PSE issues using stochastic heun integrator

[Stefan Lang]

Dear TVB community. 

I am attempting a whole-brain simulation using the default TVB connectome in the GUI of TVB version 2.7.2.

I am using the Stefanescu-Jirsa 3D model, with parameters set according to doi: 10.1523/ENEURO.0158-15.2016. I am using the mean values from healthy controls for K12, K11, and K21. 

I am using the stochastic huen integrator with a dispersion of 0.1. The integration step size is 0.012. I am monitoring the temporal average of Xi state variable. The sampling period is the default 0.9765625. Simulation length is 1000ms. 

I am attempting to perform a parameter space exploration, varying the coupling value. When I perform single simulations with various global coupling values between 0 and 0.01 I get reasonable output of time courses with variation dependent on the coupling value. 

However, when I perform a parameter sweep using the GUI PSE function (0.001 - 0.1, step size 0.01) there is no difference across coupling values.  When I click on the discrete PSE visualizer, it states the minimum and maximum global variance are the same. When I export the time-series and plot them, the time courses are exactly the same for every value of the global coupling. When I examine this carefully it seems that only the coupling value that was defined at the top of the simulation set up is being used, rather than performing a sweep over the range defined in the PSE section at the bottom of the GUI.

However, if I perform the exact same simulation with the PSE using the non-stochastic Huen integrator, I am able to see some variability across coupling values. 

It seems to me that the GUI PSE function may not be working when using the stochastic integrator? 

Has anyone else noticed this or have any suggestions? 

Any assistance is appreciated.