CODEWITHSUNDEEP
pythonmathematical-optimizationscipoperations-researchpyscipopt
orpanter
orpanter

Reputation: 115

Can I set the SCIP constraint handler to work only after a feasible solution is found?

I read the SCIP constraint handler documentation, and reads that CONSHDLR_CHECKPRIORITY= -1 makes that the constraint handler only works when the solution is integral. However, my handler is still adding constraints before finding a feasible solution. So, even if the solution is integral, is not satisfying all the constraints.

I couldn't find another parameter to set the constraint handler to start only after the presolve is over and a feasible solution is found, or to only add constraints when the solution satisfies all the model constraints.

Is there a specific parameter to set this? or some trick to put with restriction?

Upvotes: 0

Views: 306

Answers (1)

Leon
Leon

Reputation: 1688

You simply need to give your constraint handler the lowest Priority out of all the constraint handlers (both for checking and enforcing). That way your constraint handler will only be called when all other constraint handlers have acted.

You can see all the default priorities by typing disp conshdlrs in the interactive shell. The default values are

 constraint handler   chckprio enfoprio sepaprio sepaf propf eager prestim description
 ------------------   -------- -------- -------- ----- ----- ----- ------- -----------
 benderslp            10000000 10000000        0    -1    -1   100     fme  constraint handler for Benders' Decomposition to separate LP solutions
 integral                    0        0        0    -1    -1    -1     fme  integrality constraint
 cardinality               -10      100       10    10     1   100     f    cardinality constraint handler
 SOS1                      -10      100     1000    10     1   100      m   SOS1 constraint handler
 SOS2                      -10      100       10     0     1   100     f    SOS2 constraint handler
 varbound              -500000  -500000   900000     0     1   100     fm   variable bounds  lhs <= x + c*y <= rhs, x non-binary, y non-continuous
 knapsack              -600000  -600000   600000     0     1   100     fme  knapsack constraint of the form  a^T x <= b, x binary and a >= 0
 setppc                -700000  -700000   700000     0     1   100     fme  set partitioning / packing / covering constraints
 linking               -750000 -2050000   750000     1     1   100      m   linking constraint x = sum_{i=1}^{n} c_i*y_i, y1+...+yn = 1, x real, y's binary
 or                    -850000  -850000   850000     0     1   100      m   constraint handler for or constraints: r = or(x1, ..., xn)
 and                   -850100  -850100   850100     1     1   100     f e  constraint handler for AND-constraints: r = and(x1, ..., xn)
 xor                   -850200  -850200   850200     0     1   100     fme  constraint handler for xor constraints: r = xor(x1, ..., xn)
 conjunction           -900000   900000        0    -1    -1   100     f    conjunction of constraints
 disjunction           -900000  -950000        0    -1    -1   100     f    disjunction of constraints (or(cons1, cons2, ..., consn))
 exactsol              -999999 -9999999        0    -1    -1   100     fme  constraint to ensure that primal solutions report back exact solutions
 linear               -1000000 -1000000   100000     0     1   100     f e  linear constraints of the form  lhs <= a^T x <= rhs
 orbisack             -1005200 -1005200    40100     5     5    -1       e  symmetry breaking constraint handler for orbisacks
 orbitope             -1005200 -1005200    40100    -1     1    -1      m   symmetry breaking constraint handler relying on (partitioning/packing) orbitopes
 symresack            -1005200 -1005200    40100     5     5    -1       e  symmetry breaking constraint handler relying on symresacks
 logicor              -2000000 -2000000    10000     0     1   100     fme  logic or constraints
 bounddisjunction     -3000000 -3000000        0    -1     1   100     f    bound disjunction constraints
 cumulative           -3030000 -2040000  2100000     1     1   100     fme  cumulative constraint handler
 nonlinear            -4000010      -60       10     1     1   100     fme  handler for nonlinear constraints specified by algebraic expressions
 benders              -5000000     -100        0    -1    -1   100     f    constraint handler to execute Benders' Decomposition
 pseudoboolean        -5000000 -1000000        0    -1    -1   100      m   constraint handler dealing with pseudo Boolean constraints
 superindicator       -5000000 -5000000        0    -1     1   100      m   constraint handler for indicator constraints over arbitrary constraint types
 indicator            -6000000     -100       10    10     1   100     f    indicator constraint handler
 countsols            -9999999 -9999999        0    -1    -1   100     fme  constraint to count feasible solutions
 components           -9999999        0        0    -1     1    -1          independent components constraint handler

Upvotes: 2

Related Questions