[Cbc] Problems with reading large model from file

[Bjørn Sigurd Johansen (Spider Solutions AS)]

Hi,

FYI: The problems I had with reading large model from file seem to be clarified / have a workaround.


After having ran around 10.000 different optimization problems, I think it’s safe to conclude as follows:

  *   The CBC file input parser does not always properly handle file inputs containing around 9.000 characters in a single line. At least not for objectives and/or constraints.
  *   The CBC file input parser seems to never get into line length problems if maximum line length is 255 characters.

(I have not investigated line lengths between 255 and 9000 characters).


If you need to split long lines (i.e. insert additional line breaks), note the following:

  *   Don’t split a coefficient and a model variable.  I.e. split on the space character before the coefficient.
  *   There is no need the keep the space character before the coefficient. I.e. the space may be substituted with a line break.
  *   I don’t know if a limit may be split.  (E.g. if “<= +123.45” may be split on the space character).



Thanks,
Bjorn
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/cbc/attachments/20191220/8ca82d68/attachment.html>