TSOpt in QMERA,model is unreasonable

Dear All,

Before I optimized a TS model,the atomic position has been fixed.But after optimization,the fixed part of the atoms has moved. What is the reason?

Regards
Yi Wang