Anything you've done
- Home
- Board: /sci/ - Science & Math
- Reading: Anything you've done
We respect your right to privacy. You can choose not to allow some types of cookies. Your cookie preferences will apply across our website.
Hi guys i'd like to share some of my lame undergrad simulations with you, I guess we could share some shit we did in this thread.
Be aware that my gifs will be pretty shitty cause I just learned how to do them.
Also they will be in portuguese because it is my main language.
The first one is a magnus effect simulation, the balls are exactly the same, except that the light blue ball has zero angular momentum while the dark one has non zero. Perspective is bad in this one.
not something I did
>bovine aerodynamics
In this one I tried to recreate this situation
https://www.youtube.com/watch?v=5ZnKIPfIhAQ
Just pretend there is a rope linking them like in the video. The mass that starts below has 80kg while the other ones 8kg if I remember well... There is no resistive force and the rope is not rolling on the rod. I tried to set the initial conditions similar to the ones in the video.
this one is a wave function spreading in a square barrier
>>8095865
Wow thats impressive, I really hope I have this much of free time if I ever get a job
>>8095865
not something I did
> robotic balance
not undergrad, but I was developing a new type of optimization algorithms to solve problems that have contraints with a volume of 0.
>>8095867
It looks like a swinging atwood's machine
>>8095876
shit size didn't work
This one is a bar going towards equilibrium, the small mount in the left is a glitch due to bad boundary conditions
>>8095879
I dont know what that is but it looks very complicated to me, nice gif thought, did you use gnuplot?
>>8095878
Never realized that, I think that in fact they are the same system, except for the "anchor" point
This one I dont know what it is but it looks dope so im gonna post it anyway. It was meant to be a wave function spreading in a coloumbian potential but I did something wrong
>>8095898
heated bar*, its basically a graph of temperature vs position evolving in time
>>8095898
>I dont know what that is but it looks very complicated to me, nice gif thought, did you use gnuplot?
nah just matlab
basically an optimization problem (a simple parabola) with a constraint that the optimal point should lie on the sinusoid.
A gradient method (for example) would start somewhere and get stuck on one of the branches of the sinusoid before getting close enough to the center.
I can't remember what I did here, but this is a genetic algorithm with a twist that allows it to move past the constraint to see if the grass is greener on the other side.
I don't have an animation for the best algorithms I found though.
>>8095909
>I can't remember what I did here, but this is a genetic algorithm with a twist that allows it to move past the constraint to see if the grass is greener on the other side.
Sounds & looks like bringing the constraint in with small but increasing weight? Good way to avoid having to use a more complicated method to avoid getting stuck at local max.
>>8095910
probably! There was also something with a penalty function
this is one of the final ideas I kept, somehow it didn't have a .gif name so I didn't see it.
How did you make the gifs, OP? What reference did you use? I am also interested in computer simulations but my plots suck ass.
>>8096321
Not OP but plotting in Matlab is super easy and so is making gifs out of them. And since Octave and Scilab are very similar to Matlab I wouldn't be surprised if it's also easy in them
>>8096321 Here.
So this is a mass rotating in a spring. I will try to gif it later.
>>8095858
oh my god please do not draw 3d plots like that.
3d plots are inherently garbage on 2d screens and they're even worse when you don't include any kind of xy axis placement indicator
>>8096321
Hey OP here, I use gnuplot to the animations. All you need is a file with some columns of data to plot.
In pic related I used a "txt" file with 4 columns of data, they were basically x1,y1,x2 and y2. Which are the corresponding positions of both masses. Then you can use a command like this in gnuplot: just paste dot it / ulr9
The commands inside the "do for" are just parameters for the plot, you can learn about gnuplot plotting on the internet.
You can find more uself information about gnuplot animations here: gnuplotting dot. org / tag / animation /
Data file I used: just paste dot it / ulr1
Fortran 95 algorithm for a static central mass and a orbiting one: just paste dot it / ulqz
>>8097642
I know right, perspective really sucks in this one, I didn't know how to improve it
>>8097702
>I know right, perspective really sucks in this one, I didn't know how to improve it
that's what i meant about them being inherently garbage. there ISN'T a way to make it better, because rendering it as a static 2d image necessarily throws information away
your only alternative is to try to add that information back with a new dimension, like color-coding depth or adding a down-tick to the plane below, but those are kludges
>>8097718
Disagree. Dividing a vector to the point into three parts aligned with the basis and drawing 1-3 of these vectors to the point would give some perspective, and is better than color-coding or scaling or whatever other magic.
Yes, 3d to 2d is lossy, but we're good at interpreting it when there is a better attempt to encode the information in a way we can recognize.
>>8097718
Maybe a tracker following the object on each axis would be helpful
>>8097731
>>8097718
Op here I think this is way better
>>8097817
I like that, though I would add a vertical line from the end of the path to the position of the ball to give a tiny bit more perspective of height.
>>8097829
thats what you meant? or did you mean to actually draw the path?
>>8097880
how about this
>>8097829
>>8097880
>>8097909
ok im done after this one, i think its way better now, thank you
>>8097909
Almost perfect. Anything else I'd suggest is almost pedantic, like tiny "shadow" (single black pixel is fine) connecting the two projected lines for each ball so that in cases where there is not a sharp angle between them the boundary is still clear. I had imagined the vertical line being gray or something which would also work, but eh. Don't bother re-rendering for something so minor.
>>8097917
This is really nice
Made this for fun the other day.
Behold my godly Matlab plotting skillz.
>>8097909
yeah that's better
>>8097937
Are those new maps for each time step?
>>8097917
really nice! Not what I meant with the trackers but I like it
