Question about isConverge implmentation

[hdi-amiri]

I was wondering why isConverged function is not based on RMSE of training set, instead it is based on value of objective function. with this isConverged function, I guess it is almost impossible for algorithm to converge.

for example this implementation is using rmse of training set as stopping condition.

I wanted to know does librec have this capability of such convergence functions? What is the reason behind that you guys took another approach for isConverged function?