[Cad] [CAD] Solvespace

Nancy Ouyang nancy.ouyang at gmail.com
Tue Nov 17 19:23:46 EST 2015


Based on group interest, this week we'll read about solvespace, which as of
March 2015 has a native linux port. It's Brep based.

I'll throw out these links and if someone has input on any particular
reading / activity for the reading group, let the list know -- otherwise
I'll pick something tomorrow night.

Thanks,
--Nancy
==================

Youtube demo: https://www.youtube.com/watch?v=2r8Y4tOeYZQ
Source: https://github.com/jwesthues/solvespace
Windows: http://solvespace.com/download.pl
Linux: http://solvespace.com/forum.pl?action=viewthread&parent=574
OSX: http://solvespace.com/forum.pl?action=viewthread&parent=586

Constraint solver as python library:
https://github.com/BBBSnowball/python-solvespace

Brief description of Solvespace's constraint solver:
http://solvespace.com/tech.pl

Matt's explanation of constraints:
http://www.mattkeeter.com/projects/constraints/

Example of adding a new constraint:
https://github.com/ruevs/solvespace/commit/9963d9bb25d1e17ed5d5bbe8cd42fc73c8326f7c
(from http://solvespace.com/forum.pl?action=viewthread&parent=1005)

My friend's complaints about solvespace:

>
> it seems like the solver might be mostly symbolic
> with some sort of parser to evaluate
>
> well i need to look at the source
> but the equations should be numerically represented
> otherwise you spend most of your time parsing a large system rather than
> doing math
> even worth parsing is slow because it has lots of branches
>

>
might look at solvespace in more detail later, it at least sounds hopeful
> depends on whether the newton iteration is done on a matrix or by
> symbolically evaluating the constraints
>
> ew solvespace stores constraints and jacobians as expression trees
>

>
if i really cared about this i would start rewriting the Expr and System
> classes of solvespace
>

~~~
narwhaledu.org, educational robots <http://gfycat.com/ExcitableLeanAkitainu>
 [[<(._.)>]] my personal blog <http://www.orangenarwhals.com>,
orangenarwhals
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/pipermail/cad-rg/attachments/20151117/afc9d5fb/attachment.html


More information about the CAD-rg mailing list