It’s good to have something like this. Last time I worked with libav the only available good documentation was the official examples and the ffmpeg (the tool) source code itself, along with the comments in the headers.
1,220 karma · joined November 17, 2017
The graph in the article can be seen as a factor graph. VSLAM systems usually have a (kind of bipartite) factor graph with vertices/variables that are either keyframes or ‘good’ features, with edges/factors between features and the frames that see them and between adjacent frames; each of these edges are the factors in the graph. This structure results in very large but sparse graphs, and there are factor graph optimization libraries that take advantage of this sparsity (e.g. g2o or GTSAM.) These libraries also use specialized optimization techniques for some of the nonlinear manifolds (e.g. SO(3)) that arise in SLAM problems.
This is not a bug, it's a part of how Git fundamentally works. If you want to mitigate it you have to sign your commits. GitHub could only attribute commits in the UI if they're signed, but I suspect that this is considered too much friction to enable.
He also has a ton of other really great talks on controls engineering on his main channel.
> A vaccine is also an attenuated version of the virus so if becoming immune from the virus doesn't last long it won't last long from a vaccine.
A vaccine can be an attenuated virus, or a piece of mRNA or DNA, or a distinctive viral protein. Which type of vaccine you use+the adjuvants that go with it can result in very different immune responses and efficacy over time.
> This is false, immunity is strong and long-lasting.
Immunity to COVID-19 varies a lot person to person and is poorly understood. It does however appear that many people can be reinfected (especially people who had mild symptoms the first time.)
But we’ve had things like this for a long time. The police have acted like they do for generations, we had concentration camps for Japanese people during World War II, and we’ve always done a variety of other reprehensible things (propping up brutal dictators, destroying native civilizations, institutional racism of every possible flavor.) Frankly, I’m shocked that our word ever meant something.
They aren’t necessarily to be blamed for that. This blog post isn’t really assigning blame to white homeowners.
Trotting out a nebulous, totally unsourced claim and then implying that anyone who questions you is wrong beyond help is… not exactly a convincing tactic.
What information are the people angry about the current situation missing?
> It's not about the reality of the situation; it seems like more of a tribal thing.
This seems needlessly dismissive. The “reality of the situation” is that police brutality needlessly kills people frequently, most of the people who are killed are black, the police involved are rarely held accountable, and nothing has seemed to change for years.
In the US policing doesn’t even make it in the top 20 most dangerous jobs by fatalities [1] or injuries [2].
[1]: https://advisorsmith.com/data/most-dangerous-jobs/ (original source is BLS; this source just did the per-worker math) [2]: https://www.bls.gov/iif/oshwc/osh/os/snr02_00_2018.xlsx (2018 is the latest data available from BLS)
For students however I think that Java is probably the better call (fewer footguns, more resources like the AP curriculum.)
You have to be ignorant or deliberately acting in bad faith to say that signatures are an issue, because about 10 minutes of research will show you that states which do vote by mail get around these issues pretty easily.
Vote-by-mail is cheaper (no polling places to run), increases voter turnout (which is why it’s universally opposed by the Republican party), and actually reduces fraud risk because it leaves a paper trail and removes the opportunity for voter intimidation at polling places.
I prefer C++ to Java, but I think there’s absolutely a niche in the US educational market because the AP Computer Science A test uses Java. It could be pretty nice to have a Java-compatible MCU for the high school students I mentor because that’s the language they know, and it would make some cool hardware projects more accessible.
I talked to a couple of friends about this, and about 1/2 had similar course offerings.
On another note: the FIRST Robotics Competition’s official support library (WPILib) includes a number of advanced control features (trajectory generator, extended/unscented/“regular” Kalman filters, LQR) that were written almost entirely by students. I was one of the students involved in this work, and I can say with certainty that students can get a tremendous amount out of the program.
I avoid integral control because it’s less effective and much harder to deal with than good feedforward. Most of the mechanisms we use (in the FIRST Robotics context) have a feedforward model and good system-identification tools so this works out well.
It’s also certainly true that a LQR won’t help you if you have no a priori knowledge of your system, but for many of the mechanisms we need to control this isn’t a problem.
I’m curious, do you mean that there will be more black swan events due to hyperconnectivity and they’ll be weird, or that there’ll be the same number of black swan events but they’ll be weirder due to hyperconnectivity?
I understand the reasoning behind the latter, but if it’s the former I would be interested to hear why hyperconnectivity results in more black swan events.
This bit seems important. The “only 400 tested!” statistic people are citing seems much more reasonable in this context.