Most performant way for boolean-AND constraint

[adihe]

Hello all

I am using the following code in CP-SAT to implement X = A && B.

IntVar x = model.NewBoolVar("");

model.Add(x == 0).OnlyEnforceIf(new ILiteral[] { a.Not() } );

model.Add(x == 0).OnlyEnforceIf(new ILiteral[] { b.Not() });
model.Add(x == 1).OnlyEnforceIf(new ILiteral[] { a, b });

Is this the optimal way or is there a more performant approach?

Thank you for your feedback in advance.

Best regards,

Adrian

[Laurent Perron]

model.AddBoolOr(new ILiteral[] {a.Not(), b.Not(), x });

model.AddImplication(x, a);

model.AddImplication(x, b);

--
You received this message because you are subscribed to the Google Groups "or-tools-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to or-tools-discu...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

--Laurent

[adihe]

Thank you very much Laurent.

It works like a charm.

For my understanding: 

I checked the documentation for AddImplication at https://github.com/google/or-tools/blob/master/ortools/sat/doc/reference.md but I could not find details on why it would/could be potentially faster.

Could you give me a hint why it is so?

Thank you very much

Adrian

[Laurent Perron]

your model is not Boolean, you add integer linear equations. Especially with the enforcement with 2 literals. they should be presolved into something simpler (I need to check).

This is way to complex for my liking. 

The fastest you can get is by using clauses (BoolOr). Implications will be transformed because a => b is actually a.Not() or b.

[adihe]

Thank you Laurent.

[wtv...@gmail.com]

model.AddImplication(x, a);

model.AddImplication(x, b);

It seems means X => A && B but not X = A && B?

[Laurent Perron]

model.AddBoolOr(new ILiteral[] {a.Not(), b.Not(), x }); 

implements a && b => x

the two implications implement the opposite direction.

Laurent Perron | Operations Research | lpe...@google.com | (33) 1 42 68 53 00

[wtv...@gmail.com]

Sorry, I forget this sentence. I have one puzzle when I test my code.

I have a constraint base <=> pack & flag.  

when I write this code,  it doesn't work.

cpPack.AddBoolAnd([pack, flag]).OnlyEnforceIf(base)

cpPack.Add(base == 1).OnlyEnforceIf([pack, flag])

then I change the code to following, it works.

base = cpPack.NewBoolVar("base")

flag = cpPack.NewBoolVar("flag")
cpPack.Add(z[j] - ze[i] == 0).OnlyEnforceIf(flag )
cpPack.Add(z[j] - ze[i] != 0).OnlyEnforceIf(flag.Not())
cpPack.AddBoolAnd([pack, flag]).OnlyEnforceIf(base)

cpPack.AddBoolOr([pack.Not(), flag .Not()]).OnlyEnforceIf(base.Not())
# cpPack.Add(base == 1).OnlyEnforceIf([pack, flag])

What is the difference between them? Do not two half-reified constraints equal to a full-reified constraint in ortools?

[Laurent Perron]

It should be equivalent.

On tested on master, and it works fine.

Laurent Perron | Operations Research | lpe...@google.com | (33) 1 42 68 53 00

--