[Clp] code sample for feasible point

[LMH]

As I understand it, an example of one of the problems I am trying to solve is,

minimize (over x): 1
s.t. Ax = P
     x^T * [1] = 1
     x[i] >= 0

where [A] = { x1, x2, xn }
x^T is the transpose of x
[1] is the all-1 vector

in this case for for the data,

-0.9422, 2.02568, -0.86967, 0.855568, -0.601532, 0.323411, 0.596289
-0.879254,-6.62066,-7.34114,0.585723,2.78475,-0.387155,2.99808
 8.42321,-3.20765,-1.0542,-4.1625,2.92884,-0.327967,3.50838
10.774,0.545379,-1.12841,-1.5456,-0.697098,-0.74527,-0.774246
 4.94671,0.26215,-3.77845,-0.191081,6.03136,3.66745,-6.22697
 8.72006,-4.73847,-0.93998,-3.85806,3.76628,0.499046,2.02952
 4.02597,-3.96582,-3.72078,-5.20739,7.33248,-0.41624,2.88221
 8.16238,-2.51468,-6.8052,-2.9021,2.28175,-2.72354,-0.313697
-0.0458385,-2.56746,-4.00668,-4.6656,8.62974,-1.00593,4.6638
 6.81399,-3.65807,-4.21049,-3.78946,4.7107,-1.2893,5.65669
-2.03821,-5.98271,-6.61833,-2.62184,5.20874,-1.79854,5.31864
 5.45169,-6.62274,-0.878738,-5.38384,4.09504,-0.626197,5.56643
 1.44436,-9.92098,4.99388,3.21483,-1.66346,3.64611,-0.0439071
-2.53507,-3.18565,6.33243,5.70343,5.6815,-5.98565,-0.239556
-1.06284,-6.71054,-7.89752,-1.71425,5.42546,-1.42733,5.34596
 9.96752,-5.18027,-1.43649,-4.85739,3.4546,-0.765721,3.39876
 6.14065,-6.80276,-4.45474,-4.45472,4.96714,-1.12068,5.68639
 0.789644,4.46502,0.667421,-0.753405,11.1041,5.58855,-2.47415
 3.4069,-2.47077,-6.1166,-3.97326,10.5417,1.11918,0.915666
 9.90365,-3.71723,-4.98727,-4.18883,5.12366,-2.19175,2.14106
 7.50469,-6.45069,-1.76295,-5.8613,4.92941,-1.28936,5.62019
 0.784531,-4.41507,7.81695,6.00103,5.11466,-7.41083,-0.859709
 8.76947,-6.89332,-2.22446,-5.56305,4.37828,-0.904622,5.32845
10.361,-5.37848,0.280192,6.50686,0.295681,5.4023,0.903526
 4.23739,-2.91606,-5.48661,-4.01339,-0.250081,-9.08589,-7.44781
10.3283,-2.70868,2.08743,9.13842,0.496487,4.09732,0.356068
 6.29274,-10.344,3.09637,3.86458,6.52654,-4.4275,2.03636
 2.77549,-5.52054,-10.1449,-5.1293,5.97665,-5.38324,2.84052
-1.53211,-8.28862,5.89797,8.91702,5.26826,-7.60942,-1.76916
 6.55035,-5.0121,-9.87386,-5.39377,3.79171,-10.0618,-3.29243
 6.16032,-9.05789,-5.40621,12.851,-0.897329,8.45366,2.49965
-2.0535,-7.28366,8.4009,10.4692,8.09746,-13.0538,-2.21631
-0.100108,-2.19943,-5.25283,0.665639,-2.74879,-1.88559,-2.01111
-6.68807,-2.0858,-1.67507,1.09343,-0.425115,1.13627,1.38262
 5.52471,6.96936,-0.195879,2.15077,0.580036,-2.21135,5.11126
 3.03433,-5.15217,-1.53586,-6.28597,4.69399,-1.2563,6.01448
 0.143787,-1.69845,-3.46485,-3.37431,-1.68215,-8.23386,-7.67879
11.971,1.98691,-2.03594,5.68318,-1.72722,2.49254,0.985464
 7.28228,-12.0396,-4.96597,14.202,-1.74251,8.69396,2.39749
 7.82083,-8.19497,-13.3157,-5.08318,9.88666,-5.91352,7.07704
 0.329034,-11.9767,9.11734,15.1535,12.8102,-17.7227,-2.09866

where the first row is not part of the convex set of the total data.

I believe that I understand how to declare a model and add rows and data but I don't
understand how to add the constraints listed above. I am not sure where the all-1
vector goes or what its dimensions should be. I don't understand what the "objective"
of the model is. It seems as if there should be two data objects here. One with the
first row of the data above and a second with the rest of the data and the model
testing if the first row fits into the solution of the rest.

Can anyone help with this? Am I not even close?

LMH



LMH wrote:
> Hello,
> 
> I need to add some code to a program that will take a set of data points and
> determine the feasible point. More specifically, I think I only need to know if the
> model is feasible or not. Something like isProvenPrimalInfeasible() is probably
> sufficient.
> 
> Currently, my data is stored in a vector of row objects where each objects holds a
> vector<double> with the column values for the row. I need to translate this data to
> the proper clp container (it looks like clp uses its own matrix class,
> ClpPackedMatrix), construct a model, and then retrieve weather or not the model is
> feasible or infeasible.
> 
> I have some sample data where all points are part of a convex set where the model
> should be feasible, and another set where one point is not part of the convex set and
> the model should be infeasible. The data is in R7 if that matters. This version of
> the data is not scaled but I can do that if it would be helpful.
> 
> I have read most of the users guide and understand what allot of the methods do.
> 
> Can I just do something like what is shown here,
> 
> https://urldefense.proofpoint.com/v2/url?u=https-3A__projects.coin-2Dor.org_Clp_browser_trunk_Clp_examples_addRows.cpp&d=DwICAg&c=Ngd-ta5yRYsqeUsEDgxhcqsYYY1Xs5ogLxWPA_2Wlc4&r=G-rWCwMCBm3eUuw9HwVHnyOeSnSun6y799F3zRs5-x4&m=xIVNhd8I_SwjEMcx2vtdWlw2EbNdQHXPLVrTrvdOaQ8&s=LgeCQqzGcQj_rHVBNMgcnKvf5vk_DYRouZ-amKezzK0&e=
> 
> which is
> 
> - declare an empty model
> - resize the model to the proper number of columns
> - declare a row array
> - populate the row array
> 
> If this is correct, there are a few things I don't understand.
> 
> What is the purpose of the objective coefficient set on lines 35-41?
> 
> In lines 47-50 where the data is added to each row, it doesn't look like the two rows
> have the same dimensions. This doesn't make sense. The code starting on line 155
> looks more reasonable.
> 
> Is there not a way to just declare and populate a matrix and pass that to a model
> method? The code starting on line 199 likes more like this, but all of the different
> implementations in this sample are a bit confusing.
> 
> LMH
> _______________________________________________
> Clp mailing list
> Clp at list.coin-or.org
> https://urldefense.proofpoint.com/v2/url?u=https-3A__list.coin-2Dor.org_mailman_listinfo_clp&d=DwICAg&c=Ngd-ta5yRYsqeUsEDgxhcqsYYY1Xs5ogLxWPA_2Wlc4&r=G-rWCwMCBm3eUuw9HwVHnyOeSnSun6y799F3zRs5-x4&m=xIVNhd8I_SwjEMcx2vtdWlw2EbNdQHXPLVrTrvdOaQ8&s=tlvDeytN6AJ-dFWUzJiOqVZh5IohQp0tQtu99WKWo7c&e=
>