Re: [sdpd] The (small) budget makes the law

[Armel Le Bail <alb...@cristal.org>]

Lubo wrote :

>going through yesterday's e-mails I missed one more quantity : you count
>minutes for that or this program, but where is the time for THINKING of
>the problem ? Or is it done blindlessly by MC method :-) ?

Time is counted for all of us. And if you want to save time, don't
read that mail !

This is certainly what Lachlan wants : more time for thinking by
reducing the programs setup time. IMHO, even with 10mn for ESPOIR,
the longer part of the whole setup time (not for me), he should keep
enough thinking time during the programs execution.

Several things were lacking in my "how many time list" :
ESPOIR is declared to be a last chance program, so that,
one of these things is :
  - How many time for failing in classical Patterson and
    Direct methods (or in structure prediction for Lubo) ?

You may argue that "failing" does not occur during the
programs setup time, however, setup will have to be made
again and again with different setup parameters. That is
true also with indexing of search-match, etc.

My own minimum average time for realizing a simple 10 atoms
SDPD is 1440 minutes (synthesis not included ;-).
And then, I need 43200 minutes before submitting a paper
for publication, at least. But my maximum time for a SDPD
seems to be unlimited (I have still examples for that in my drawer
which I cannot propose for a Round Robin : I am even not
sure that they are single phases).

Writing the tutorial showing how setting up ESPOIR in 10
minutes will take probably > 120 minutes (I am a bit meticulous ;-).

About time, in 1998 for the SDPD Round Robin, people
had 43200 minutes for solving 2 structures. And only 2
succeeded : this talk more about the real time necessary
for a SDPD. Two years later, may be more people would
succeed, but this is not sure since not much more
software are available at reasonable cost (or for free).

The Round Robin organizers would have to give 2 months
instead of one maybe...

The Size/Strain Round Robin gave also one month, and
the deadline was October 20. But some people have asked for
more time already... And this is not the program setup which
needs time, it is the thinking time allowing to produce the
correct setup, the one which will be successful at execution
time.

Probably, if artificial intelligency progress up to give
us mechanical slaves, we will have then more time for
thinking. The bad side is that if this program really succeeds,
then those slaves may well think better than us !
I am not sure if we should speed up the setup time or slow
down it... Is that not great thinking (thanks to only 10mn for
setting up ESPOIR, much shorter than the time for drinking
a beer during a workshop) ??

Best,

Armel Le Bail
http://www.cristal.org/course/


-------------------------- eGroups Sponsor -------------------------~-~>
eLerts
It's Easy. It's Fun. Best of All, it's Free!
http://click.egroups.com/1/9699/14/_/40740/_/972371022/
---------------------------------------------------------------------_->