Thursday Throughput: AAS Bumper Crop Edition

Michael Siegel

Michael Siegel is an astronomer living in Pennsylvania. He blogs at his own site, and has written a novel.

Related Post Roulette

10 Responses

  1. Murali says:

    A gallon is not that much. I easily consume a gallon of fluids a day. If I replace all of that with soy milk, I’d be drinking a lot of soymilk. But I don’t like soymilk (even though I like tofu, especially the firm kind). And while 5-10 soy burgers is a lot, that is because the actual soy content in these burgers is less than in firm tofu. to get some statistics, in a soy burger, there are 19g of protein. Let’s suppose all of this is soy protein. a 1/4 block of extra firm tofu contains 26g of soy protein. In order to get 95-190g of soy protein, I have to eat 1-2 blocks of extra firm tofu. Over two meals a day, I could easily see myself eating 1 block of extra firm tofu. Tofu shrinks when its fried so its not that filling. Just google tahu goreng. 1 tahu goreng is a block of deep fried tofu covered in peanut sauce. It is a meal (and sometimes only part of a meal). If I didn’t have peanut allergies, could I have tahu goreng (or gado gado or any other dish made from a block of firm tofu) for lunch and dinner? It’s not unimaginable. Some combination of the two can easily set you over the safe limit.Report

  2. Doctor Jay says:

    [ThTh13] I love this.

    I just wanted to say…

    I love this.Report

  3. veronica d says:

    If soy products could cause feminization, trust me, a certain segment of the population would be able to demonstrate it empirically.

    It doesn’t.Report

  4. dragonfrog says:

    [ThTh8] – I don’t think the conclusions in the article add up, and I think it comes down to using old data for the energy efficiency of modern networks.

    The only figure with an effective denominator I found in there was 1.6 kg CO2 / 0.5 hrs of video. Assuming 1 GB / hr of video, that’s 3.2 kg CO2 / GB. The Shift Project document they link to estimates 0.519 kg CO2 / kWh, so a bit over 6 kWh / GB of data transferred.

    So, where does that power / GB estimate come from?

    Googling around, I found a few papers from about 2010 – 2012 that had estimates in that range. (Possibly relevant – 2011 is when Netflix started their transition to streaming-only, splitting off the DVDs-by-mail business)

    I also found this from 2017

    This article derives criteria to identify accurate estimates over time and provides a new estimate of 0.06 kWh/GB for 2015. By retroactively applying our criteria to existing studies, we were able to determine that the electricity intensity of data transmission (core and fixed‐line access networks) has decreased by half approximately every 2 years since 2000 (for developed countries), a rate of change comparable to that found in the efficiency of computing more generally.

    Moore’s Law. It’s a hell of a thing. But what’s a couple orders of magnitude among friends?

    It takes the equivalent drive from the 3.9 miles in [ThTh8] (“you probably didn’t have to drive that far for your VHS fix”) to driving 70 yards (if you lived 2/3 of a block from the video store you probably walked).Report

  5. North says:

    ThTh8 and 9 go together well and I just wanna make a point of agreeing entirely with both points.
    I swear to God(ess?) that some of the loonier cohorts of the eco-left make me seriously wonder if there’s something to the socialcon shibboleth that if you get rid of organized religion then people will just organize new religions or religious-ize existing social structures.Report

  6. [ThTh1] I’d think the Sun-Earth-Moon system is also a three-body problem, since the Sun attracts the Moon twice as powerfully as the Earth does.Report

    • veronica d in reply to Mike Schilling says:

      Yep. However, the distance to the sun makes planet-moon systems generally tractable.Report

    • Oscar Gordon in reply to Mike Schilling says:

      Didn’t some one just get an AI, or a neural network, to solve a three body problem?Report

      • As with most things AI or neural net, sort of. Given a sufficient library of previously solved problems for training, the AI takes a new problem and builds a prediction of the trajectories. With lots of weaknesses: fundamentally, underneath, it’s a statistical thing; no one knows exactly what parameters the statistics are based on, or how; it is, apparently, an iterative process and there are no established convergence criteria.

        That said, I have started a real-time control project that has an image recognition component, and will no doubt be training some sort of AI/NN for that. Or using someone’s already trained open-source one. At least initially, the software just has to answer the (relatively common) question, “Is that a feral cat inside the target zone?”Report

      • veronica d in reply to Oscar Gordon says:

        I’m pretty sure it is formally unsolvable, at least in terms of analytic functions, although I should probably check that. I recall that Poincare argued that it was unsolvable, mostly based on topological considerations of phase space —

        Which leads to a really cool bit of math trivia. A lot of Poincare’s motivations to explore topology was in service to understanding the geometrical aspects of differential equations. Think of it this way, any system of N second-order ODEs can be transformed into a system of 2N first-order ODEs. (This is what happens when you move from classical Lagrangian dynamics to Hamiltonian dynamics.) So, if you think of those 2N variables as dimensions in a 2N dimensional space, and then if you think of all the dx/dt’s and dy/dt’s as describing a vector field in that space, then you can think of the various solutions to your equation as flows in that space. Families of solutions form hypersurfaces. Those hypersurfaces often have an interesting topology.

        (I assume you’ve at least seen “phase space” described during your engineering training.)

        See also the Liouville’s Theorem: https://en.wikipedia.org/wiki/Liouville%27s_theorem_(Hamiltonian)

        (You’ll get to use your knowledge of fluid dynamics to think about ODEs. Ain’t math grand!)

        Anyway, this was the kind of thinking that led to the conclusion that there was no analytic solutions. (Formally: “analytic” more or less means something can be solved as a power series.)

        I just checked. This according to the wiki page:

        On the other hand, in 1912 the Finnish mathematician Karl Fritiof Sundman proved that there exists a series solution in powers of t^1/3 for the 3-body problem.[6] This series converges for all real t, except for initial conditions corresponding to zero angular momentum. (In practice the latter restriction is insignificant since such initial conditions are rare, having Lebesgue measure zero.)

        That’s kind of cool. It’s not analytic, but it is a series.

        In any case, we can readily solve three body problems using numerical methods. In fact, because we can represent spheres as point-masses, and because we can usually represent oblate planets as point-masses plus a quadrapole moment, we can actually get pretty freaking accurate for a lot of problems. In the Hamiltonian form, the numerical methods can exploit the “symplectic geometry” of the problem to get wicked accuracy. Google “geometric integration” for more.

        (I like math because I get to use words like “symplectic” non-ironically.)Report