Run module

Once the model is defined, one can run the reactor code to simulate the photochemistry over the defined reaction time. Two controls are available:

• # MC runs defines the number of Monte-Carlo samples of the chemistry for which the code is to be run.

  • if set to 0, only the nominal run is performed.

  • for positive numbers, an additional control appears, asking if one wishes to append those runs to the exisiting ones. In this case, the sample numbers will be chosen accordingly, to avoid double runs with the same sample…

• # snapshots per run defines the number of snapshots of the species concentrations to be saved along the simulation. The points are regularly spaced on the logarithmic time scale.

• Log relative error and Log absolute error control the error levels of the IRKC implicit-explicit integrator (see https://github.com/ppernot/Reactor)

  Note: the default values are chosen to provide rapidly reasonable results. Fot a better accuracy on the minor species, it might be better to lower these values.

• Use SR and Max. SR: by default (Use SR unchecked), the reactor code estimates the spectral radius (SR) of the jacobian of the explicit contribution (see https://github.com/ppernot/Reactor). The estimation can fail, with a “Convergence failure in estimating the spectral radius.” message. To circumvent this, one might shortcut the automatized estimation of SR by checking Use SR and providing a value in Max. SR. If an appropriate integration statistics file exists, and if the reported value of SPRAD is not constant, SRmax is estimated as 1.2*max(SPRAD). This choice can be overridden by providing a value in Max. SR. If available, evolution of SPRAD can be seen in Analysis>Sanity checks>Spectral radius.

Local tab

• Log tail only enables to see the last lines of the reactor code log when ckecked or the full log when unchecked. It is checked by default to enable to visualize the progress of the task.

The Start button in the Local tab will launch the simulations on the local computer. The text outputs and error messages should be captured and displayed in the right panels.

Warning: it might happen that no output is displayed, although the simulations are perfectly running in the background. If this happens, check that the MC_Output folder is being populated according to your expectations… Never press the Start button twice, as this might start several concurrent runs of reactor and mix up their outputs.

Cloud tab

There is provision for a Cloud mode to launch the simulations on the Virtual Data cloud. To be done… For now, one needs to do it externally from the ReactorUI interface.

Note: to run reactor on the Virtual Data openstack cloud:

1. perform a nominal run on the local machine to ensure and check that the project is correctly configured;

2. follow the manual procedure explained on CloudReactor in this document;

3. when the runs are done, proceed to the Analysis tab.