When you calculate the net flux you have to calculate the forward flux for each edge, and throw out the edges that have negative flux. When you do this (https://github.com/brycecr/msmexplorer/blob/master/MSMExplorer/src/edu/stanford/folding/msmexplorer/tpt/TPTFactoryCM.java#L302-L310) you have fFlux that contains the total flux along each edge, but you are modifying it on the fly to compute the net flux.
The problem with doing this, is when you modify fFlux(i, j), now the net flux on edge (j, i) will be calculated incorrectly.
When you calculate the net flux you have to calculate the forward flux for each edge, and throw out the edges that have negative flux. When you do this (https://github.com/brycecr/msmexplorer/blob/master/MSMExplorer/src/edu/stanford/folding/msmexplorer/tpt/TPTFactoryCM.java#L302-L310) you have fFlux that contains the total flux along each edge, but you are modifying it on the fly to compute the net flux.
The problem with doing this, is when you modify fFlux(i, j), now the net flux on edge (j, i) will be calculated incorrectly.