idetc-2013-paper

The repo for the paper:

Constrained Multibody Dynamics with Python: From Symbolic Equation Generation to Publication

Gilbert Gede, Dale L. Peterson, Angadh Nanjangud, Jason K. Moore

• There is a 10 page limit.
• LaTeX template: http://www.asme.org/kb/proceedings/proceedings/author-templates
• Draft paper info: http://www.asmeconferences.org/IDETC2013/DraftPaperPrep.cfm
• Paper types: http://www.asmeconferences.org/IDETC2013/PaperTypes.cfm
• Conference website: http://www.asmeconferences.org/IDETC2013

Things to perhaps address:

• Motivate the need for streamlining the comprehensive study of a system.
• Explain ‘symbolic’ dynamics. People may not know. Reference AutoLev and MotionGenesis. There are probably other packages for motorcycle dynamics.
• Michael Sayers wrote the original code that AutoSim/VehicleSim/BikeSim etc are based on. It is a Kane’s based symbolic manipulator for multibody EoMs.
• MBDyn is an example of a numerical open source mutlibody (and flexible i think) dynamics package.
• Include a code sample for a problem.
• PLoS One software paper guidelines: http://www.plosone.org/static/guidelines#software

---

Paper Outline

• Abstract

• Introduction

  • What do we mean by analytical dynamics?
  • Why do we need a tool for symbolic dynamics?
  • Who benefits from this?
  • What follows in this paper?

• Demonstration Problem

  • What is the problem, and what does it look like?
  • Why did we pick it?
  • How do you (start to) write the problem by hand?
  • What does the code look like to generate the equations of motion?
  • What do the equations look like?
  • What can you do next?
    • Simulate & Visualize
    • Pretty/Latex Printing
  • Workflow for problem
    • sympy.physics.mechanics Code
    • SciPy (or ndsolve code)
    • plotting code
    • visualization code (we could potentially use something like d3.js to have an animation inside the IPython notebook...will take a little more effort, but be badass).
  • Results of problem
    • Plots
    • Example of 3D
  • It’d be pretty cool to have this whole problem in an IPython Notebook that you can easily download and run and play with. This could be included as supplementary materials.

• Software Validation (Why should I trust you?)

  • tests within sympy
    • ensures stability/consistency during development
  • benchmark validations (rolling disc, bicycle, other...)

• Software Design (How does it work?)

  • How do I get it?
    • Download and installation (keep it simple and short)
      • sympy (talk about usefulness and stability of dev branch)
      • basic PyDy tools: i.e. the SciPy stack
      • http://scipy.github.com/install.html
      • http://numfocus.org/projects-2/software-distributions/
  • How do I learn to use it?
    • SymPy Docs for sympy.physics.mechanics
    • PyDy.org for start to finish problems (and accompanying pydyexamples git repo)
  • What is it made of (modules, classes, and functions)?
    • SymPy basics that it uses?? Just a sentence or two. cite Sympy for more info.
    • list or simple figure
    • how about a definition list: name of item + plus sentence or two describing it. A list alone seems like too little info. sure
  • How do these classes interact with each other (probably figures)?
    • ReferenceFrame/Vector & Dyadic interactions (detailing essential.py, visually)
    • ReferenceFrame Tree (how a tree is formed, possible pitfalls)
    • Vector Assemblage (vector is a list of parts of frames and frames - shown visually)
    • Point Tree (similar to RF)
  • How is this translated into equations?
    • container classes (Particle, RigidBody)
    • KanesMethod/LagrangesMethod
      • Describe the classes and their methods, probably a page for each class. At least give the basic understanding of how the classes work.
      • Talk about the methods for constraints and auxiliary speeds.
      • mass matrix
      • cite the paper that Luke and Gilbert wrote about linearization instead of writing about those methods here
      • Explain the form of the EoMs that these methods produce and speak some on solving them for the u dots and the pitfalls associate with that, why it may be better to do it numerically
  • What else can it do?
    • Custom indices for RefereneFrames
  • What can’t it do?
    • problem size limitations - unknown
    • defining problems visually (future problem design a graphical body assembler that builds mechanics code dynamically)
    • fast translation to 3D visualization

• Conclusions

• Acknowledgements

• NSF