Kinematic simulation of a Binomial Distribution

A Spot for User Mechs and Kinematic discussions
Post Reply
BillM
Old Timer
Posts: 191
Joined: Wed Jan 06, 2016 10:12 am
Location: Mystic CT

Kinematic simulation of a Binomial Distribution

Post by BillM »

Art

I've decided to learn more about the Kinematic capabilities of the Augie Vector processor. Over the last few days I viewed some of you videos regarding the processor and duplicated some of your examples.

I always like to stress the limits of programs so I built a simulation to demonstrate the binomial distribution.  The inspiration for this model was a box that I had built decades ago when my kids were in 4th or 5th grade. The box contained BB's, a bed of nails, and catch bins that demonstrated the effect of a sequence of binary events that over time form the bell shaped binomial distribution.

The first two attachments are jpg's of the simulation near the end of the run.  The attached .vcp file contains the simulation and I also encapsulated the whole mess in a .mech file.  It took eight tries to tweak the model to get a reasonable approximation to the distribution.

The simulation starts out with 800 small balls that fall onto a funnel object. The funnel is an attempt to create a single stream of balls to drop into the grid of pins.  I lost count of the number of pins.

I run the simulation ~40% rather than 100%. I do see some anomalies at higher speeds where some of the balls drop through the catch bins at the bottom of the model. Perhaps these anomalies are caused by the decimation of some of the objecs and the sheer number of collisions the  program has to calculate.

I don't think I'll do much more with this model.  The comb structure of the catch bins should be taller.  The sides of the funnel should be longer since some of the balls spill over the edges.

Is there any way to delete empty layers?  The method I stumbled upon was to create a mech file of the entire project, start a new project and then import the mech file.

I'm still a bit confused over the layers and the relationship to the simulation.  I first built up my simulation model by importing DXFs each of which ended up on a different layer.  In order to run the simulation I moved all of the chains to the same layer.

If I build any more simulation models I'll probably build up the components of the model and create .mech files from the components.

Bill
Attachments

[The extension has been deactivated and can no longer be displayed.]

[The extension has been deactivated and can no longer be displayed.]

BinomialDistributionEx-02.jpg
BinomialDistributionEx-01.jpg
BMeyers
Old Timer
Posts: 84
Joined: Sun Oct 26, 2014 4:41 am

Re: Kinematic simulation of a Binomial Distribution

Post by BMeyers »

B-CNC (before CNC and before gearotic) I made a few - Galton Box/quincunx.  And drew a few but never cut them. 

Didn't consider using Gearotic to create a dynamic model.  THANKS!
Attachments
IMG_8909 Quincunx with milled UHMW stops and modified funnel.jpg
User avatar
ArtF
Global Moderator
Global Moderator
Posts: 4586
Joined: Sun Sep 05, 2010 6:14 am
Contact:

Re: Kinematic simulation of a Binomial Distribution

Post by ArtF »

Bill:

  Im impressed that you got to that level. :) , I also found it funny as its the same test
I once used to test the entropy functions. Ill check on your question, I havent been there in
awhile so Im cant recall how to change levels.
  Ive turned off the -CON switch in general release as I was finished with the vector
processor experiment and have plans to redo it in a better form at some point. Auggies
drawing capability was too bad for it and was the impetus to start Vexx, which is much
better that drawing and such and is close to ready for such things as vector capability
should I manage to get there.

  Thx for the photos and the description, Ill get back to you with the level changing
capability which as I recall was in a dialog somewhere..

Thx
Art
BillM
Old Timer
Posts: 191
Joined: Wed Jan 06, 2016 10:12 am
Location: Mystic CT

Re: Kinematic simulation of a Binomial Distribution

Post by BillM »

BMeyers & Art

I am not a CNC person myself.  My interest in Gearotics and now the Kinematic capabilities of the Augie Vector processor is to model/simulate physical systems.

My first attempt used Gearotics to model an 1860's E Howard Tower Clock as documentation of the clocks design.  The gear train used actual gear measurements in the design of the model.

I liked BMeyer's note about the Galton Box/quincunx that he built.  I was unfamiliar with that particular box.  My inspiration for the box I built was a large scale device at a Science Museum that demonstrated the principle.

Out of curiosity I dug around in our closets and attic and discovered that I still had the device that I built decades ago. The attached two jpgs show the device in its initial and final states.  The zip file contains an MP4 movie shows the device in action.  I edited the file to show the first few seconds and last few seconds of the device in order to shrink the size of the file.  Note the shape of the final distribution compared to the Vector Processor model is quite similar...not that I am surprised.

It took a while for me to get used to operating my device. The firse time I released the BB's too quickly and they jammed at the mouth of the funnel.  The Vector Processor model of the device had similar problems (just goes to show that the physics of Art's program really work).  Notice that the balls in the model are spaced into groups to slow down the volume of balls hitting the funnel.

I still have a lot to learn about the Vector Processor and its features.  Art's videos are a great help to get one started using the Vector Processor.  Experimenting with the processor is the best way to master its complexity.

I'm not sure of what I'll try to model next...I'll continue to explore.

Bill
Attachments
MAQ03238-Edited01.zip
(3.29 MiB) Downloaded 298 times
DSC03239.JPG
DSC03231.JPG
User avatar
ArtF
Global Moderator
Global Moderator
Posts: 4586
Joined: Sun Sep 05, 2010 6:14 am
Contact:

Re: Kinematic simulation of a Binomial Distribution

Post by ArtF »

Bill:

  Yes, there is something just very interesting about chaotic systems like that one.

  FYI, the reason greater speeds make things break is that Auggie doesnt ray trace a move, it computes
next position based on speed, so if a point goes into an object, that is sensed, but if a point is fast
enough it can go into , then back out of that object on the other side and that will not be seen as a collision.

  Its one of the many limitations in trying to do things fast..

Art
User avatar
ArtF
Global Moderator
Global Moderator
Posts: 4586
Joined: Sun Sep 05, 2010 6:14 am
Contact:

Re: Kinematic simulation of a Binomial Distribution

Post by ArtF »

Hi Bill:

Just checked and you should be able to move things to layers using the menu item layer. Create a new
layer, copy or move an item to a layer is in that menu. Even if from a saved mech file the item selected
should be able to be assigned to a layer at will.

  If this doesnt seem to be the answer, yell and Ill dig deeper.

Thx
Art
BMeyers
Old Timer
Posts: 84
Joined: Sun Oct 26, 2014 4:41 am

Kinematic simulation of a Binomial Distribution

Post by BMeyers »

BillM:
Thanks for digging around your attic and posting photos of the device you built! Appreciated.
Post Reply

Who is online

Users browsing this forum: No registered users and 4 guests