the complexity of the formal path to the solution, or at least what the human thinks their formal path was.

I’m generally of the mind (also, ha) that we won’t really ever be able to quantify any of this in a meaningful way in the short term and if anything which qualifies as AGI does emerge, it might only be something which is an “I know it when I see it” kind of evaluation…

Where are you getting 300W from? The body only dumps 100W of heat at rest and uses like 300-400W during moderate physical activity, so I’m a little confused about what you are describing there. The typical estimates I’ve seen are like 20W or so for the brain.

Edit: I should also say that what you describe does seem like a great way to compare solutions between computational systems currently being developed and a good one to use to try to push development forward; it just seems quixotic to try to be able to use it comparatively with human cognition or to be able to meaningfully use it to define where AGI is, which might not be what you were advocating for at all, in which case, sorry for misinterpreting!

szvsw··on Arc Prize 2024 Winners and Technical Report
> That said, I think there should be consideration via information thermodynamics: even with TTT these program-generating systems are using an enormous amount of bits compared to a human mind, a tiny portion of which solves ARC quickly and easily using causality-first principles of reasoning.

This isn’t my area of expertise, but it seems plausible to me that what you said is completely erroneous or at the very least completely unverifiable at this point in time. How do you quantify how many bits it takes a human mind to solve one of the ARC problems?

That seems likely beyond the level of insight we have into the structure of cognition and information storage etc etc in wetware. I could of course be wrong and would love to be corrected if so! You mentioned a “tiny portion” of the human mind, but (as far as I’m aware), any given “small” part of human cognition still involves huge amounts of complexity and compute.

Maybe you are saying that the high level decision making a human goes through when solving can be represented with a relatively small number of pieces of information/logical operations (as opposed to a much lower level notion closer to the wetware of the quantity of information) but then it seems unfair to compare to the low level equivalent (weights & biases, FLOPs etc) in the ML system when there may be higher order equivalents.

I do appreciate the general notion of wanting to normalize against something though, and some notion of information seems like a reasonable choice, but practically out of our reach. Maybe something like peak power or total energy consumption would be a more reasonable choice, which we can at least get a lower and upper bounds on in the human case (metabolic rates are pretty well studied, and even if we don’t have a good idea of how much energy is involved in completing cognitive tasks we can at least get bounds for running the entire system in that period of time) and close to a precise value in the ML case.

szvsw··on Diátaxis – A systematic approach to technical documentation authoring
I think this idea is so common now that it is almost the standard for technical documentation at this point even if people don’t know it by that name.
szvsw··on We can mine asteroids for space food
To be fair, people will still be people… if the sludge/protein mash can be mildly upgraded to diversity/quality comparable to MREs (not that I’ve ever had one) sure, but for long term space flight it seems plausible that the psychological/morale detriments of eating sludge every day (or any single meal for that matter) would be significant.
szvsw··on NeuralDEM – Real-Time Simulation of Industrial Particulate Flows
Take a look at the paper I linked, specifically section 4, which illustrates finding force laws to match the desired dynamics of a real physical material (in this case, Delrin, including elastic and plastic deformation.

I guarantee you will like this paper!

szvsw··on NeuralDEM – Real-Time Simulation of Industrial Particulate Flows
Okay, fair, I was using fluid loosely (and inaccurately) to mean both granular and fluid behavior. But there’s nothing inherently incompatible between fluid dynamics and the discrete element method as far as I am aware, just like there is nothing inherently incompatible with solids. Sure SPH or LBM or FVM are the more traditional choices for fluids and computationally more tractable in most cases, but they aren’t necessarily “more right.”

Awesome paper on how powerful particle based methods can be:

https://www.sciencedirect.com/science/article/pii/S187775032...

And a fun image of a DEM solid model of fracture:

http://www.cba.mit.edu/media/DEM/index.html

szvsw··on NeuralDEM – Real-Time Simulation of Industrial Particulate Flows
Interesting. I wonder what parts of this approach could be adapted to DEM models of solids. For those unaware - even though DEM is naturally chosen for fluids, depending on how you configure the force laws between particles you can easily model solids as well, where each particle is essentially a chunk of material. There are then some interesting choices to make about (a) what kind of lattice you set the initial particles up in, and (b) how you tune the force flaws to get the macroscopic properties you want around stiffness, etc and (c) if you use multiple “types” of particles forming a composite etc.
szvsw··on Large-Scale Dimension Reduction with Both Global and Local Structure (2021) [pdf]
I agree with you that it is correct from a logical perspective; I’m talking more about a narrative perspective. Any paper benefits from having at least one or two non-toy/bon-benchmark problems IMO.
szvsw··on Large-Scale Dimension Reduction with Both Global and Local Structure (2021) [pdf]
Very cool paper, code in the repo looks good and easy to use as well. After a quick skim of the paper, I feel like it suffers from a pretty common flaw (one which my PI often points out to me in my work, so I guess I’m just extra attuned to it right now): the authors make a pretty convincing argument (to me at least, but I’m more of an applied ML than theoretical ML person, so grain of salt) from a mathematical/methodological perspective that PaCMAP is better than common popular DR algorithms, and has various desirable properties in terms of simultaneous global/local scale preservation etc, but they more or less accept it as a given that we need better DR algorithms and that being better than the existing methods makes the work interesting, while failing to really convincingly illustrate actual use-cases where PaCMAP unlocks some sort of insight or delivers some sort of meaningful result that t-SNE and friends could not do.

I think doing so would be especially important in a paper on DR techniques which are already so fraught in how they are deployed (often with little thought) in many applied contexts, and when so much of their putative utility comes from their interaction with human visual perception. I would have loved to see some discussion of actual engineering use cases where PaCMAP proves more useful than t-SNE - I’m sure there are many! Really just nitpicking from me though, will probably try it out on my own cases in the next few days.

szvsw··on The Analog Thing: Analog Computing for the Future
There’s no better introduction to signals and systems than a modular synthesizer IMO - the combination of tactility and audibility for multi-sensory learning is so great at building intuition - and more importantly, excitement! - for signal processing.

This looks like a cool project in the same spirit!

szvsw··on Numpyro: Probabilistic programming with NumPy powered by Jax
I would say that, at least for me, PyMC’s main advantage was in DX. I just found model construction much more straightforward and better aligned with how I wanted to assemble the model.

This textbook/walkthrough is great: https://a.co/d/9dWXDTK

szvsw··on MIT engineers make converting CO2 into useful products more practical
There are plenty of papers which look at precisely the economic and carbon tradeoffs of these sorts of processes. It all just depends on where you are. When you have a clean grid and a cheap grid, these methods become pretty feasible.

Take a look at De Luna et al, Science 364 2019 [1]

[1] https://www.science.org/doi/10.1126/science.aav3506

szvsw··on Model Predictive Control in the Browser with WebAssembly
Another thing to keep in mind is that having AI/ML surrogates which can evaluate expensive functions faster can also be integrated as an information source in model predictive control algorithms.
szvsw··on VersaTiles – a complete FLOSS map stack
Only mildly related, but has any one else been using Martin [0] for tile serving from Postgres/PostGis? It’s been a huge benefit to me and probably my favorite open source map tool right now. Martin + Deck.GL is such an awesome combo.

[0] https://maplibre.org/martin/introduction.html

szvsw··on Defining Statistical Models in Jax?
> make Bayesian inference more manageable

Discovering PyMC and the excellent accompanying textbook was game changing for me! Being able to write full hierarchical models in a handful of lines of code hooked up to pandas data frames already is so wonderful.

The more tools for this the better!

szvsw··on Is the attack helicopter dead?
Obviously this is just a binomial distribution, but another thing to consider I suppose would be if all trials are performed sequentially or simultaneously. If performed sequentially, on the one hand, you have a non-zero chance of not needing to expend the subsequent trials; on the other hand, it seems reasonable to think there might be a degraded (or increased!) probability for each sequential trial. If conducted simultaneous, similarly, it seems reasonable to think that that the individual chance of success is higher due to saturation of one form or another, but you are also guaranteed to expend all resources.

Point is just that it seems a little silly to try to reductively do these calculations - seems meaningless to try to compare without more information…

szvsw··on Ventilation and Temperature Control (1999) [pdf]
Haha, yes I know! You also have airspeed, perspiration, metabolic rate, clothing levels, and transients (eg moving from a hot space to a cold space) which all affect the phenomenological experience- but we do consider all of these effects to be wrapped up within the “thermal comfort” term of art.

See the CBE Thermal Comfort Tool to interactively determine comfort bands in relation to various ASHRAE standards for computing thermal comfort with all of these affecting parameters:

https://comfort.cbe.berkeley.edu/

szvsw··on Is AWS S3 having an outage?
Yeah the account is from like 2019 or so. But still, I’m spinning everything up with Terraform or AWS Copilot, so it is still an active choice on my end to put it in us-east-1.
szvsw··on Is the attack helicopter dead?
Am I the only person who is shocked to realize that this is the same Perun who does strategy gaming YouTube tutorials on a related channel!? I had seen a bunch of his Terra Invicta content… had no idea he also did actual military analysis? Is that what he does professionally in some capacity or is he just an amateur interested in this stuff with a fairly well researched YouTube channel?
szvsw··on Is AWS S3 having an outage?
us-east-1 gang rise up!!

For once I can feel mild pleasure at seeing the tables turned…

To be honest I don’t know why all my projects are always in USE1, I guess it’s just because that’s where we have always had them for my lab so I’ve stuck with it for no good reason…

szvsw··on Can you get root with only a cigarette lighter?
There are some great scenes in Rebels of the Neon God [1992] by Tsai Ming-Liang (Taiwanese filmmaker) where the main characters steal the main pcbs from some arcade machines and try to resell them to the arcade owner lol. Wonderful film, recommend it - some great scenes in those arcades.
szvsw··on Ventilation and Temperature Control (1999) [pdf]
Thermal comfort is still the number one thing. Visual comfort (primarily glare, but also the frequency spectrum of the light/overall light levels matter too) and ventilation do matter too though. Everybody knows what it’s like to be in a crowded lecture hall with poor ventilation and high CO2 levels leading to drowsiness!

There’s been a big boom in research related to dispersal of things (CO2, biological materials etc) lately though due to the increasing accessibility of CFD to building scientists + COVID.

szvsw··on CoRncrete: A corn starch based building material
Makes sense!
← PreviousPage 5 of 10Next →