MathGroup Archive 2010

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: How export/recompile expressions and calculation code to run

  • To: mathgroup at smc.vnet.net
  • Subject: [mg111418] Re: How export/recompile expressions and calculation code to run
  • From: Albert Retey <awnl at gmx-topmail.de>
  • Date: Sat, 31 Jul 2010 02:42:46 -0400 (EDT)
  • References: <i2ub5f$jfb$1@smc.vnet.net>

Hi,

> We have some Mathematica program (with Mathematica-based GUI) and we
> need to sell it to customers.

> We would like to make our program independent of Mathematica (on
> running time).

if you have everything including a GUI ready, isn't the Player Pro a
reasonable alternative? You would just have to create an encoded package
from your code and it would cost something around 300$ per license (you
should talk to WRI for details).

> I see two opportunities:
> 1) We can Export all key mathematical expressions from mathematica to
> java classes and then use some java numerical libraries to carry out
> the calculations (+ GUI on java).
> 2) We can Export all key mathematical expressions from mathematica to
> some free and powerful package for numerical calculations: like Scilab
> and R. And then write a program to carry out the calculations (+GUI).
> 
> Any suggestions?

It really depends a lot on what functionality that Mathematica provides
you are using in your code. What kind of numerical routines do you use
(NSolve,FindRoot,NDSolve,NMinimize,Fourier...)? Do you use any symbolic
methods? Do you use any of the visualization functionality, if so which?

What are the priorities and boundary conditions for creating such an
application: target time to market, dependencies on other products/ease
of installation, target platforms, costs, execution speed, prior
knowledge and preferences of developers...

Only if you can provide more details a good choice can be suggested, in
principle you can do everything with any of the possibilities (and a
couple others), it is just a matter of how much time and effort you are
willing to spend.

My personal experience tells me that the knowledge of the developers and
their preferences is crucial to success: results will be so much better
if they try to proof their suggestion will work than when they try to
proof something they don't like will not work :-)

hth,

albert


  • Prev by Date: Re: Surface integral on a 3D region
  • Previous by thread: ODE solution transfer