v2.6.3

Warning! JSON and SQLite outputs have been updated. Changes are strictly additive.

Changes:

• Public transit legs are now available in output. Among other things, this allows for the analysis of the involved foot distance and in-vehicle time.
• The home location is now directly given next to the person's features and does not have to be inferred from the activities anymore.
• Route choice calibration has been fixed and is available again.
• Reduced the progressbar tick rate when no console is registered to reduce spam when stdout is piped to a file.
• Performance: Increased the degree of parallelization of several initialization routines.
• OSM file for road network can now be supplied separately. Useful when traffic counts must be at a specific point in time, but POI data should be the latest.

Fixes:

• Mode choice calibration is now weekday-aware.
• Assignment now correctly includes mode_speed_up

v2.4.8

Changes:

• Significant speed-up of surrogate-assisted calibrations; achieved by implementing the surrogate in TensorFlow (x300 times faster than previously)
• New debug option for calibration that helps find errors with calibration data.
• Simplified custom parameterization. Individual files can now be supplied through the CLI. Parameter sets can now inherit from one another.
• Smaller plattform-specific binaries are now available. A platform-independent "All"-binary is still available.

Minor:

• Small logic update for finding a destination in the same cell that the agent is currently in.
• Bug fixes

v2.3.0

Changes:

• Moved to Java 21
• New experimental calibration feature
  • Calibrateable parameters:
    • Attraction of each location
    • Mode choice utility coefficients
    • Probability of route alternatives
  • Required data: Traffic counts
  • See also Readme
• Mode choice option: FAST
  • Same as GTFS but without including travel time.
  • Less realistic mode choice but runtimes decrease from hours to seconds.
• Bugfixes

v2.2.2

Support for ordinary osm.xml files.

Fixed:

• Error when a hole of a osm relation split the relation into multiple polygons.
• Error where clipping the GTFS file created invalid trips.

v2.2.1

Name change

v2.2.0

New map data option: Overture Maps

v2.1.1

• Run parameters are now part of the output to promote reproducibility.
• New CLI Option:
  • Usage: --mode_speed_up mode=factor
  • Use: Slow down or speed up one or more modes to determine the impact of travel time on mode choice.
  • Example: --mode_speed_up CAR_DRIVER=0.3 (will slow down car travel durations by 70%)

v2.0.19

New features:

• Mode choice: It is now possible to let OMOD determine each agent's transport mode between activities with a multinomial logit model. To do so, set the flag --mode_choice to GTFS. The resulting trips will also include the distance and the free-flow duration of the trip with the determined mode. The GTFS option requires a valid GTFS file that covers the area of interest to determine travel times with public transit (for Germany see: https://gtfs.de). For convenience, you can also set the --mode_choice flag to CAR_ONLY to force all trips to be done by car or to NONE to create trips with an undefined mode.
• Parallelization: Most computationally heavy parts of OMOD run in parallel now. Set the number of parallel processes with --n_workers. By default, this value is set to the number of CPU-Cores available.
• Refactoring: The source code has been restructured to introduce more modularity. We introduced interfaces that define the requirements for each major model component, simplifying the process of replacing these components (see doc/codeStructure.md).
• Output formats: The JSON output format has been overhauled to allow for the optional inclusion of trips between activities. Additionally, two new formats are available: SQLite and MATSim (Unstable). Select the desired format by changing the output file's extension to .json, .db, or .xml.
• Population Definition: The frequency of features in the population can now be defined more flexibly through the definition of subgroups. For example, you could define different generations and assign each generation different car ownership probabilities.

Warning!! The changes required a change in the output format.

v1.8.2

• Bugfixes

v1.8.1

Warning! Delete old caches when using this version.

Changes:

• Added OSM Features for destination choice attraction (Unused in the default calibration):
  • Areas for: Office, retail space, schools, universities, shops
  • Numbers for: Place-of-worship, cafés, fast food places, restaurants, kinder gardens, tourism POIs, retail units
• Bugfixes: OD-Matrix calibration
• Performance improvements