Mhmh yeah that. Should be easy enough to test, take same setup and replace the tapered cavity with equivalent cylinder. If it still produces the effect it must be some error. Their null test was with apparatus mounted completely differently and resulting in completely different thermal error graph. Or a test apparatus that is not as susceptible to thermal warping as they proposed in the paper.
Yeah, they pulled out the good ol pilot wave theory and pretty much said, we like this one, maybe our results have something to do with it. Well anything is possible, but there sure is nothing in the paper indication it might actually be so. Regardless, they do have something of a impulse response in the measurements and they are getting some results that seem to be more than just thermal error. But they have only whooping 9 measurements with pretty awful variance and tons of thermal error.
Would it have been so hard to keep the test cycling overnight to get more data points? Make a dummy test showing only thermal error? Better setup and more repetition is really required, but the paper has better than nothing and that is quite a bit, when it comes to potentially physics textbook rewriting experiments.
The question is not how much effect there is, or how it might one day be used, but they need to really prove that there bloody well is an effect and the paper imho falls just a bit short. If you squint right yeah there is something, but it would be nice if the say replaced the tapered resonance chamber with a simple cylinder and showed the "thermal error only" result for comparison and if they ran more test cycles.
Whats wrong with neural nets, its the only known technique that can produce programs capable of performing tasks a programmer doesn't know how to do. Deepmind programmers were not in fact better Go players than the world champion, but with neurals nets they did indeed produce a program that beat the world champion. That's pretty awesome, its one thing to use a machine to do something faster than a human can, its completely different thing do use a machine to do what human doesn't know how to do.
Silly people, if China won't manufacture iPhones Vietnam will, or India or Indonesia or some other similar place where you can open shop and hire 1000 people per day to work as a biorobot. Manufacturing these things in USA or Europe is flat out not viable, especially USA what with workers unions and all.
Obviously its better, all the politicians have to concentrate on watching their back and they get away with much less stupid crap. Politicians are a necessary evil, but best have them restrained by their coalition mates and have the opposition breathing on their necks instead of giving all the power to one bunch of morons or even worse, one individual.
Huge number of PLC-s run on a windows machine, TwinCAT ftw! Best way to develop a machine by a long shot. If an industrial machine has a PC its a safe bet it runs windows, so all of them run windows.
Someone physically goes on location and does it manually. If you update machine software you better verify that it actually changes what and how you want to change anyway, getting undesired results from software update is not exactly a rarity, especially if you don't have an actual production machine to test the software on first. Normally you hire a team of local service guys anyway, traveling half a world away to update PLC or swap some minor part in the machine gets rather pricey.
The difficulty of messing with industrial equipment is not how to mess with the software, but in how to get access to it in the first place. These days most newer machines don't actually have a physical PLC(unless you count safety plc that only handles the safety), instead they run a soft plc on a PC, generally side by side with windows using VT-x. Once you have access to such a machine messing with it is really not hard at all.
Slashdot Mirror
Mhmh yeah that. Should be easy enough to test, take same setup and replace the tapered cavity with equivalent cylinder. If it still produces the effect it must be some error. Their null test was with apparatus mounted completely differently and resulting in completely different thermal error graph. Or a test apparatus that is not as susceptible to thermal warping as they proposed in the paper.
Yeah, they pulled out the good ol pilot wave theory and pretty much said, we like this one, maybe our results have something to do with it. Well anything is possible, but there sure is nothing in the paper indication it might actually be so. Regardless, they do have something of a impulse response in the measurements and they are getting some results that seem to be more than just thermal error. But they have only whooping 9 measurements with pretty awful variance and tons of thermal error. Would it have been so hard to keep the test cycling overnight to get more data points? Make a dummy test showing only thermal error? Better setup and more repetition is really required, but the paper has better than nothing and that is quite a bit, when it comes to potentially physics textbook rewriting experiments. The question is not how much effect there is, or how it might one day be used, but they need to really prove that there bloody well is an effect and the paper imho falls just a bit short. If you squint right yeah there is something, but it would be nice if the say replaced the tapered resonance chamber with a simple cylinder and showed the "thermal error only" result for comparison and if they ran more test cycles.
Whats wrong with neural nets, its the only known technique that can produce programs capable of performing tasks a programmer doesn't know how to do. Deepmind programmers were not in fact better Go players than the world champion, but with neurals nets they did indeed produce a program that beat the world champion. That's pretty awesome, its one thing to use a machine to do something faster than a human can, its completely different thing do use a machine to do what human doesn't know how to do.
Silly people, if China won't manufacture iPhones Vietnam will, or India or Indonesia or some other similar place where you can open shop and hire 1000 people per day to work as a biorobot. Manufacturing these things in USA or Europe is flat out not viable, especially USA what with workers unions and all.
Obviously its better, all the politicians have to concentrate on watching their back and they get away with much less stupid crap. Politicians are a necessary evil, but best have them restrained by their coalition mates and have the opposition breathing on their necks instead of giving all the power to one bunch of morons or even worse, one individual.
Huge number of PLC-s run on a windows machine, TwinCAT ftw! Best way to develop a machine by a long shot. If an industrial machine has a PC its a safe bet it runs windows, so all of them run windows.
Someone physically goes on location and does it manually. If you update machine software you better verify that it actually changes what and how you want to change anyway, getting undesired results from software update is not exactly a rarity, especially if you don't have an actual production machine to test the software on first. Normally you hire a team of local service guys anyway, traveling half a world away to update PLC or swap some minor part in the machine gets rather pricey.
The difficulty of messing with industrial equipment is not how to mess with the software, but in how to get access to it in the first place. These days most newer machines don't actually have a physical PLC(unless you count safety plc that only handles the safety), instead they run a soft plc on a PC, generally side by side with windows using VT-x. Once you have access to such a machine messing with it is really not hard at all.