FMB is returning partial/incorrect result

[mmaroti]

Just searching for all 2-element poset and encoding the problem as

tff(declared_dom, type, dom:$tType).
tff(declare_dom0, type, dom0: dom).
tff(declare_dom1, type, dom1: dom).
tff(dom_elements, axiom, ![X:dom]: (X=dom0 | X=dom1)).
tff(dom_distinct, axiom, $distinct(dom0, dom1)).
tff(declare_rel, type, rel: (dom * dom) > $o).
tff(axiom6, axiom, ![X:dom]: rel(X,X)).
tff(axiom7, axiom, ![X:dom, Y:dom, Z:dom]: ((rel(X,Y) & rel(Y,Z)) => rel(X,Z))).

Running this with vampire -sa fmb gives the result with :

tff(predicate_rel,axiom,
           rel(dom0,dom0)
%         rel(dom0,dom1) undefined in model
         & ~rel(dom1,dom0)
%         rel(dom1,dom1) undefined in model
).

This is not a model, as the undefined values cannot be arbitrarily set to any value. How can I force the output relation to be fully defined?

[quickbeam123]

Thanks for the report, Miklos!

Restoring the values of eliminated symbols is an open issue that we plan to resolve within the next couple of months. (We will probably we able to point you to workarounds sooner, but current master has a small issue with model printing in general. Please stay tuned!)