[Dip] Phase 1 Bound history

[Matthew Galati]

Hi Kim,

I implemented this in trunk. I cleaned up the phase history quite a bit and
added a column for phase - so you can distinguish 1 vs 2. I added a new
parameter "LogObjHistory" so you can toggle it on (LogObjHistory=1) separate
from the other log levels.

I might add a "phase 3" for when the price+branch heuristic runs. I'll do
this later.

I haven't had much time to test it - but on some basic examples, it seems to
work. Feel free to give it a try.

I'll test it more soon. Let me know if the new format looks OK.

Matt

Example:

========== OBJ History Node 0 [BEGIN]: ====================================
Phase Cut   Price Time      ThisLB    BestLB    ThisUB    ThisIP    BestIP
1     0     1     0.031     -619.56   -619.56   281.14     INF       INF
1     0     2     0.038     -10.61    -10.61    1.46       INF       INF
1     0     3     0.042     -2.36     -2.36     0.47       INF       INF
1     0     4     0.048     -2.18     -2.18     0.21       INF       INF
1     0     4     0.049     -2.18     -2.18     0.00       INF       INF
2     0     5     0.058     2203.56   2203.56   4472.65    INF       INF
2     0     6     0.075     1889.35   2203.56   4262.89    INF       INF
2     0     7     0.087     765.51    2203.56   3920.28    INF       INF
2     0     8     0.093     1339.63   2203.56   3830.60    INF       INF
2     0     9     0.108     953.78    2203.56   3411.36    INF       INF
2     0     10    0.117     2453.32   2453.32   3157.57    INF       INF
2     0     11    0.129     1806.68   2453.32   3040.10    INF       INF
2     0     12    0.137     2409.12   2453.32   2958.03    INF       INF
2     0     13    0.142     2467.32   2467.32   2925.39    INF       INF
2     0     14    0.151     2554.71   2554.71   2886.51    INF       INF
2     0     15    0.160     2542.20   2554.71   2842.33    INF       INF
2     0     16    0.171     2716.49   2716.49   2829.79    INF       INF
2     0     17    0.178     2692.15   2716.49   2815.30    INF       INF
2     0     18    0.189     2677.33   2716.49   2800.87    INF       INF
2     0     19    0.197     2709.81   2716.49   2790.81    INF       INF
2     0     20    0.206     2704.59   2716.49   2778.01    INF       INF
2     0     21    0.216     2720.29   2720.29   2776.49    INF       INF
2     0     22    0.226     2692.33   2720.29   2774.58    INF       INF
2     0     23    0.236     2714.38   2720.29   2773.82    INF       INF
2     0     24    0.245     2766.16   2766.16   2771.85    INF       INF
1     1     25    0.252     -2.25     -2.25     0.75       INF       INF
1     1     26    0.255     -1.73     -1.73     0.20       INF       INF
1     1     27    0.263     -0.82     -0.82     0.12       INF       INF
1     1     28    0.269     -0.29     -0.29     0.05       INF       INF
1     1     28    0.277     -0.29     -0.29     0.05      4691.00   4691.00
1     1     29    0.278     -0.31     -0.29     0.04       INF      4691.00
1     1     29    0.279     -0.31     -0.29     0.04      4691.00   4691.00
1     1     29    0.279     -0.31     -0.29     0.00       INF      4691.00
2     1     30    0.286     1252.06   1252.06   3729.79    INF      4691.00
2     1     30    0.287     1252.06   1252.06   3729.79   4691.00   4691.00
2     1     31    0.294     1539.67   1539.67   3598.28    INF      4691.00
2     1     31    0.295     1539.67   1539.67   3598.28   4691.00   4691.00
2     1     32    0.305     1973.85   1973.85   3502.56    INF      4691.00
2     1     32    0.306     1973.85   1973.85   3502.56   4691.00   4691.00
2     1     33    0.320     1798.31   1973.85   3222.31    INF      4691.00
2     1     33    0.321     1798.31   1973.85   3222.31   4691.00   4691.00
...




On Wed, Feb 9, 2011 at 6:30 AM, kim <kim at hamilton-vejlin.dk> wrote:
>
> I can easily access the phase 2 bound history easily through the
> m_nodeStats member of the DecompAlgo class, but is the phase 1 bound
> history recorded and accessible anywhere?
>
> I realize the bounds of the phase 1 problem are not directly useful for
> problem solving, but I think I am generating phase 1 problems with a lot
> of degeneracy in them and would like to plot the bound history for
> confirmation.
>
> - Kim
>
> _______________________________________________
> Dip mailing list
> Dip at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/dip
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://list.coin-or.org/pipermail/dip/attachments/20110223/22054f6f/attachment.html