Constraint-based Causal Discovery by using Path Constraints gained from Signal Injection and Recovery

The discovery of causal relations via interventions has proven to be simple when only one observed variable is affected or unaffected. However, in a multivariate setting, it is likely that more than one variable is affected by the intervention. Thus drawing conclusions about the true causal graph becomes far more difficult as we can not retrieve information of any obvious causal relationship or causal order. We demonstrate, that causal discovery with multiple affected variables is possible by introducing a novel definition of path constraints for constraint-based causal discovery. We exercise our novel technique on a combustion engine simulation, were we inject wavelets of our choice in a variable of investigation and try to rediscover this wavelet in the other, observed variables to gain such path constraints and thus to restraint the causal graph search.