I have a project where we need to do many forecasts in a parallel setting. In predict.varest there are multiple calls to lm.summary. This has multiple side-effects, first it is a very expensive way to retrieve the degrees of freedom, as that would only require counting the number of coefficients and subtracting that number from the observation count. The second problem which I guess is more specific to my application is that I sometimes get a problem when running multiple threads in mclapply, where the execution stops inside the lm.summary function.
I have forked the repository and created a pull request for you to evaluate. Please let me know if you need any more information or any adjustments to the code. Please see https://github.com/bpfaff/vars/pull/12
I have a project where we need to do many forecasts in a parallel setting. In
predict.varest
there are multiple calls tolm.summary
. This has multiple side-effects, first it is a very expensive way to retrieve the degrees of freedom, as that would only require counting the number of coefficients and subtracting that number from the observation count. The second problem which I guess is more specific to my application is that I sometimes get a problem when running multiple threads inmclapply
, where the execution stops inside thelm.summary
function.I have forked the repository and created a pull request for you to evaluate. Please let me know if you need any more information or any adjustments to the code. Please see https://github.com/bpfaff/vars/pull/12
Best regards Jon.