Potential inconsistency in MPG and docs

[Yuri Gribov]

Hi,

I might be reading it wrongly but meaning of RM_IMP and RM_PMI seems to be swapped in MPG (section 4.3.4 "Half reification") and ReifyMode docs (https://www.gecode.org/doc-latest/reference/group__TaskModelInt.html#ga8904e2a3f216089daeaa60652a6429a5).

MPG claims that

  half reified constraint 'b = 1 => x = y' ... can be posted by
  Reify r(b, RM_IMP);

whereas docs say that RM_IMP

  For a constraint 'c' and a Boolean control variable 'b' defines that 'b = 1 <= c' is propagated.

So which of them is correct?

-Yuri

[Filip Konvička]

Hi,

I believe it is just a notation/clarity issue.  The MPG is right and the "=>" in 'b = 1 => x = y' is a logical implication operator.

The 'b = 1 <= c' seems to me like a logical value constraint; so when "b = 1" holds, the "b=1" yields "1" (for true) and the half reification constraint enforces that "c" must also be true.

Does this make sense?

Cheers,

Filip

čt 26. 1. 2023 v 20:21 odesílatel Yuri Gribov <tetr...@gmail.com> napsal:

--
You received this message because you are subscribed to the Google Groups "Gecode" group.
To unsubscribe from this group and stop receiving emails from it, send an email to gecode+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/gecode/fe4e26a0-6bbd-4c34-99a5-b5b451d129ffn%40googlegroups.com.

[Filip Konvička]

P.S. - and now I noticed that the HTML docs actually use the implication symbol instead of "<=" which I thought was a "less-than" applied to constraint logical values.  So I think that the HTML docs need to be fixed.

Thanks!

Filip

pá 27. 1. 2023 v 13:56 odesílatel Filip Konvička <filip.k...@gmail.com> napsal: