n | dt
|
Strategy used to solve the algebraic equations: every Nth step or periodically for Adams–Bashforth methods.
|
||
n | dt
|
Strategy used to solve the algebraic equations: every Nth step or periodically for Runge–Kutta methods.
|
||
Global absolute tolerance as a factor of the relative tolerance, if atolglobalvaluemethod is factor.
|
|||
scaled | unscaled
|
|||
factor | manual
|
|||
Absolute tolerance for time derivatives per field. Only applicable if atoludotactive is on. See below.
|
|||
true | false
|
|||
on | off
|
|||
Name of the controlling study step or user if the feature is controlled manually.
|
|||
Interpolate the last time specified in the list of output times, if true. If set to false, the last output time is solved and not interpolated. In particular, the solver does not take steps past the last time.
|
|||
include | exclude
|
|||
Expression for time stepping when tstepping=elemexprs.
|
|||
on | off
|
|||
on | off
|
|||
on | off
|
|||
on | off
|
|||
on | off
|
|||
on | off
|
|||
on | off
|
|||
on | off
|
|||
The fraction of the time step for the initial step of a manual time stepping for BDF. The name and the default depend on the BDF order; for example, initialstepfractionbdf-2 for BDF order 2.
|
|||
The growth rate for the initial steps of a manual time stepping for BDF. The name and the default depend on the BDF order; for example, initialstepgrowthratebdf-2 for BDF order 2.
|
|||
on | off
|
|||
on | off
|
|||
yes | maybe
|
|||
Maximum time step for Dormand–Prince 5: automatic (auto), constant (const), or an expression (expr).
|
|||
Maximum time step for generalized alpha: automatic (auto), constant (const), or an expression (expr).
|
|||
1 | 2
|
|||
direct | lumped
|
|||
automatic | manual
|
|||
Number of time steps between updates, when updtlvl is set to manual when timemethodexp is not set to ab3loc.
|
|||
Number of time steps between updates, when updtlvl is set to manual when timemethodexp is set to ab3loc.
|
|||
implicit | explicit
|
|||
on | off
|
|||
tsteps | tout
|
|||
tsteps | tout
|
|||
Probed to use if probesel=manual.
|
|||
linear | constant
|
|||
on | off
|
|||
on | off
|
Rescale after initialization for the BDF and Generalized alpha methods, when consistent is set to bweuler.
|
||
Runge–Kutta method: RK34, Cash–Karp 5, or Dormand–Prince 5. Only available when timemethodexp is set to rk.
|
|||
on | off
|
|||
on | off
|
|||
on | off
|
|||
auto | manual
|
|||
auto | manual
|
|||
Times to store: output times by interpolation; every Nth step taken by solver; or steps taken by solver closest to output times.
|
|||
manual | elemexprs
|
|||
false | manual
|
Update time step: off (false) or manual, when tstepping is set to explicit for the timemethodexp methods erk and ab3.
|
•
|
With implicit (the default), an implicit time-stepping method such as BDF is used.
|
•
|
With explicit, an explicit time-stepping method, such as the Runge–Kutta family of explicit methods is used.
|
•
|
With timemethod=bdf, the IDA solver (which uses a variable order backward differentiation formula) is used.
|
•
|
With timemethod=genalpha, the generalized-α method is used. With generalized-α, the numerical damping can be controlled by giving a value, 0 ≤ ρ∞ ≤ 1, by which the amplitude of the highest possible frequency is multiplied each time step (hence, a small value corresponds to large damping while a value close to 1 corresponds to little damping). This is done through the property rhoinf. Also, the initial guess for the solution at the next time step (needed by the nonlinear solver) can be controlled through the property predictor when generalized-α is used. With predictor=linear, linear extrapolation using the current solution and time-derivative is used. With predictor=constant, the current solution is used as initial guess.
|
•
|
When timemethod is set to init the solver computes consistent initial values (for the start time, as defined by the property tlist) for the system and then stop. Time derivatives of algebraic variables and indicator functions might still be uninitialized after this operation. Such uninitialized quantities are represented by NaN (not a number) in the solution object.
|
•
|
With timemethodexp=rk, an explicit Runge–Kutta method is used: RK34, Cash–Karp 5, or Dormand–Prince 5 (specified using the rkmethod property).
|
•
|
With timemethodexp=erk, a classical Runge–Kutta method of order 1–4 is used.
|
•
|
With timemethodexp=ab3, an explicit Adams–Bashforth 3 method is used.
|
•
|
With timemethodexp=ab3loc, a local time-stepping version of Adams–Bashforth 3 is used (for the Wave Form PDE interface).
|
For more information about the Time-Dependent solver; see Time-Dependent Solver in the COMSOL Multiphysics Reference Manual.
|