Changeset 1717

Timestamp:
Sep 29, 2009 12:31:01 PM (11 years ago)
Author:
wehart
Message:

Merged revisions 1692-1715 via svnmerge from
https://software.sandia.gov/svn/public/coopr/trunk

........

r1692 | jwatson | 2009-09-23 09:03:44 -0600 (Wed, 23 Sep 2009) | 5 lines


Promoted the efwriter to a full script, stored in coopr/scripts. Deleting all local drivers.


Some misc mods to network flow pysp example that I forgot to check in.

........

r1693 | jwatson | 2009-09-23 13:21:54 -0600 (Wed, 23 Sep 2009) | 9 lines


Fixed an issue in the WW PH extension involving variable slamming. During initialization, variable values are assigned the associated stage for purposes of slamming to force convergence.


When picking a variable to slam, the loop looks for the "_stage" attribute - which is only defined for variables appearing in all but the last stage.


However, the specification for variable slamming in wwph.suffixes input files may necessarily inadvertently (for reasons of simplicity) specify variable values in the final stage, e.g., foo[*,*].


I now check for the existence of the "_stage" attribute before blinding accessing it.

........

r1694 | jwatson | 2009-09-23 14:37:14 -0600 (Wed, 23 Sep 2009) | 3 lines


Inverted the order in which cycles are detected in the WW PH extension - we want to report the longest observed cycle, which are indicative of real issues.

........

r1695 | jwatson | 2009-09-24 09:04:24 -0600 (Thu, 24 Sep 2009) | 5 lines


Modified the WW PH extension to (1) threshold the mipgap to 1.0, the maximal allowed CPLEX value and (2) spit out a warning if the mipgap for any iteration is greater than the initial mipgap.


While both shouldn't happen given nice/sane inputs, we don't always provide nice/sane inputs!

........

r1696 | jwatson | 2009-09-24 10:03:37 -0600 (Thu, 24 Sep 2009) | 3 lines


On David's advice, simply threshold mipgap in the WW PH extension to a maximum of the initial mipgap. If we go above this value in a subsequent PH iteration, the behavior is pathalogical and should be treated/flagged as such.

........

r1697 | jwatson | 2009-09-24 10:52:19 -0600 (Thu, 24 Sep 2009) | 3 lines


Fixed a bug in index-template matching occuring in one-dimensional variable manipulation.

........

r1698 | jwatson | 2009-09-24 14:02:01 -0600 (Thu, 24 Sep 2009) | 5 lines


Changing the order of valid CBC problem formats from MPS-NL-CPXLP to NL-CPXLP-MPS. The solver plugins by default take the first specified, and if that's MPS, then we have an issue translating from MPS-encoded names back to the pyomo variable names. With NL and CPXLP, there isn't an issue.


In the long term, we need to at a minimum modify pico_convert to store an auxilliary map file from which we could properly store the results from an MPS-based solve.

........

r1699 | jwatson | 2009-09-24 16:18:58 -0600 (Thu, 24 Sep 2009) | 3 lines


Fixed one-d index-versus-tuple issue in another location (isolated to pysp).

........

r1700 | dlwoodr | 2009-09-24 21:29:05 -0600 (Thu, 24 Sep 2009) | 2 lines


add a parameter to allow multiple PH iterations between cycle induced slams

........

r1701 | jwatson | 2009-09-25 10:23:28 -0600 (Fri, 25 Sep 2009) | 3 lines


Missed pyspbody.tex in some previous check-in. Whoops.

........

r1702 | jwatson | 2009-09-25 10:29:11 -0600 (Fri, 25 Sep 2009) | 3 lines


Fixed minor issue with latest WW PH checkin.

........

r1703 | dlwoodr | 2009-09-25 12:12:29 -0600 (Fri, 25 Sep 2009) | 2 lines


Fix the errors in pyspbody.tex; there are still some overflows in file displays done via {verbatim}

........

r1704 | dlwoodr | 2009-09-25 12:41:22 -0600 (Fri, 25 Sep 2009) | 2 lines


better wwph.cfg for the sizes tutorial

........

r1705 | jwatson | 2009-09-25 14:12:28 -0600 (Fri, 25 Sep 2009) | 3 lines


Restructuring of pysp network flow example.

........

r1706 | dlwoodr | 2009-09-25 14:35:07 -0600 (Fri, 25 Sep 2009) | 2 lines


add the data files and update the reference model to use continuation lines

........

r1707 | dlwoodr | 2009-09-25 15:29:10 -0600 (Fri, 25 Sep 2009) | 5 lines


self.InitialMipGap? = 0.10

--This line, and those below, will be ignored--


M pysp/wwphextension.py

........

r1708 | dlwoodr | 2009-09-25 15:31:44 -0600 (Fri, 25 Sep 2009) | 2 lines


spell check, better formatting for sizes {verbatim} example files

........

r1709 | dlwoodr | 2009-09-25 15:59:10 -0600 (Fri, 25 Sep 2009) | 2 lines


fix a few errors

........

r1710 | dlwoodr | 2009-09-25 16:02:05 -0600 (Fri, 25 Sep 2009) | 2 lines


I'm going to quit looking for little mistakes now so as to avoid finding any more today

........

r1711 | jwatson | 2009-09-26 17:09:54 -0600 (Sat, 26 Sep 2009) | 7 lines


Forgot to add ReferenceModel?.py in the network flow restructure.


Modified param.py to correctly handle case where singleton parameters are initialized via a rule. Before, they (triggered by a network flow model simplification) were triggering a failure (.value being None) in the presolve.


Updated the 1ef10 extensive form based on the new maximal-M parameter definition.

........

r1712 | jwatson | 2009-09-26 19:29:38 -0600 (Sat, 26 Sep 2009) | 3 lines


Added more instances of the pysp network flow test problem.

........

r1713 | jwatson | 2009-09-28 07:43:41 -0600 (Mon, 28 Sep 2009) | 3 lines


Miscelleanous cleanup - with CHILE2 pysp forestry example stable/tested, eliminating older Davis-originated CHILE example.

........

r1714 | jwatson | 2009-09-28 07:58:54 -0600 (Mon, 28 Sep 2009) | 3 lines


Re-did the pysp CHILE forestry example with label-based time stages, in prep for testing with set "ord" functionality.

........

r1715 | jwatson | 2009-09-28 09:09:09 -0600 (Mon, 28 Sep 2009) | 5 lines


Added an "ord" method for sets, to compute the positional index of an input element in an ordered set.


Added time-discounting of money to pysp forestry example (CHILE instance) to use it!

........

Location:
stable/1.2
Files:
24 deleted
13 edited
116 copied

Note: See TracChangeset for help on using the changeset viewer.