nancy collins wrote:
> sorry to have to say this, but turning positions into doubles
> is an almost impossible idea.  almost every dx module
> opens the positions and reads through them, using floating
> point local variables.   it would touch thousands of lines
> of code.

Apologize for bothering again with this, but I'm missing something.
I assume I can safely cast from my double positions
to a float position. (I could insert a check and raise an error if this
is not the case. Is box.c the right place for such check?)
Then every dx module shold be able to read through the position array
and store positions in float local variables.

> 
> but it is possible to make your own components.  you could
> make a field with a "double positions" component in your
> input file, and use it for saving/starting/restarting your simulations.
> 

Yes, but I will end up with duplicated data. And I will have
to manage the cast from double to float in my program when writing
positions.

Maybe the solution I'm looking to is to modify the dx importer
in suche a way that, when it reads from a binary file a double array
it stores it in memory as a float array?

Thanks again for your suggestions,

Marco Morandini

Reply via email to