forum.alglib.net
http://forum.alglib.net/

Tolerance for quadrature and dll export
http://forum.alglib.net/viewtopic.php?f=2&t=883
Page 1 of 1

Author:  MikeH [ Mon Jul 15, 2013 10:29 am ]
Post subject:  Tolerance for quadrature and dll export

Hi,

Just downloaded and trying out ALGLIB - so far it is working very well. Very pleased with how easy it was to build and use (though see my 2nd question below).

However I have a couple of queries.

1)
When using the autogkintegrate function, I gather it calculates at close to machine precision. This is really too much for me, I would prefer less precision and greater performance. I edited line 3395 in integration.cpp...

state->eps = 100000*ae_machineepsilon

... to increase the multiplier, it seemed to work just fine and give me acceptable results in a much quicker time.

However I don't want to really make a global change like that as the level of precision required will be context dependent. How easy would it be for me to set up a method SetPrecision or something similar? Or are there plans to add functionality like this in the near future?

2) I have built alglib as a separate dll - but how then can I easily export all the required classes and functions? Even for the single integration there are 2 classes and 3 functions. I don't want to make extensive modifications to headers etc as I would have to merge in later releases. What I have done so far is to include the necessary source code in my project, but if I get to the point of having several projects which use alglib this isn't really a good solution.

Any thoughts or ideas on either of these would be very welcome.

Many thanks,

Mike

Page 1 of 1 All times are UTC
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/