s1dev

Quantum computers don’t have such a compelling case for optimization. Classical optimization algorithms are really good. Usually if you look hard enough you can find a sqrt(n) speed up but I don’t think these are the applications that will change the world

Mainly quantum simulation. The (most industrially relevant) dream is that you can do direct simulation of large, complicated molecules to better understand them. This is a place where the best understanding is that quantum computing is an exponential speed up although this is quite difficult to prove

How do you propose to power a fleet of spacecraft at 550AU? Would you be able to acquire enough RTGs? What about DSN time for the downlink?

For the moment the “killer apps” involve simulating quantum systems: Chemistry (catalysts etc), microbiology (protein folding), nuclear / particle physics

The problem with the 100LL phase-out is that you have over 70 years of engines and fuel systems designed to work with leaded fuel.

In order to complete the phase-out you either need to: force people to put a new $30K+ engine into a <$50K plane, totally replace the fuel system, and get a supplemental type certificate or: Require that the new fuel is a drop in replacement. One would have to show that the new fuel is 100% compatible with 100LL without modifications to the aircraft.

The FAA has been attempting the second route for some years, but to date has not been terribly successful. The first route has been done successfully in other regions such as the EU, but I would suspect there is far less legacy general aviation aircraft. I personally look forward to the mainstream introduction of diesel engines into the US market that are able to run on Jet-A

Here are a few ways to kill you and all your passengers under instrument flight rules. Most of these involve a single mistake and a "controlled flight into terrain"

- Turn too early/late past an arbitrary point in space

- Turn the wrong direction

- Descend too early past an arbitrary point in space

- Use the wrong altimeter setting

- Failing to recognize and troubleshoot a failed instrument

- Failing to recognize signs of hypoxia

Not to mention dealing with emergency procedures while you don't do any of the above

Here is an example from a few years ago where a 777 missed a mountain by a few seconds after some non-standard instructions by ATC

https://www.youtube.com/watch?v=tFdXax7Zh_g

A good example of a recent technical failure where one likely alternative is "everyone dies" https://en.wikipedia.org/wiki/British_Airways_Flight_38

The engine needs to be able to run without electrical power, or you need to convince the FAA that your backup will fail with an extremely small probability. This is one of the reasons magnetos + carburetors are common in general aviation

CPUs you can implement on FPGAs or something else whereas FPGAs are really only useful if you can get it made. Getting the design rules from a foundry usually involves an NDA and you will need (proprietary) EDA tools.

Also the utility of open source CPU cores is much higher than an open source FPGA since you're not going to be able to customize it anyway because of mask costs.

$100K is a lot of money for some experimental low temperature labs. The other issue is that vibrations from compressor units can interfere with sensitive experiments like scanning tunneling microscopy. Helium is expensive enough as it is that I think most labs would like to recycle it, but there are drawbacks.

You can't get thousands of variables with all to all connectivity on a quantum annealer. The noise will kill it. I.e. you will never have such a device with reasonable coupler precision.

The techniques for decomposing a problem and solving the sub problems are a big (exponential) overhead because of frustration effects which require you to look at the entire problem at once.

And for context the PUBO -> QUBO mapping (high order to low order) + topology embedding can blow up the number of variables by an order of magnitude or more. The solution space scales exponentially with the number of variables so a naive solver will again deal with a messy overhead.

It remains to be demonstrated that D-Wave has an asymptotic advantage over classical optimization algorithms. Even synthetic spin glass benchmarks specifically designed to give the DW an advantage show no advantage.

Furthermore, real optimization problems have high connectivity, high weight cost function terms, and require higher precision than available. The first two will require an unacceptable overhead for fixed topology solvers. DW is not useful for real world optimization problems.

Google on the other hand has actually done something a classical computer can not do

For what it’s worth, Burbank actually has a really short runway.

Journals often charge extra publication fees per figure. Any supplementary material also needs to be available decades from now

The real paper is https://arxiv.org/pdf/1810.05338.pdf

1810.05337 is a summary of results for PRL

Both are very good