Equilibrium parameters

If it is applicable, contains the type of assignment algorithm.

• 0 = Trajectory DTA with MSA.
• 1 = Temporal layer DTA with gradient projection.
• 2 = Temporal layer DTA with stochastic route choice.

If AssignmentAlgorithm = 2, the listed parameters are ignored:

• → ConvergenceAlgorithm
• → GradientFunction
• → MinRelativeGapDTA
• → SplitModelCoef

Default is 1.

Caches the destination flows and destination demand matrix on disk.

Dumping of cached data is used to improve memory handling.

Important:  A Solid State Disk (SSD), significantly improves computation times, reducing I/O operations time.

• true: Data cached.
• false: Otherwise.

Default is true.

Convergence algorithm.

If it is applicable, controls the convergence algorithm.

Admitted values:

• 0 = MSA (Method of Successive Averages).
• 1 = Exact Gradient Projection.
• 2 = Quasi Gradient Projection.
• 3 = Reduced Gradient Projection.
• 4 = Min-Max Gradient Projection.

Default is 3.

Represents the interval length (in seconds) for the demand event binary data.

The smaller the interval, the larger the file size.

DemandEventBinaryDataPeriodization< ResIntD are not useful (→ ResIntD).

DemandEventBinaryDataPeriodization=-1 means use the same value of ResIntD.

Default is -1.

Maximum number of DNL iterations (DNL resolution problem, see → Running day type simulations using DNL mode).

After DNLnITE, the equilibrium algorithm is stopped, disregarding the relative gap value (see → TRE convergence formulas.

Default is 0.

Maximum number of DTA iterations (DTA simulation).

After DTAnITE, the equilibrium algorithm is stopped, disregarding the relative gap value (see → TRE convergence formulas).

Default is 30.

Defines the number of parallel DTA threads.

Improves computation performance, reducing route choice computation time.

Special values:

• −1 = Use all logical cores (see Windows Task Manager > Performance tab).
• 0 = Single thread.

Default is -1.

Initial instant of the DTA simulation.

The format is expressed as a pattern [HHmmss].

Special values:

• −1 = Equal to the initial instant of the simulation.

Default is -1.

Duration of DTA intervals, measuered in seconds [s].

Driver route choice is assumed to be constant within each DTA interval, then better convergence is achieved using a value consistent with supply variability.

Default is 3600.

Span of DTA, measured in seconds [s].

Represents the time extension of the DTA simulation.

Special values:

• −1 = Equal to the span of the simulation.

Default is -1.

Equilibrium assignment.

Controls the computation of the dynamic equilibrium assignment in TRE.

Admitted values:

• 0 = Dynamic network loading with fixed route choice. Does not perform an equilibrium but only GLTM dynamic network loading based on the route choices described by the turn probabilities in the TPRB table.

• 2 = Dynamic user equilibrium. Operates as a standard equilibrium.

Default is 2.

Gradient scaling function.

If a gradient projection algorithm is used for convergence, GradientFunction determines the function used as cost derivative.

Admitted values:

• 0 = wmin

• 1 = wmin0
• 2 = avgw
• 3 = wa
• 4 = wa0

Default is 2.

Contains the binary snapshot file path (name) to be loaded.

If empty, the snapshot is saved with the name Visum_filename-Snapshot.bin, where Visum_filename is the name of the Visum file involved in the process.

For example: NetworkName-Snapshot.bin.

Default is "" (empty string).

Preserves bush (for example, the topological order) between equilibrium iterations.

Important:  If AssignmentAlgorithm = 2, KeepBush MUST BE SET at true, otherwise TRE provides a diagnostic about it.

Default is false.

Linear system solution algorithm.

If it is applicable, determines the solution algorithm of the linear loading system.

The linear loading system determines the destination network loading map, given the destination turn probabilities.

Admitted values:

• 0 = BiCGSTAB (BiConjugate Gradient Stabilized method)
• 1 = Jacobi
• 2 = Gauss-Seidel

Default is 2.

Minimum gap linear loading problem.

Linear loading problem error threshold.

Defines a stop criterion of the network loading linear system solution algorithm based on the achievement of a minimum error.

Default is 1e-10.

Enables preconditioning of the linear loading algorithm to improve the network loading map solution.

• false: On networks not significantly congested, preconditioning can be skipped to avoid the additional computation cost. The price of this choice consist of a smaller coherence of TPRB according with the route choices computed by the RCM.
• true: Otherwise.

Default is true.

Maximum number of iterations of the linear loading problem.

This sets a stop criterion of the network loading linear system solution algorithm based on the achievement of a maximum number of iterations.

Default is 10.

Associated to DTA input assignment information.

• true: Enables as input (starting point) the assignment information exported by a previous DTA.
• false: Otherwise.

Default is false.

Relative gap threshold of the DNL problem (see (→ Dynamic Traffic Assignment (DTA) > DNL iterations)).

Sets a stopping criterion based on the achievement of a minimum relative gap.

Default is 1e-12.

Relative gap threshold of the DTA problem (see → Dynamic Traffic Assignment (DTA)).

Sets a stopping criterion based on the achievement of a minimum relative gap.

Default is 1e-5.

Contains the binary snapshot file path (name) to be saved.

If empty, the snapshot is saved with the name Visum_filename-Snapshot.bin, where Visum_filename is the name of the Visum file involved in the process.

For example: NetworkName-Snapshot.bin.

Default is "" (empty string).

Number of internal iterations per root.

Default is 1.

Associated to DTA exported assignment information.

• true: Export assignment information to be used as input (starting point) for another DTA (see → Dynamic Traffic Assignment (DTA)).
• false: Otherwise.

Default is false.

Associated to the binary data of the demand events.

• true: Export the binary data for the demand events at the end of the simulation.
• false: Don't export the binary data.

Default is true.

Defines how are handled rerouting data.

• true: Destination-based split rates and other data are stored at the end of an equilibrium run, to enable simulation of rerouting events (DoRerouting = true) on subsequent network loading runs for the same day type.
• false: Otherwise.

Default is true.

Related options:

• EquAss=2 to enable SaveReroutingData=true.
• DoRerouting=true only works after rerouting data has been produced by an equilibrium with (SaveReroutingData=true) for the relevant day type.

Important:  If SaveReroutingData is true, there are constraints on other TRE configuration parameters (see → Rerouting event model > Off-line equilibrium to get binary files to enable rerouting).

Represent the number of incremental assignment iterations before starting with standard iterations.

WarmIte iterations are used to mitigate the effects of unnatural congestion generated by the first AoN (All or Nothing assignment) iterations of MSA (Method of Successive Averages) assignment.

AoN identifies the minimum cost assignment, when everyone takes the shortest path (in time) from origin to destination.

During warm-up iterations, in the supply model a constantly increasing share of the input demand is loaded until reaching 100% in the last iteration.

Default is 0.