MIMO and nonlinear control theories are probably some of the hardest topics in all of engineering. SpaceX control system also has to compensate for the fuel moving inside their rockets so the control algorithms probably involve some kind of fast numerical fluid simulation.
Another interesting thing SpaceX is doing is to use consumer-grade chips in triple redundancy configurations instead of using $100,000+ radiation-hardened aerospace/defense grade chips.
For everyone reading neither the article nor the paper:
- both show neural networks can learn the game of life just fine
- the finding is that to learn the rules reliably the networks need to be very over-parameterised (e.g. many times larger than the minimal size needed for hand-crafted weights to perfectly solve the problem)
This is not really a new result nor a surprising one, nor does it say anything about the kinds of functions a neural network can represent.
It's an attempt to understand an existing observation: once we have trained a large overparameterized neural network we can often compress it to a smaller one with very little loss. So why can't we learn the smaller one directly?
One of the theories referred to in the article and paper is the lottery hypothesis, which states that a large network is a superposition of many small networks and the larger you are the more likely at least one of those gets a "lucky" set of weights and converges quickly to the right solution. There is already interesting evidence for this.
Apologies; I confused BPC-157 with TB-500, and made a horrible amalgamation.
What do you want to know? Do you have some specific goal you want achieved or curiosity that you would like satisfied?
It's been a bit since I've been involved with these things. Most of it is underground, but I can give you a quick rundown.
BPC-157 and TB-500 are regenerative peptides. BPC-157 seems to be more "global" and neuro-involved throughout the body, while TB-500 is more local and structural (joints, tendons, etc.). BPC-157 also has a (prolonged) effect in some that negates the effects of amphetamines. Subcutaneous injections of BPC-157 have helped get rid of my recurring ganglion cysts and golfer's elbow.
Semax would fall under "slavic nootropics," along with Selank, and if I remember correctly Epithalon. All have "sub-versions" of varying efficacy. F.e. all have "N-Acetyl" and "N-Acetyl Amidate" versions. NASA would be the shortened version of "N-Acetyl Semax Amidate" -- which in my experience is the "strongest." With NASA, while I was injecting it subcutaneously it brought a sort of structure to my mind I hadn't had since I was a child. It's like feeling everything is falling into place, and a loss of the feeling of helplessness.
If you've ever used noopept, it's like that except with more real and long-lasting effects. I would liken it to bromantane, too.
If not, it's difficult to explain what they are, because they're such a different class of drugs that there's no reference point to base their effects off of. Imagine that you have a drug, but instead of giving you a few hours of a noticeable "high" or "low," instead you get a small, but perceptible shift in how you view the world, and how you filter all the information coming in. Like a micro-micro-micro dose of LSD. No high, no impairment, just a beneficial "shift" in your perception that lasts for an indeterminate, but long time.
A few of my friends were career-researchers and likened these effects to be genomic (subtly altering the expression of genes all around the body) rather than physical (that is, simple physical reactions like consuming more electrolytes would cause you to hold more water, and become bloated, because electrolytes attract and "hold" water; or how drunkenness is simply a temporary shift in the delicate GABA/glutamate balance in your brain). The purely "physical" drugs require constant re-dosing to be effective, while the more genomic ones (such as peptides) can have long-lasting effects even after they've been ceased.
(Unrelated but this is why I don't believe singularities exist in the universe - we don't know enough about quark degeneracy pressure to know if it's actually possible for a star to collapse - it's possible the quark pressure keeps the matter from compressing.)
Every proton would have the energy of a baseball. It's bananas. Granted, space is really empty, but it's not that empty, somewhere around a hundred atoms per cubic meter. Your ship might look like a very, very long shooting star. Probably dialing the speed down a touch would be worth it for whatever your shielding material is, but who knows? We're talking miracle engines here. I think in the "Valkyrie" ship-on-a-string concept, they had some sort of magnetic thingummy that generated more power as more stuff smacked into it, then as they decelerated they let out this sort of gas mist to go ahead of the ship and smack into things.
Live poker is still quite soft. While you do have to be better compared to years past, it's also far easier to get better because you can study solvers. I used to be a "good but not great" online player, but the extent to which you could truly understand hands was limited due to the limitations of the software at the time (mostly equity calculators that could not truly calculate EV). That changed with the release of Piosolver in 2015 that lets you calculate the EV of various strategies and give you the "correct" answer assuming various assumptions. Those assumptions are always wrong but it's still informative in the "all models are wrong, some are useful". I used to "plateau" in my undertsanding of the game but you no longer plateau because the game can now be understood at a more complete level.
Live poker will always be soft because there will always be plenty of people there to have fun and gamble no matter how much software exists. Furthermore, only a slim minority of players even actively trying to win actually put serious effort into studying (see also, dan luu's post on how it's relatively easy to reach the top 5% of any endeavour, even if very difficult to reach the top 1%).
I can't help but use the opportunity of poker on the front page of Hacker News to note that I have a strategy blog and training app on this exact topic:
www.livepokertheory.com
There's a lot of beginner to advanced strategy articles on applying practical solver outputs to live poker games as well as a preflop training based on solver generated preflop charts. Poker is currently my only significant source of income and I "eat my own dogfood" on this site so I'm very incentivized for it to be good!
I think it is possible to build up a rational case for an objective universe starting with just perception. Firstly if the only thing that exists is conscious awareness, where does the informational content of the world you perceive come from? It doesn't come from your awareness, because you are not aware of it until you perceive it. You can say it comes from the subconscious, but the subconscious is not part of your conscious awareness. It's external to it, in the same way that your hand is external to your conscious awareness. There has to be an origin for perceptions that is external to conscious awareness of those perceptions.
From there we observe that these perceptions are of a consistent and persistent form, so it’s rational to conclude that they have an origin in a consistent and persistent source. From there, and taking into account our ability to test our perceptions through action, we can build up knowledge about the world of our experiences.
As for trusting logic and rationality, does it give consistent and useful results? Test it and see if it continues to work reliably over time. If applying logic provides random, contradictory or unreliable results that’s a problem, but maybe you can correct that by modifying how you reason about things and trying again. That’s learning. So I think we do have the cognitive tools we need to build up a robust account of reality starting from base perception.
We didn’t start off human society and civilisation with scientific laws as our founding axioms. We inferred them from sense data, including the process of physically testing our ideas in the world.
It is said that a visitor once came to the home of Nobel Prize–winning physicist Niels Bohr and, having noticed a horseshoe hung above the entrance, asked incredulously if the professor believed horseshoes brought good luck. “No,” Bohr replied, “but I am told that they bring luck even to those who do not believe in them.”
In the US, where Twitter & Facebook are dominant, the current consensus in the public mind is that political polarization and radicalization are driven by the social media algorithms. However, I have always felt that this explanation was lacking. Here in Brazil we have many of the same problems but the dominant social media are Whatsapp group chats, which have no algorithms whatsoever (other than invisible spam filters). I think Yishan is hitting the nail on the head by focusing the discussion on user behavior instead of on the content itself.
I live in Vanuatu and is it very far to the west side of the Pacific yet the almost southern most island of Vanuatu, Aneityum, has stories of what they called the "Yellow People" that were on the island before they, Melanesians, arrived from northern islands. These people on the island were excellent stone carvers and could make stone walls which the current locals admit they never learned from the "yellow people". Old engravings exist still of these original people that to me sound like those may have come from the east, South America. I don't have photos though, this is a story I just heard recently from family members of that island.
Another interesting thing SpaceX is doing is to use consumer-grade chips in triple redundancy configurations instead of using $100,000+ radiation-hardened aerospace/defense grade chips.