thirdhaf

My assumption is it's referring to mathematical tables and the historical makers of them. https://en.wikipedia.org/wiki/Mathematical_table

As a quick and dirty rule of thumb measuring parts per million in anything except time or frequency will get expensive. Temperature drifts will cause expansions and contractions on that order if you’re measuring lengths.

Yes fun story, Jack was an assistant signalman and got paid for the job, partially in beer! https://en.m.wikipedia.org/wiki/Jack_(baboon)

By 1870 pi was known to several hundred decimal digits, for something like this calculation where you have other large sources of error Archimedes approximation from 2 millennia earlier would probably be fine. (<1% error)

https://en.m.wikipedia.org/wiki/Chronology_of_computation_of...

Oxide makes a rack-scale computer called the Oxide Computer, here's the intro page: https://docs.oxide.computer/guides/introduction

That phrase distinguishes the internal group responsible for that part of the architecture, don’t think it’s a marketing term.

Could have been Earl Muntz, who was in the TV manufacturing business, learned about this from one of Bob Pease’ columns.

https://www.electronicdesign.com/technologies/industrial/boa...

The explanation is really well done, it captures the essence of the Pauli exclusion principle without delving too deeply into the weeds. In my opinion the best part of the video is the explanation of the "hole" quasiparticle at 6:10 (I learned this as a pseudo-particle but will defer to Wikipedia [1]).

While a great introduction to semiconductor behavior this does gloss over a very important detail namely direct vs indirect semicondoctors as some others have mentioned. In the video the detail that's glossed over relates to the nature of crystals, namely that they're highly ordered repeating structures but that they don't look the same when viewed from every direction. This means that there isn't a single band-gap but multiple ones depending on the direction of the crystal you're contemplating.

At this point you may reasonably ask why the direction matters and now we unfortunately get deep into the weeds with quantum mechanics again. When a single photon is absorbed in the semiconductor system both momentum and energy must be conserved. The momentum of the photon for something like the Silicon bandgap is quite small (something like the equivalent of an electron traveling at 1500m/s) while the momentum of room-temperature conduction electrons is substantially faster [2] so as a very slight simplification transitions due to the absorption of photons are not accompanied by a change in momentum and so we only care about the band structure (and the accompanying free carriers) associated with a particular crystal direction.

In particular in Silicon you have what's called an indirect bandgap, namely the minimum energy conduction band electrons have a different momentum from the valence band holes ([3]) and as a consequence while you can _absorb_ a photon in order to make a detector you cannot make it efficiently _emit_ a photon as an LED should (something the video got wrong).

None of this matters for the heart of the video, which focuses blue LEDs in the GaN materials system which is definitely a direct bandgap material, however if someone does manage to create a manufacturable light emitter in pure Silicon expect an absolute revolution with regards to optical computing and photonics. (Not for lack of trying, this has been the holy grail for at least 20 years, possibly longer)

[1] https://en.wikipedia.org/wiki/Quasiparticle [2] https://www.chu.berkeley.edu/wp-content/uploads/2020/01/Chen... [3] https://www.iue.tuwien.ac.at/phd/wessner/node31.html

NREL has a great visualization for the research cell (as opposed to module) side of things here: https://www.nrel.gov/pv/cell-efficiency.html

Funny story, that demo only exists because the designer was mis-categorized as a software engineer after an acquisition by Apple. After nearly getting the person in question fired it ultimately resulted in a successful job category change for them.

Was it this one perhaps? https://news.ycombinator.com/item?id=7261003

From the paper it looks like most of the interesting work with the laser and optical train use an external board to recreate the feedback and control circuitry already present and required for normal operation in a DVD/BluRay drive. It would be great if you could have more control over the existing hardware in these drives.

I found a project that started the reverse engineering process on a popular bluray drive [1] but it really looks like an uphill slog against undocumented CPUs and motor control chips among other obstacles. Anyone know of any other resources for reusing the existing hardware but modifying the control software?

1. https://github.com/scanlime/coastermelt/

Neat hack, but why would you want a fan on the internet?

There's an episode of James Burke's Connections that explores this idea with a bit of depth, I recommend watching the whole series if you have the time.

James Burke Connections, Ep. 4 "Faith in Numbers"

With regards to how you locate it, the paper proposes looking at existing data from FERMILAT, a gamma ray telescope which could possibly pick up anihilations from a dark matter cloud surrounding such a black hole. Since the object itself is so tiny (I’ve never seen a 1:1 scale figure in an astronomy paper before) you need to detect its presence in some other way.

Arecibo is and has been used for radar astronomy as well https://en.m.wikipedia.org/wiki/Radar_astronomy

You can use the phenomenon of negative differential resistance to improve the design of some very common circuit elements. For example you can move from a classic six transistor SRAM cell to a two transistor design. It's a bit dense reading but here is a research synopsis from one group working on this in silicon/SiGe. http://www2.ece.ohio-state.edu/~berger/summ_ritd05.html

This is a pretty thorough description of generation, not just the classification problem that is the usual introduction to machine learning.

I just loved the map of the generated place names [1] the illustrations are hilarious.

[1] http://colinmorris.github.io/assets/rbm/kicksville.png

Link to the other HN thread on this topic: https://news.ycombinator.com/item?id=5391667

If the comments are to be believed someone's already lost their job over this.

Unless the planets happen to orbit very stable stars in the plane that sometimes puts them between earth and their star. With Kepler and some earth-bound telescopes you can discern the brightness dips from these and sometimes glean minimal spectroscopic information on the planet. Direct imaging is possible with interferometric methods and some more innovative approaches [1] but require funding.

[1] http://en.wikipedia.org/wiki/New_Worlds_Mission

I use Better Touch Tool [1] which is mostly about expanding defaults for multitouch gestures. I haven't gotten around to using the Better Snap tool which focuses on window management yet.

[1] http://blog.boastr.net/

That's exactly the application they're after.

When these discussions happen I like to calculate some hard systems numbers since they're rarely in either the literature or the popular science articles. All the power available naturally comes from the sun so we need to know the solar spectrum to start. Conveniently there's an ASTM standard [1], so let's use that.

Here we're only interested in capturing energy outside the human visible spectrum so only radiation above 750nm is interesting.

Integrating from the spectrum data we find that the absolute maximum power that we can ever get from this system is 464W/m^2 (Compared to 1000W/m^2 without wavelength restriction)

The paper [2] talks about 4.5% PCE (power conversion efficiency), let's be generous and say they can get to 10% before commercialization which implies 46W/m^2 MAX from these window tints.

Since these films are on the sides of buildings we also can't track the sun. Solar angle varies throughout the day but let's integrate for a south-facing building in North American latitudes and we get a power factor of about 0.5 over the course of a day so we're back at 23W/m^2. of electrical energy from these films.

They're going to have to be exceedingly inexpensive to make this effort worthwhile.

[1] http://rredc.nrel.gov/solar/spectra/am1.5/ [2] http://onlinelibrary.wiley.com/doi/10.1002/adma.201203827/ab...

Trace amounts perhaps. Generally speaking radiative heat scales as T^4 where your temperature is measured on an absolute scale like Kelvin. In addition most of the heat flux being radiated by a human falls in the the 10um range which is somewhat difficult to detect and most certainly difficult to convert to electricity.

In short, think microwatts.

If you want to implement it in a general way you still need to know how javascript supports closures though.

Completely agreed about 44.1kHz being a sampling rate. To make the rest of the numbers nicer I'm going to pretend we're talking about 40kHz sampling for the rest of this comment. It's clear that you can't distinguish between a 0-degree phase shifted 20kHz signal and a 0-degree phase shifted 60.0kHz signal (I thought we weren't going to talk about Nyquist here). However by the same token you CAN represent a 90-degree phase shifted 20kHz just fine. By way of example consider the bit stream (I'm also pretending 1-bit sampling for now) [0 1 0 1 0 1] which in this format represents a 20kHz signal. It's pretty obvious that we can represent the 90-degree phase shifted signal as [1 0 1 0 1 0] and there's nothing in the input stage that stops the original source from creating this particular type of signal. In fact we can unambiguously represent ANY phase shift in the same fashion, but as you point out then we start arguing about whether your sample has enough precision to unambiguously represent the input. With a 90-degree shift (corresponding to a ~23usec delay) we have none of the precision problems you allude to.

Assuming 44.1kHz sampling rate the smallest path length difference between the ears is approximately 6.8mm. This corresponds to 2.3 degrees (with sounds source at infinity). Humans can place sounds with about 3 degrees of resolution so unless you have some citations I am seriously skeptical about the claim that higher sampling frequencies gives you anything whatsoever.

Zero loss is only true for DC transmission. It's true, loss is lower than copper at 50-60Hz AC but radiative losses don't magically disappear. You're also limited in how much current you can carry since superconductivity breaks down in sufficiently high magnetic fields. For all that room temperature superconductors are revolutionary because of the jump in power density in things like motors and yes power lines.

This sounds suspiciously like what one of my friends did about twelve years ago for our freshman engineering project. Granted I believe these mice were more like $50 back then. See the 21 Aug comment here as proof that I'm not hallucinating: http://news.lugnet.com/robotics/handyboard/?n=*8563,-100&...

In a strange way I sort of did this with my EE and math degrees. I did draw the line at the Hartree-Fock approximation for doing ab-initio band structure calculations for silicon though so there's still room at the bottom for me. Since then I've come to realize that programming is more fulfilling to me but I do feel prepared for this particular line of rabbit-hole questions :-)

I'm sure part of the issue was that I was trying to teach myself Labview under a deadline as a fresh college grad with zero oversight from anyone with experience with it.