Thanks for you answer Thomas.
I was missing the fact that it was not 10^e but 2^ e, and i was missing my 
eyes too, you are right.

Just one last question for personal knowledge, why if ±3.4028235×10^38 is 
max/min finite number, -1*10^10 return empty ?



Really sorry if i bother the bandwith with this thread, it was really not my 
purpose. I was just questionning myself about precision.

Just for my defence, this is not so esoteric i hope :
I agree totally, it is special case (Polynomial fitting for exemple, or like 
on what i am working compute coefficients for autoregressive model) which 
need to be traited by script, dll,...
But it doesn't matter how many pixels you got on the chart. If you have many 
recursion wich increase rounding error you can see the difference even in 
640*400 (i.e: PolyFit fred posted, if in double precison VBS there was need 
to center X axis on zero ! and just changing this translation you can see 
diff. on the chart).
I am aware that more precision is more slower, and prefer many times the way 
you make Amibroker work... it is so fast ! Thanks Thomas. It is the more 
important for trading purpose.


Cheers,
Mich.

_________________________________________________________________
Retrouvez tout en un clin d'œil avec Windows Desktop Search ! 
http://desktop.msn.fr/

Reply via email to