Thursday Throughput: Tensor Calculus 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

33 Responses

  1. Motoconomist says:

    ThTh6. I just do not understand folks who are worried about climate change and then fight to close nuclear plants…Report

    • Oscar Gordon in reply to Motoconomist says:

      Some of those plants need to shut down (end of life, no feasible way to breathe new life into them).

      What I don’t get are the people who make it nearly impossible to bring new plants &/or new nuclear tech online (that isn’t the holy grail of fusion).Report

      • You mean like Vogtle 3 and 4, where every stage has been plagued by major construction errors? Vogtle was licensed promptly, never dragged through court (other than during the Westinghouse bankruptcy), got $8B in federal government bond guarantees, and is still taking 14 years to construct. At a bit over $12B per GW of capacity. Next door in South Carolina, the Summer 2 and 3 construction was abandoned after some billions of dollars had been spent because it was cheaper to buy power from basically any other source than to finish construction.Report

        • Oscar Gordon in reply to Michael Cain says:

          Why are Vogtle and Summer over budget? Why are they plagued by construction issues? Could it be because the construction of a nuclear power plant is a rather specialized project, kind of like building a nuclear submarine, or a nuclear aircraft carrier, and if there are large time gaps between when such places or things are built, you lose the people with that critical skill set and experience. Thus the people trying to do it have to relearn all the lessons learned long ago, while also making the mistakes that go hand in hand with whatever changes have happened since then.

          I know you have this soapbox regarding the cost of nuclear, but that cost is not magical, it is an artifact of both the regulatory burden (even if entirely justified) and the loss of critical knowledge among the contracting community because few reactors get built and the time between them is long. It is a demon of our own making, because there is a strong fear bias against nuclear.

          Why is coal/nat gas/co-gen cheap? Because we build those regularly. It’s practically rote by now, and the designs are effectively modular. Even wind turbines get installed so fast because there is a whole industry that has matured to make sure they go up as fast as possible.

          If we want more nuclear, we need to suffer the expense of learning how to build more nuclear, OR we have to allow engineers to actually develop better nuclear (like we are with NuScale) that doesn’t have such a steep learning curve every damn time.Report

          • Oscar Gordon in reply to Oscar Gordon says:

            PS Anyone (not specifically you, Mike) who thinks fusion plants will be cheaper than fission is not in touch with reality. I know it’s a sci-fi staple, but even when we finally tame that process, those first few commercial scale plants will cost billions to build. And they will stay that way until we’ve got a few under our belt and the construction industry knows what they are doing.Report

          • Jaybird in reply to Oscar Gordon says:

            I hope that this is most of it.

            My worry is that we don’t know how to do stuff anymore.Report

              • Jaybird in reply to Michael Cain says:

                I feel a little bit better, I guess.

                But if I heard that Elon Musk was going to start nuclear reactors, my response would be mixed.

                Part would include “oh, good… maybe they’ll be completed.”

                But there would be other parts that would be asking “what the hell?”, among other questions.Report

              • InMD in reply to Jaybird says:

                Are you kidding? My response would be invest invest invest!Report

              • veronica d in reply to InMD says:

                Musk seems to have very high personal risk tolerance. After all, SpaceX is rather known for blowing shit up, and Musk being fine with this. After all, it’s testing. Tests fail. That’s why we test.

                That’s one sort of thing when the worst case scenario is a predictable explosion on a rocket launch pad. A fission reactor is quite a different level of risk.

                For example, there is this one reactor in Japan …

                “But surely Musk would behave differently with nuclear!”

                Um, I have zero trust in such a statement. Over time he’s demonstrated very poor impulse control. I mean, seriously. The guy is a bit of a nutcase. He honestly thinks cramming a bunch of cars in a narrow tunnel with very limited access is a good idea. It’s such a fundamental lack of “engineering common sense” combined with a stubborn refusal to consider the opinions of subject matter experts. Yikes.

                It is simply this, Musk’s crazy pants risk tolerance combined with the fact there would be no way to hold him properly accountable for disaster …

                No. Just no.

                The people who thrive in SV startup culture are not the same people we want working on nuclear power. “Move fast, break things” is a fine ethos for a provider of online entertainment. It is not the right kind of ethos for high risk public works.Report

              • Oscar Gordon in reply to veronica d says:

                Michael is right about the costs, but I wouldn’t get too stressed about things blowing up, unless the reactor type is one that has a real risk of blowing up.

                The thing about the reactor designs the nukes have been developing these past few decades is that they are designed to be explosive-accident-proof. Mostly by removing the things from the reactor cores that tend to rapidly expand when they get hot, and/or creating cores that will naturally slow down a reaction that is getting too hot.

                So it would really depend on what Musk wanted to play with, and who he hired to play with it (because Musk is not a nuke engineer and has no business pretending he is one).Report

              • Michael Cain in reply to InMD says:

                Now let’s go through why Oscar has a point, and why fission reactors are different than rockets.

                Lots of places will let you blow up rocket motors and even entire rockets. No state or private land owner will let you build a fission reactor without a federal license that basically says, “The NRC guarantees it won’t blow up.”

                You can do a lot of rocketry development and testing for $100M. An NRC license for a new large reactor class will run about an order of magnitude more. Not just do you have to design it down to the last nuts and bolts, you have to pay for the NRC to learn enough to verify that design. Then you can start building.

                Currently, all known licensed designs run on the order of $8B per GW of faceplate power (yes, including NuScale’s first module). So you need someone to pop for the billion-dollar license, and the $8B (assuming a GW plant). If SpaceX had needed $9B up front, and eight or more years before there was any revenue, they’d have failed*.

                If you don’t have a national government involved, new reactor designs aren’t ever going to be done. Eg, Obama guaranteed $8B in bonds for the Vogtle plants; DOE is providing the land and a billion dollars towards the first NuScale module. Reagan said the US is not in the reactor design business, and Clinton nailed that down. Trump’s administration took some quarter-assed steps to reverse that, but clearly aren’t serious.

                * The Georgia PSC, in order to keep Southern Co. from pulling the plug on Vogtle 3 and 4, allowed for the first time ever a utility to charge customers for the construction costs of a plant that may never generate a single watt.Report

          • I note that NuScale is up to $1.3B for a 60 MWe first module. They no longer talk about when or even if the prices will ever be lower than that. Some of the increase, interestingly, has nothing to do with nuclear per se. They decided that the politics of the feds confiscating water from the Snake River were too ugly, that the river is so overcommitted they couldn’t buy sufficient water rights, so went to air cooling. That increases the costs and decreases the thermal efficiency.

            It’s entirely possible that some day fission prices will come down. I assert that it’s far enough away that we can’t afford to wait. Either build some other low- or no-carbon tech, or pay really high prices for nuclear electricity.Report

  2. Brandon says:

    ThTh3: of course he was distracted. Houses were burning, thugs were in the streets. Afghanistan suffered too (apparently “home defense” does “paranoia planning” when they’re not busy looking for agents provocateur).
    Project Salus is the sort of thing DoD does for fun. When DoD is distracted by fires in the streets, not so much.Report

  3. veronica d says:

    Gosh, when I saw this was titled “Tensor Calculus Edition,” I was expecting to fully understand the Reimann curvature tensor just by reading it.

    Anyway, I find Tensor analysis deeply beautiful, despite the fact I have quite a meager understanding of the machinery. That said, the “geometric” view of GR to me seems profound. I’ve cracked the spine of the Misner/Thorne/Wheeler/Kaiser a few times, but I’ve never really made it past the first third, and that’s just the “easy track.” But still, I think I can kind of glimpse what is happening. It’s really cool.

    Part of me wishes that the Wheeler-esque geometric view of physics was true “deep down” (whatever that means). From my few brushes with QM, I find all the wave function stuff somehow unsatisfying.

    Of course, none of this has anything to do with how nature actually works. The world isn’t here to please us. But anyway, I’m just musing.Report

    • veronica d in reply to veronica d says:

      Just to add, my first statement was meant to be amusing and friendly, not snide. I hope it came across that way. If not, I apologize in advance.

      (The joke is that I find diff. geometry really hard to grasp, but it is something I would like to understand someday.)Report

      • I’ve never read Michael Spivak’s books on differential geometry, but if they’re as good as Calculus on Manifolds, they’re pretty good.

        Trivia: Spivak is also the guy that created the E, Em, Eir pronouns.Report

        • veronica d in reply to Mike Schilling says:

          Yeah, Calculus on Manifolds is pretty great. Honestly, though, I find it strangely easier to learn “physics math” from physics texts rather than math texts. I find the physics texts tend to play to my visual intuition more than pure math texts.

          Likewise I learn a lot of “logic/formal-systems” stuff from CompSci texts. In this case, it’s not my visual intuition, but just my experience writing software that lets me “grasp” the subject.

          I still prefer to approach algebra from a pure math perspective. I’m not sure why.

          And you just out-trans-knowledged me. Congrats!Report

    • Tenser, said the Tensor.
      Tenser, said the Tensor.
      Tension, apprehension,
      And dissension have begun.
      Report

  4. J_A says:

    Honestly, though, I find it strangely easier to learn “physics math” from physics texts rather than math texts. I find the physics texts tend to play to my visual intuition more than pure math texts.

    It is the same to me. I studied line and surface integrals in math and, though I passed the course, it made no sense to me. It was like Gauss was bored or something. The following term I studied the same concepts in Electromagnetic Theory and it was bloody obvious what those integrals meant in the real world 😇Report

    • veronica d in reply to J_A says:

      Yeah, that’s a perfect example. Moreover, that’s really how that math developed. Field theory emerged from trying to solve physical problems.

      As a total math nerd aside, I see some clear parallels between complex analysis and field theory, particularly the Cauchy Integral Theorem, and various similar things. I’m also aware that Hamilton discovered the Quaternions while trying to find something like complex numbers, but in 3 dimensions (so naturally he found 4 dimensions, because the gods are cruel). Anyway, I’ve always kind of assumed that Hamilton wanted to apply stuff like Cauchy to something like field equations, although I’ve never seen that spelled out.

      Modern math texts are often quite divorced from the historic context of how the math was developed. For example, modern Galois theory looks absolutely nothing like what Galois actually did. Most Galois theory texts will mention the historic context, but not explain it. For that you have to seek out specialized texts.

      This is fine, I guess. I suspect most students care more about how they can apply these things in their current work, and delving into outdated notation and ideas would take up valuable brain real estate without much payoff. Still, it interests me quite a lot.

      I think physics texts are somewhat better at this, but only somewhat.Report

    • veronica d in reply to J_A says:

      Oh and I love this:

      It was like Gauss was bored or something

      One can hardly quantify the amount of wonderful math we have nowadays because one afternoon either Guass or Euler were bored.Report

      • My own opinion is that almost all of the useful stuff would have been developed by someone else anyway. I became even more convinced of the time the prof asked me to stay after the graduate topology class. “Have you been reading old topology texts?” he asked. As the class was being taught using the Moore method, that would have been a serious violation of the rules. When I asked why, he told me that the proof I’d done in class was the way that particular theorem was proven up until 20 years earlier. It was correct, but a dead end. Eventually a different proof opened up a variety of new things. The thing was that for that proof there had been an “Aha!” moment for me when all the pieces fell together in my head.Report

        • veronica d in reply to Michael Cain says:

          You were actually not allowed to read other topology texts? That so weird.Report

          • That’s the Moore method. Interesting to experience once. At least at that time, that’s the only way UT-Austin ever taught the first graduate topology class.Report

            • veronica d in reply to Michael Cain says:

              Ah. I guess I can see a class like that being fun, at least for topology.

              I’m not sure I’d enjoy learning, say, complex analysis that way.Report

              • Note I said “interesting” rather than fun, and I intended that in hindsight. It certainly wasn’t fun that semester in Austin. The Moore method requires a very large investment in time and effort by the students, I was also taking algebra, which I was never strong in, graduate analysis taught as a fail-out class, and TA’ing for a demanding (but fair) professor. They didn’t fail me out, but they did chase me out. The recently upgraded operations research group was happy to take me.Report

  5. Kazzy says:

    ThTh4: Have we tried turning it off and turning it back on? Or maybe blowing into it?Report