E comes with a multitude of options, many of which affect proof search and interact in often unexpected ways. The easiest way to get good performance out of the prover is to rely on the automatic mode. For a problem specified in the current TPTP-3/TSTP syntax, the suggested command line (as of E 1.9) is
eprover --auto problem.p
If you have more time than memory, you can also add
--memory-limit=Auto
or
(e.g.) --memory-limit=1024
. Note that E will
typically perform much better in automatic mode than in the
default mode.
Even more problems will typically be solved with strategy scheduling. Use
eprover --auto-scheduling=<cores> --cpu-limit=300 problem.p
<cores>
is the number of CPU cores to be
used. Both cores and CPU limit are used to adapt the schedule,
but will assume reasonable defaults if omitted.
Starting with version 1.3, E is able to provide answer substitutions. It currently adheres to the Tuple version of the TPTP proposal for answer extraction. To use this feature, you need to specify an existentially quantified conjecture formula with the formula type question, as in the example below. This implies the use of the TPTP-3/TSTP syntax.
fof(socrates,axiom,(philosopher(socrates))). fof(plato,axiom,(philosopher(plato))). fof(hume,axiom,(philosopher(hume))). fof(philosphers_exist,question,(?[X]:(philosopher(X)))).
By default, E will stop searching after the first answer. The
--answers
option can be used to specify for how many
answers E will keep searching :
eprover --answers=2 ans_test01.p
E comes with UNIX man pages for the major executables,
including eprove
r, eground
,
and e_axfilter
. If you perform a full install
e.g. into /usr/local, man should be able to find them. If not,
you can directly display the man pages via e.g.
man E/DOC/man/eprover.1
The programs all support the --help
option, which
will generate output that is roughly equivalent to the man
page.
E/DOC/eprover.pdf
.eprover --version.