Forgot your password?
typodupeerror

Submission + - Google tests AI processor in space (npr.org)

Geoffrey.landis writes: Google just put a satellite testing AI chips into space. The refrigerator-sized satellite is part of a research project, "Suncatcher," the company hopes can pave the way for orbiting AI data centers powered by the sun. The prototype satellite has four specialty chips in it called Tensor Processing Units, or TPUs, that Google designed for machine learning and has already deployed in data centers on Earth. The chips will run a version of the company's Gemma AI model for 15 minutes at a time due to heat management constraints, using the open weight model to answer simple queries. In a blog post, Google said the mission will gauge how the TPUs handle "the physical stress of spaceflight and the radiation and thermal extremes of space." The satellite launched October 1 is a prototype that Travis Beals, senior director and lead of Project Suncatcher, called "a very minimal test" to make sure the chips can run in space. A SpaceX rocket carried it into orbit, and Planet will get it up and running. Beal said Google will then fire up and test the TPUs. The goal is for the mission to be operational for a year.

Comment Hot electrons [Re:Perovskite thin films] (Score 3, Informative) 42

I have some amount of skepticism too. Nanosecond lifetimes of hot electrons are amazing, but to date there's no way for turning hot electrons into electrical energy*. So they'd essentially have to invent a technology from scratch.

--

Your last sentence... I was just about to post something similar... far as I remember, heat doesn't generate electricity in PV cells... it might even hurt their efficiency, but you beat me to it.

Correct on both, in fact heat decreases efficiency of PV cells (thermal dark current is opposite to photovoltaic current). However, keep in mind that "hot electrons" aren't exactly hot in the sensible-heat sense-- it means that the electrons are in a high-energy state well above the bandgap of the semiconductor. The solar photons come in at an equivalent temperature of 6000K, and all of that energy gets transferred to the minority-carrier electrons, but they quickly shed their excess energy (by transferring energy to the lattice and to other electrons or holes) and relax to the edge of the conduction band (or, ~kT above it). So, using "hot electrons" means somehow finding a way to transfer energy out of the highly-excited initial state immediately after photon absorption, rather than the relaxed quasi-equilibrium state of the conduction band.

So, there's energy there, and in principle it could be used. It's just going to take a new approach to use it. I can't think of a way to do that without adding junctions, and then you end up essentially just re-inventing the multi-junction solar cell.

Now... you could route a water line behind the PV cell, and that'd give you warmer (depends on conditions) water (and might help cool the PV cells which might help their efficiency).

Yes, but of course now you're harvesting the sensible heat, not the hot electrons. This can be useful, if you need low-grade heat.

Comment Perovskite thin films (Score 5, Informative) 42

The perovskite semiconductors here are a very different technology than the old thin photovoltaics. They have the potential to be much more efficient, and the deposition techniques are cheap and comparatively low-tech

Whether they can succeed in simultaneously be highly-efficient, resistant to degradation in the environment, and still be low cost is the subject of a lot of work. We'll see.

They were supposed to be manufactured for the price of a lollypop per square meter back in the '90s.

To be more specific, the target back in the '80s and '90s was clearly articulated as being fifty cents per watt (where "per watt" meant, when illuminated at standard conditions of 1 kW/square meter.) Silicon photovoltaics can now be purchased at 12 cents per watt. In today's dollars. They not only hit the target, they blew it away.

The reason the thin films of the '80s and '90s lost was not that they were bad, but that silicon simply outcompeted them.

So forgive my skepticism regarding new breakthroughs.

I have some amount of skepticism too. Nanosecond lifetimes of hot electrons are amazing, but to date there's no way for turning hot electrons into electrical energy*. So they'd essentially have to invent a technology from scratch.

--
  *(thermoelectrics convert hot electrons into usable energy, of course, but there not just the electrons, but the whole lattice is hot. And the efficiency is worse than a solar cell).

Comment Re:Loss of contact with reality (Score 1) 260

The only thing resembling "babble" is the fact, once the unwashed masses steal a technical term to use wrong, there is no possible hope to convince them otherwise,

Except it wasn't the unwashed masses who labeled large language models "AI". It was the AI corporations (maybe "AI" should be in quotes here), who wanted it to sound sexier.

Comment Re:Only two? (Score 3, Insightful) 67

Fascinating; three completely disparate things, none of which have any relation to teleology, the nominal subject of the sentence.

"Wigner's Friend" is a thought experment, pointing out that in a strict interpretation of quantum mechanics, even conscious observers must be in an indeterminate quantum state until observed by an external observer. The least action principle states that the path of an object moving through a potential field is the one in which the action, the time integral of the kinetic energy minus the potential energy, is minimized. The second law of thermodynamics says that in any closed system the change in entropy, defined as the change in heat divided by the temperature, must be greater than or equal to zero; for a not-closed system, in order to decrease the entropy in one place, energy is needed (typically to be radiated as waste heat; i.e., life takes energy from the sun, and rejects waste heat to the envionment (which ultimately radiates it to space).) None of these have any relation to teleology, the philosophical concept that causality is driven by purpose.

Comment Re:Mercator projections are infinite (Score 1) 274

And which part: during my school time, I never have seen such a map, do you not grasp?

Correct: I do not grasp. I think you simply don't notice where the equator was, because it's so common for it to be below the centerline that nobody notices or pays attention.

I gave links to six examples of the common ordinary maps you see all the time; I could have given a hundred. Did not even one look famililar?

Here's a clue. If on a Mercator projection Antarctica isn't larger than all the other continents combined... the equator isn't centered.

All maps I have seen: have the equator in the middle of the paper. Where else?

You already explained where, and why. This is why they put the equator below the center:

So obviously "to save paper space" or "to have more details in the north" they put the "base line" down.

Comment Re:Just my opinion (Score 1) 88

I've always assumed dark *anything* was nonsense.

"Dark" just means "we can't see it". Anything that isn't either emitting light or other electromagnetic radiation, or reflecting light from something it's close to that's emitting light or other electromagnetic radiation, is "dark".

The data is, if you look at the motion of the stuff we see, and calculate the motion based on the masses of stuff we see, the trajectories deviate. The obvious thing to guess is that it's the gravity of something (or things) we don't see. This has worked before-- for example, this is how Neptune was discovered; it's how close binary stars were discovered.

An alternate hypothesis is that there's some other force pulling or pushing on them, or else gravity doesn't work the way we think it works. Either way, you're hypothesizing something unknown.

Comment Every (large) galaxy has a black hole (Score 1) 88

every galaxy has a black hole. This is incorrect. Who says this?

Every large galaxy is believed to have a black hole. Dwarf galaxies (like the Magellanic Clouds) may or may not.
  https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fwww.astronomy.com%2Fscie...
  https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fastronex.net%2Fwhy-is-th...
  https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fscience.nasa.gov%2Fmissi...

Comment Mercator projections are infinite (Score 1) 274

The first links excluded Antarctica.

Exactly. That's what happens when you move the center of the map up above the equator; the bottom part of the map goes off the page. If you kept the bottom part on the page, the equator would be in the centered. By definition.

Did you not actually read what I posted that you're responding to?

...
If a map would be a world map, the equator would be in the middle. Anything else would not make any sense.

A Mercator projection which actually showed the entire world would be infinite in size, because the north and south poles project to + and - infinity. Every Mercator projection leaves out parts of the world near the poles. However, the standard Mercator projections leave out more of the southern polar regions than they do of the northern polar regions. Which means the equator is below the center of the map.

Comment Mercator Equator [Re:There's no good answer...] (Score 1) 274

I never saw a Mercator projection map where the equator was not in the middle.

I suspect you're so used to it that you don't even notice that in most Mercator projections the equator isn't in the middle:
  https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fwww.worldatlas.com%2Fr%2Fw...
  https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fprintable.rhinoplax.co...
  https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fimage.shutterstock.com...
  https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fscholarlykitchen.sspne...
  https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fwww.mapsales.com%2Fmap-i...

For comparison, here's a Mercator projection with the equator in the middle: https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fwww.alamy.com%2Fpolitica... Antarctica is larger than all the other continents combined!

Comment There's no good answer (other than a globe) (Score 3, Insightful) 274

I am not sure that the UN should be in the business of defining a "standard" map projection-- why shouldn't everybody use the map that they find most useful? But if they have to set one single standard map, I'm happy it's not Mercator, because, indeed, the Mercator projection is misleading on area.

In its defence, though, while Mercator distorts areas, it keeps shapes correct (locally), and it also keeps directions correct. Small-area accuracy and correct directions are more important than the large-area distortions for most practical use.

Any projection that keeps areas correct must distort shapes. You can't mash the surface of a sphere onto a flat surface without distorting something.
My favorite is the Lambert equal-area projection, which actually gives you the impression of being a projection of a globe, but you can't win-- all the projections have flaws.

I'll also note that the new UN standard map has moved the equator back up to the center of the map. Pretty much every Mercator projection you ever see has the equator moved toward the bottom, because (1) the far southern reaches are not important to people in the northern hemisphere, and (2) if you put the equator in the center, the Mercator distortion makes Antarctica absolutely HUGE, which puts too much of a highlight on how inaccurate the projection is.
 

I will also point out that the UN logo itself shows a polar view, different from ANY of the other views (but very clearly is favoring the northern hemisphere).
https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fglobalsymbols.com%2Fuplo...

Comment Re:Pretty impressive (Score 1) 73

While true, personalized genetic alteration of medicine can be expected to remain quite expensive until the process is fully automated.

Exactly. Automation of processes that used to be expensive is an excellent example of how treatments that started out outrageously expensive become cheap and routine, and dirt-cheap genetic testing is the poster case for this,

Slashdot Top Deals

Yet magic and hierarchy arise from the same source, and this source has a null pointer.

Working...