search

rl-institut / smooth

Simulation framework for non-linear state-dependant energy systems
Other
6 stars 7 forks source link

no parallelisation during gradient descending #170

Closed LauWien closed 4 years ago

LauWien commented 4 years ago

To reduce the run time it would be nice if the gradient descending would run parallel on all cores - or the amount you give the script. So first one run without parallelisation to get the direction of the gradient. After that parallel stepwise on all given cores.

LauWien commented 4 years ago

Is one attribute at a time beeing checkt within the gradient descending or all decision variables at once? -- I have the feeling it already is.