Open blishko opened 1 year ago
Note: There are multiple benchmarks where this happens (besides chc-LRA-TS_111.smt2). OpenSMT spends time in strange places according to measurement. Check if the model is unnecessary complicated and could be simplified. If resolved, repeat witness validation for LAWI.
Regarding chc-comp-21/LIA-NonLin/chc-LIA-NonLin_522.smt2 and huge number of terms: This is caused by the complicated structure of the system, with large number of clauses (~1000) and large predicates (with hundreds of variables), which does not suit our normalization process very well. Some optimization has been applied to avoid creation of some temporary terms. More could be done, especially when dealing with equalities.