Dietmar Wolz

Results 20 comments of Dietmar Wolz

Indeed, in SMAC3 "Scenario" supports the "n_jobs" parameter where you also can execute multiple runs in parallel. It may indeed be a good solution. Will also have a look at...

Related to the parallelization topic: I am currently thinking how to add parallelization to ALNS itself. For instance I could implement a "parallel_repair" operator which applies several "repairs" in parallel...

> obvious parallellisation of doing independent ALNS runs in parallel With evolutionary optimization - which I am more familiar with - we are faced with the same problem. Sometimes the...

Was busy with other tasks, will add more examples, but probably not before October. There are two JSON files already uploaded: https://github.com/dietmarwo/fast-cma-es/blob/master/examples/data/sched2.json https://github.com/dietmarwo/fast-cma-es/blob/master/examples/data/sched1.json which one exactly is missing?

Hallo, attached are my slided from the ESA optimization workshop were I talked about  CR-FM-NES. CR-FM-NES is one of the very few algorithms where I don't have a good alternative...

Haven't seen this problem on python 3.9 (https://repo.anaconda.com/archive/Anaconda3-2021.11-Linux-x86_64.sh). Seems not related to the benchmark itself.

Sorry, I missed the epochOffset parameter. Backpropagating 50 years (ESAs epoch 0 is at 01.01.2000) gives consistent mean anomalies: Mecury 5.553920558995985 / 5.559496921012562 Venus 5.432267392125204 / 5.436181913693716 > it added...

> Can you be more precise about what information should add ? The most detailed approach to really see what is going on is to write for each time anything...

>My differential evolution algorithm is multithreaded using webworkers to distribute the trajectory evaluation of each >individual in the population (it increases the speed a bit, but not by much). I...

By the way: https://optimize.esa.int/challenge/spoc-trappist-tour/p/trappist-tour contains verification / visualization code for an 1DSM tour using planets defined by their keplerian parameters: https://api.optimize.esa.int/media/problems/spoc-trappist-tour-trappist-tour-1648650812274.py It probably can easily be adapted to planets used...