MATSim Code

test performance of subnetwork routing

suprisingly, there does not seem to be a big overhead of checking mode
restrictions for every link (at least in a network were all links can
actually be used). Thus not sure that building a sub-network is worth it
(what triggered my look into this seems to be one of those profiler
artefacts...)

log

also perform dilution at import

with the same input files, we can now easily try out different sample
sizes and areas of interest, without having to pass a day re-generating
the input.

Still TODO: some analysis listenners which use the knowledge about
subpopulation and AOI to produce meaningful statistics

sample subpopulation files directly at import

same idea as the other things in ScenarioMergingConfigGroup: avoid the
multiplication of input files.

just use info from activities to locate facilities

this enforces consistency and removes the need for a f2l file.

ignore pt interactions when using activities to locate facilities

remove incomplete constructor

there was only one usecase, in my playground --- probably a leftover
from a refactoring

make use of delegation in TripRouterFactoryImpl

it was duplicating ALL the code from DefaultTripRouterFactoryImpl.

Michael, I had to slightly alter your code (the new version is not
strictly equivalent). Please check if it is not breaking anything (I
don't think it should, but one never knows).

remove unsused constructor

improve facility to link