Fixing of mixing weights in the normalmixEM() function of mixtools

[WimVDE001]

trafficstars

Hi,

Thanks a lot for this useful package!

I am using the mixtools package to fix a mixture of 4 normals with the normalmixEM() function (in the context of validating potential surrogate biomarkers; this methodology will be implemented in the Surrogate R package).

In the normalmixEM() function, the lambda parameters can be specified as starting values for the mixing weights in the EM algorithm. I wonder whether it is possible to ‘fix’ these mixing weights. So for example, is there a way to fix e.g., lambda=c(.5, .3, .2., .1) such that the 4 obtained normal components will have exactly these mixings weights?

Thanks for your consideration,

Best regards, Wim Van der Elst