[sdpd] Any idea ?

[Armel Le Bail <armel...@fluo.univ-lemans.fr>]

Hi,

Improving the efficiency of molecule location software or
structure solution starting from random configurations
programs like ESPOIR may just be a question of algorithm.

For instance, in scratch mode (random starting configuration),
it is clear that the more complex the structure is, and the less
a simple move of one atom will have effects on the cost function.
One may imagine to move randomly 2 atoms or more instead
of one and then perform a test on the R factor in order to see if
keeping those moves is advantageous. Hard to predict if this will
lead to any improvement without trying.

Many other strategies can be imagined. Only time is lacking
for testing them.

Any genius suggestion, idea ?

Armel

PS1 - remember that Espoir is GNU Public License, any
contribution is welcome.

PS2 - BTW, a Linux version is available, thanks to Lubomir Smrcok :
http://www.cristal.org/sdpd/espoir/index.html#linux