Flux sampling

[Sara]

Hello, 

I am running Flux sampling with a very big model (~3000 reactions). FBA gives feasible solutions using glpk as solver, but not using cplex. 

When I try to run Flux sampling, it takes very long and I get this error:

Cannot escape sampling region, model seems to be numerically unstable. 

Or with cplex:

Also, range cannot be empty (low>=high) unless no samples are taken. 

I have tried to remove all the infeasible loops, but still it seems to be unstable. What do you recommend? Shall I make the model smaller or just assume that it is not feasible?

Thanks a lot,

Sara 

[ch.d...@gmail.com]

Hmm it looks like your model is very close to infeasible or actually infeasible. Do you have any constraints that do not include zero or demand reaction in there that may not be getting fulfilled? Does the solution that GLPK give you make sense?