Bliksem Labs

Working on the next generation journey planners.

[Bliksem Labs @BliksemLabs]

Sometimes we write code that will never get merged upstream. It is not a wasted effort to write an elegant optimisation, or a poem.

[Bliksem Labs @BliksemLabs]

@xavierraffin @t_grote @navitia @LiveMap24 @opendatasoft @OpendataF @overflorian @OKFN @tblasig @Raildar_fr @KisioDigital We noticed that they don't use our pretty shapes, but recomputed their own.

[Xavier Raffin @xavierraffin]

Wonderful PT traffic world map by https://t.co/I51sv02J7r
PT OpenData Availability is not the same everywhere
#livemap24 #GTFS

[openOV @openOV]

Nederland zit nu ook in de @yandexcom app! :) https://t.co/IA0pkJd1TD https://t.co/QpgWorrI5N #Trots

[Andrew Byrd @globalvoid]

GTFS guidelines published at https://t.co/Z05SrhOtiP coordinated by @trilliumtransit for @RockyMtnInst.

[Waar is de Trein @waarisdetrein]

Het is alweer tijd voor de 20e versie van de Shifts app. Nieuwe dienstregeling = nieuwe data, incl. een nieuwe standplaats, Gouda. (1/3)

[Bliksem Labs @BliksemLabs]

@semilevel goed punt. In de dienstregelingen is dat ook niet altijd het geval. Maar ja, we moeten inderdaad wat met on-demand vervoer.

[Bliksem Labs @BliksemLabs]

Ik zou me als adviesbureaus zorgen gaan maken als Kim in drie zinnen uitleg, zelfstandig een mobiliteitanalyse maakt met NDOV data ;) https://t.co/G4a5G3xxW0

[Navitia @navitia]

Three years old yet OmG! "Keep calm and release the #Kraken!" as @pietercolpaert said. #Navitia #opensource #octopus https://t.co/9o5N9dChGG

[Bliksem Labs @BliksemLabs]

Lekker aan het hacken bij #NextMobility @BrabantKennis een webservice op basis van RoutingKit /cc @pietercolpaert @OSCityNL @jasperhartong

[Bliksem Labs @BliksemLabs]

@pietercolpaert @tristramg after picking some random number I knew should be in range. 100ms incl. overhead. 24-50ms excluding overhead.

[Bliksem Labs @BliksemLabs]

@pietercolpaert @tristramg sure, can you get #RoutingKit it to work? https://t.co/XueJjSBpAn

[Bliksem Labs @BliksemLabs]

@tristramg @pietercolpaert to be very clear. Our edge based A* performance is 250ms for 1:1, Dijkstra 1:N vertex based is 600ms.

[Bliksem Labs @BliksemLabs]

@pietercolpaert For A* a heuristic computation requires more math and more memory access. Still want to find the exact cause.

[Bliksem Labs @BliksemLabs]

@pietercolpaert Dijkstra would be just a priority queue, visited set and results storage at the exact size of the vertices number.

[Bliksem Labs @BliksemLabs]

@pietercolpaert because of suboptimal datastructures with respect to cpu, floating-point, cacheline and memory usage.

[Bliksem Labs @BliksemLabs]

After testing A* in C, Dijkstra in C was next on our list. With mind blowing performance.

[Bliksem Labs @BliksemLabs]

Another exciting visualisation of critical network paths.

[Open Data Nederland @Opendatanl]

Bedrijven, overheidsorganisaties én privé-personen kunnen de Stuiveling Open Data Award winnen, doe mee! https://t.co/YlqyLNqeJk #SODA2016

[Bliksem Labs @BliksemLabs]

Your laptop is so slow, it creates real-time journey planning animations in @qgis https://t.co/NdtCbHCRNU