CPLEX error message: CPLEX 20.1.0.0: logical constraint _slogcon[1] is not an indicator constraint

Support
,
1

Dear Support,

I dont know why is this interpreted as a logical constraint (it seems, this is the case):

subject to CONVEXITY {i in conv_bid_sets} :
(sum {j in start_index[i]…end_index[i] }
(X[j])<1+eps);

In the .mod file I have

param: start_index end_index :=
1 1 2
2 3 4
3 6 7
4 8 9
5 10 11
6 12 13
aggrEU_1.dat (753 Bytes)
Uploading: aggrEU_1.mod…

Thank You

1 Like
2

Hi @David_Csercsik,

Since the entire constraint is between parenthesis there is no left/right hand side of the constraint and in that case it is considered to be logical. The following should work:

subject to CONVEXITY {i in conv_bid_sets}:
    sum {j in start_index[i]..end_index[i]} X[j] < (1+eps);
3

Hi, @fdabrandao

I see, and thank You,…

So if I understand right, equality constraints are logical, right?

I have:

balance

subject to BALANCE {t in 1…T, dp_index in 1…num_DP} :
(sum {i in order_numbers: DP[i] == dp_index && period[i] = t}
(X[i]*quantity[i])=0);

I was probably mislead by this…

4

Hi @David_Csercsik,

Equality constraints are regular constraints as they have left and right hand sides, that constraint there is logical because of the parenthesis too:

If you want a constraint that ensure that the total sum is 0, you can do that as follows:

subject to BALANCE {t in 1…T, dp_index in 1…num_DP} :
       sum {i in order_numbers: DP[i] == dp_index && period[i] = t} X[i]*quantity[i] = 0;

sum {i in order_numbers: DP[i] == dp_index && period[i] = t} X[i]*quantity[i] is the left hand side, and 0 is the right hand side.

As an example of logical constraints you can see for instance the following N-Queens example where the three constraints are logical:

param n integer > 0; # N-queens
var Row {1..n} integer >= 1 <= n;
s.t. row_attacks: alldiff ({j in 1..n} Row[j]);
s.t. diag_attacks: alldiff ({j in 1..n} Row[j]+j);
s.t. rdiag_attacks: alldiff ({j in 1..n} Row[j]-j);
5

Ok I see, and thank You again. The strange thing is that it kinda worked with the parentheses as well as intended… (at least it seemed so).

6

Strict comparisons are interpreted as logical constraints because they cannot be satisfied algebraically exactly (you always need some eps.) Suggest formulating as <= 1-eps, then it’s algebraic.