Physics simulation is imho one of the things that feel like they should be easy but in reality is really difficult. And not talking about anything super fancy, just basic classical mechanics.<p>One of my dream projects would be to glue physics engine like Project Chrono[1] with modeling kernel like Open CASCADE[2], and some (FOSS) CAD (solvespace[3]?) to make an user friendly package to be able to setup simple scenes for simulation.<p>[1] <a href="https://projectchrono.org/" rel="nofollow">https://projectchrono.org/</a><p>[2] <a href="http://www.opencascade.com/content/open-source-core-technology" rel="nofollow">http://www.opencascade.com/content/open-source-core-technolo...</a><p>[3] <a href="http://solvespace.com/index.pl" rel="nofollow">http://solvespace.com/index.pl</a>
Couple of comments.<p>Seems like you are just comsol.<p>As someone who does this 50+ hours a week in industry (although only structural modelling but frequently coupled with optics/heat transfer/) and is reasonably well versed in the up and coming research I have a couple of questions (or would have if I was a potential buyer).<p>Can you mix 1D-2D-and 3D elements (I know fenics cannot). But for complex assemblies, making idealized representations is typically worthwhile (and typically necessary if anything is e.g. bolted together). Can you even mix different types of 3D elements (e.g. tets/pyramids/wedges/hexs).<p>The future of finite element modeling is probably is nurbs-enhanced or IGES-enhanced. Why are you building the past/current instead of the future?<p>Can your solvers do HP-refinement? Or even just H refinement.<p>Can you do glued contacts?<p>How much do I have to know about the time stepping scheme, or does your software work that out for you.<p>If I am doing thermal control or dynamic control can you handle that (I like that it interfaces with matlab so I can plug it into simulink--you should hammer on that in your copy--that is a really nice differentiating factor for people who would use FEA to drive control e.g. smart materials people, active optics, aerodynamic control people).<p>How easy is it to do optimizations? For instance (at home) I keep thinking about wanting to optimize a piezoelectric diagram fan like GE uses in their lights but for computers, how much easier is that going to be than just figuring out an open source software on my own.<p>Why FENICS and not DEAL.II or MOOSE (I don't know, I am just curious. I used to use FENICs but got frustrated because their input syntax kept changing on me.<p>Also really cool.<p>I have been thinking about writing software in this space but more nichy for about the past year and am beginning to get started, any interest in collaboration?<p>Also if you want to ask me questions about my thinking as a user who thinks a lot in these problems space, feel free to email me (I will want it to move to a phone call, but don't want to post my phone number on hacker news).
Here is why the "one physics simulation should cover it all" idea will never work: The word "physics" is just too broad. Is it about particle simulation (classical N-body simulations)? Or is it about fluid simulation (classical CFD = computational fluid dynamics, or classical Navier-stokes)? Is it about a game engine, digital art or just for a fancy GUI? What about the hardware, should we use GPGPUs, distributed computing or so?<p>Talking about the output, what is it the user is actually interested in? Some direct simulation outcome, such as trajectories or densities, still images, time series or some derived quantities? Is the user capable of programming in some language, is interface scripting even required? Does she actually prefer a CLI?<p>There are a lot of different industries with different needs. For certain engineering tasks (such as aviation), there are long standing tools which work fine and people are used to. For hobbyists or the gaming industry, the same applies.<p>Disclaimer: I am a theoretical/computational physicist. Maybe I am too much thinking about simulating heavy ion collisions or black hole mergers, but given the physical problem you want to solve, there are quite different numerical methods applied. This applies for fundamental concepts: Besides the traditional linear algebra/partial differential equation solving, there are statistical approaches and nowadays artificial intelligence for obtaining a solution to a problem in a shorter time. The ways how to model physical reality are as diverse as the physical problems.
It's a bit unclear who the target market is. Is it scientists? It doesn't seem to be for typical engineers who have a <i>lot</i> of very specific needs in FEA software. I can't see much aimed at them on the website. Examples of big difficult features that are useful for engineers are robust contact, automatic contact detection, plasticity, post-buckling, geometry defeaturing, automatic midplane extraction for shells, associativity with CAD models, being Ansys or Abaqus (because everyone knows and trusts them), not having to write scripts, not having to convert imperial units to a coherent system, not having to learn a huge complex pile of software (some tension with all the other requirements!), etc.<p>Easy to use FEA is a seemingly impossible goal because of the huge range of features that some users want and other users don't want as well as the fact that it has to combine a lot of concepts just don't seem to naturally fit well into any simple organizing principle for the user. Then there's the zillions of man-hours to do it all.
Don't really know how this ended up on the front page now, but the overall "idea" that I have yet to find anyone else understand and believe in (even my old prof and colleagues don't seem to "get it"), is that: wouldn't it be great if there was a unified, ideally simple, interface from which you could just select any solver by the click of a button, whether you wanted to use OpenFOAM, FEniCS, or maybe even Ansys, etc. Then you could set up your simulation model once, and run it on "any" and many solvers (to validate and check results etc.). So that's what I've been trying to do by myself for the past 7 years or so. For better or worse it initially seemed like a good idea to use Matlab/Octave, but have been planning to eventually move either to Python or Pascal/Delphi backend.