Higher dimensional geometry can show up in lower dimensional problems. This numberphile video (https://www.youtube.com/watch?v=6_yU9eJ0NxA) involves a puzzle about throwing darts at a dart board which is solved by using the volumes of 4+ dimensional spheres.
This is particularly good. Apart from the already mentioned seasonal references, a common element is that the thirst two lines should seem completely unrelated and the third one provides synthesis/closure.
Taking the difference between normal complexity and asymptotic complexity to the extreme you have https://en.wikipedia.org/wiki/Galactic_algorithm which do have the best asymptotic performance, but only on values of n so large they never come up in real life.
This seems to be similar to the usage of WAAS in the US (http://en.wikipedia.org/wiki/Wide_Area_Augmentation_System), just on a smaller geographic scale. One of the bigger sources of error in GPS today is due to various atmospheric effects that bend the signal and increase the signals propagation delay. A ground stations in a known position can generate correction for the errors but the farther it's from the ground station the less useful the correction is. I've seen GPS systems used in farming applications that use a local tower that can provide corrections to give a very accurate position for the local area.
Back in 1995 I wrote the mapping software for a "Tundra Tracker", this was the navigation software for a supply vehicle that could navigate in the arctic in whiteout conditions. The idea was to have someone drive a path while conditions were good, and we would then lock in all the waypoints. Rather than store all the waypoints for the path, we were able to eliminate about 99% of them with various splining algorithms.
One of our big concerns back then, though, was that US Military would degrade the SA (Selective Availability) signal - it had happened in the Gulf War, and our tracks, which were normally accurate to within 3-4 meters, went wacky and were almost 50 meters off at times.
The solution was to mount a radio antenna, have it calculate it's "True" location by averaging a couple days worth of GPS signals, and then, continually transmit the difference between a current GPS signal and it's known location. This Differential GPS let us calculate paths to within 1m of accuracy.
When I asked what prevented a military opponent from doing the same thing, I was told that the first thing in the battlefield that gets hit as a command center would be anything transmitting RF, particularly if it was believed to be for D-GPS. Also, D-GPS isn't that effective for missile tracking, as you need to get differential signals in areas outside of your zone of control.
It does seem like WAAS at first glance, but I actually think they are creating their own GPS-compatible satellites:
> To put it simply, with QZSS it is like the number of GPS satellites has been increased. Because QZSS has interoperability with GPS, the number of satellites that can transmit satellite signals at the same time is increased, which makes stable positioning possible. This also decreases the positioning errors as described. [1]
With L1 and L5 carriers on GPS now, you can definitely get down to cm-level positioning already. I think they are just launching more satellites in Japan-specific orbits so they have better, faster GPS coverage above Japan, and especially in dense cities where you need a high angle of elevation for positioning satellites.
The farming systems are more real-time kinematic GPS, which broadcast local corrections from a dedicated reference statiion. [2]
personally I find "short reply OK" to be preferably. It comes across as less harsh, it tells the other party that you are ok with removing the random fluff if they want to. VSRE is a demand that all the fluff be removed
I've had similar experiences with AVRs, they are quite robust and will handle shorts, reversed polarities and the like rather well. They seem to be rather conservative on the datasheets probably as they are guaranteed over a much larger temperature range than most people use their chips at.
I second the recommendation, I've been using lean's service for a rather long time (I've go boards from back before he worked out the right purple color) and I've always been happy with the speed and quality, it's a great service.
This is not entirely true, in an ideal world a CMOS device just sitting there not switching would use basically no power, but in actual application there is some leakage (both form not being fully off and some current jumping from the gate). This leakage gets worse as the transistors get smaller and when you have billions of transistors even the tiniest amount of leakage adds up, the result is leakage current can be a significant amount of the power used by a modern processor.
I completely agree with the sentiment about high college costs, the state school I went to cost ~44k for all four years, add in various scholarships (some of which I got as I'm pretty sure I was the only person to apply for them) and internships and I graduated college with no debt and more money than I started. Going to college does not necessitate taking on crushing debt, it just requires some extra legwork and picking a school that more fits your budget.
This is true, the flexibility you get form being able to reconfigure things does result in less efficiency (speed, size, power) than some dedicated hardware. An FPGA can be more efficient than general processor if it can be tailored to the specific computational task, but it will still be slower than if you made an IC with the specially tailored hardware.
This is true, and as far as I see - this is an inherent limitation. Dedicated, static ASICs will always be faster or more efficient.
However - I think there's some interesting work to be done in hybrid approaches. For example, besides FPGA cells, you can include fast on-board cache, arithmetic units, etc. So, this way, if you're switching between GPU and CPU cores, a subset of that can be reused.
For reference, a typical FPGA cell includes a 1-bit flip-flop and an LUT (look-up table). Flip-flops are memory, LUTs simulate the logic.
