Message from @Goldsteel
Discord ID: 490025517341212672
That's an exaggeration
But still
We have significantly better tools
Up until the mid 90's a lot of controllers were still analogue in areospace and even transport
Digital computers have made leaps and bounds being able to record gigabytes of data on a whim
So this FE proof is comparing the processing speed of how fast computers from 1985 can deal with satalitie data?
Compare that to 64kB of RAM
😏
Not specifically that no
I have 32 Gb in my personal computer
It's just that instrumentation has got a lot better so we don't really have to simplify the controllers too much
I wonder how many spaceships I could run
<:GWbruhGalaxyThink:405065193287319552>
Sometimes, the error on additional corrections isnt needed at low altitude because your percentage error is too low
However
PID controllers have an accumulated error and need correcting
What if you only fire a rocket for 20 minutes?
What would that cumulative error mean?
Does your maneuver need to be that specific?
Absolutely not
We even used sextants and classical mechanics on the way to the moon
Okay so, we simplify physics in some cases because we don't need to calculate everything?
You could calculate the spacetime curvature in GR
But you don't need to
Yeah
It's kinda laziness then?
That's the jist of it
Well, think of it this way
Diminishing returns on investment
You could correct all the problems but if they're not really problems because you don't need the precision why bother?
Okay that makes sense
"One is an analytical approach using a flat-Earth approximation to predict geopotential information quality as a function of spatial wavelength."
¯\_(ツ)_/¯
Could you sum up why this paper isn't FE proof like in a sentence or two?
I'd try but I'm a little lost reading your simplification
He basically summed it up with the satellite statement he made initially
It's a FE proof which requires space and satalities
<:thonk:485324336874651650>
1. We use a lot of approximations in controllers
2. Assuming a flat stationary non-rotating plane is a decent approximation.
3. Experiments report cumulative error
4. Large experimental errors are telltale of large appoximation components.
5. It's a satellite in orbit. Contradicts the point.
6. Most papers quote their approximations.
For information gathering purposes, a flat plane with all data directly displayed in relation to its surroundings makes for an easier and more efficient system. The plane contains only about 30% of the known area of the earth, and the paper specifically cites the cases and reasons for why the Flat model used has situational advantages over a spherical or elliptical model. It is not an FE proof.