1,071 words · auto-generated from the episode video
30:23the fundamental resource is surface area, right? The fundamental resource is how how can I build this? And so I still want to optimize minimum surface area because I don't want the building cost to go up. >> Yes. >> Right. This the second thing that I was telling you about the Steiner tree is all of the all of the nodes happen at 120° but in normal biological networks it should hap sometimes it happens at 90°. Why is that happening? Well, if the thing that's coming out of my tube >> is really thin. >> Again, it's that same order parameter. If the thing that's coming out of my tube is really thin, then it's actually preferable for me to be 90 degrees away from the tube. On the other hand, if the
31:04branching is like thick thick thick, then I get to the 120 degrees. And that's what they're showing. They're showing that with these neurons, what what you can see is >> if you've got these like tiny little um synapses that come out of, let's say, an axon or a dendrite spine, right? >> Those those tiny little synapses are really thin compared to the thickness of the dendrite. And so what I really want to do in that situation is come out at 90°. I don't want to go at 120 >> because that's the thing that minimizes surface area. >> It's it's funny if you think about suburban neighborhoods >> and or the design like the uh the design
31:46of the road networks. >> Yeah. >> You know, this looks like a culde-sac in a subdivision. >> It kind of does. >> You know, because you have the your main Yeah. two-lane road, then you have these offshoots that are narrower >> right off the main road at 90° angles. >> Yeah. [laughter] >> So, we've also mimicked that obviously and for, you know, Yeah. It's it's the whole image of you see people with traffic going really fast in cities and then people like map the like blood vessels and there's the similarities in these structures. >> Yeah. I mean, their structure >> all the way down, you know, all the way down. And so, I just thought this this was really cool. Um, in terms of the computational methodology, I thought the way that they did it was very cool. This is the minurf netwa. It was implemented
32:28in Mathematica. This is the new code that they use to with finite element methods to solve for surface variation problems. They actually have it available on GitHub. So, you can use it. Yes. And what what they really did was they inputed the connectivity of real maps and then they said okay what is they asked the algorithm to generate the optimal surface geometry. So it makes up a bunch of meshes >> of polygonal meshes and then tries to figure out what the optimal surface geometry is such that it minimizes surface area and they could reproduce what the biological networks were doing. So there's sort of this generative model that's based on mathematics but it is it
33:09is not it is generating net new yeah >> structures >> and those net new structures that were generated based on this underlying mathematical framework when then held up next to our observed biological uh uh comparison point >> the statistics match >> it's just it's it's like it's very nice >> so I just thought this was a very cool paper right and it's it's kind of a unification of biological physics across so many different types of organisms, fungi plants vessels brains coral all of them have the same principle, right? The structure is shaped by a universal physical constraint of minimizing the amount of membrane that I
33:49need to use to create a network. >> And because it kind of makes sense, too, because if you think, you know, um the larger your surface area, the more energy you need to make it >> to make the things and sustain the things. So it's like again if you're trying to minimize that. >> Yeah. Yeah. And it's not not just the energy in making the thing, it's the material cost. You're using more carbon atoms, right? You're using more hydrogen atoms which you could use for something else, right? So you want to optimize at all times. Um there are a bunch of really cool applications that I just want to go over. One of them is 3D bioprinting. So what you can do is design scaffolds using this algorithm to create physically correct microvasculare. If you want to create an artificial
34:30tissue and you want to improve blood flow in medicine. Yes. Right. >> So now we now basically have a pathway to 3D print things that can be integrated into our system that would map structurally >> more onto the way our biological system is is exists to to optimize for the potential for the body rejecting it and all these other >> Exactly. and and the body not having to work harder than it should to push blood through this artificial system, right? Um, another one could be hardware design like if you want to use these surface optimization principles to make 3D chip architectures, right? Like neuromorphic chips, you can you can now use these
35:12principles to do that. Yes. >> Right. And finally, what I want to touch on is like kind of the philosophy of science, right? There's string theory has gotten a lot of heat and it shows in this XKCD comic string theory summarized the string theorist said I just had an awesome idea suppose all matter and energy is made of tiny vibrating strings and his friend says okay what would that imply and the string theorist says [laughter] I don't know >> it's string theory has been the butt of so many jokes in physics and mathematics in the idea of funding fundamental science and I think that's quite dangerous. Now there there's stuff to be said about perhaps the scale at which we
35:54funded string theory versus something else. But the idea of funding something like string theory should not be totally looked down upon just because oh it doesn't have any real world applications because at the end of the day if we go into fundamental research >> yes