7.38. Restart

This section is referenced in the following other sections

• Examples: Restart Database
• Simulation Setup: Restart Reference

7.38.1. Restart Data

Scope

Aria Region, Input_Output Region

Summary

Describes the data required to output and input restart data for the enclosing region.

begin Restart Data Label

   Additional Steps {=} List_of_steps...

   Additional Times {=} List_of_times...

   At Step n {increment | interval} {=} m

   At Time Dt1 {increment | interval} {=} Dt2

   At Wall Time Dt1 {increment | interval} {=} Dt2

   Component Separator Character {=} Separator

   Cycle Count {=} Count

   Database Name {=} StreamName

   Database Type {=} {catalyst | catalyst_exodus | cgns | dof | dof_exodus | exodus | exodusii | exonull | generated | genesis | null | parallel_exodus | textmesh}

   Debug Dump

   Decomposition Method {=} {block | cyclic | external | geom_kway | hsfc | kway | kway_geom | linear | map | metis_sfc | random | rcb | rib | variable}

   Exists {=} {abort | add_suffix | append | overwrite}

   File Cycle Count {=} Count

   Input Database Name {=} StreamName

   Optional

   Output Database Name {=} StreamName

   Output On Signal {=} {sigabrt | sigalrm | sigfpe | sighup | sigill | sigint | sigkill | sigpipe | sigquit | sigsegv | sigterm | sigusr1 | sigusr2}

   Output Restart State {=} {off | on}

   Overlay Count {=} Count

   Overwrite {=} {false | no | off | on | true | yes}

   Property PropertyName {=} PropertyValue

   Restart {=} {auto}

   Restart Time {=} Time

   Start Time {=} Start_time

   Synchronize Output

   Shift To Start Time

   Termination Time {=} Final_time

   Timestep Adjustment Interval {=} Nsteps

   Use Dynamic Topology Io

   Use External Control

   Use Output Scheduler Timer_name

end Restart Data Label

7.38.1.1. Line Commands

Additional Steps

Syntax

Additional Steps {=} List_of_steps…

Summary

Additional simulation steps when output should occur.

Parameter	Value	Default
{=}	{= | are | is}	–
List_of_steps	integer…	–

Additional Times

Syntax

Additional Times {=} List_of_times…

Summary

Additional simulation times when output should occur.

Parameter	Value	Default
{=}	{= | are | is}	–
List_of_times	real…	–

At Step

Syntax

At Step n {increment | interval} {=} m

Summary

Specify an output interval in terms of the internal iteration step count. The first step specifies the step count at the beginning of this interval and the second step specifies the output frequency to be used within this interval.

Parameter	Value	Default
n	integer	–
Option	{increment | interval}	–
{=}	{= | are | is}	–
m	integer	–

At Time

Syntax

At Time Dt1 {increment | interval} {=} Dt2

Summary

Specify an output interval in terms of the internal simulation time. The first time specifies the time at the beginning of this time interval and the second time specifies the output frequency to be used within this interval.

Parameter	Value	Default
Dt1	real	–
Option	{increment | interval}	–
{=}	{= | are | is}	–
Dt2	real	–

At Wall Time

Syntax

At Wall Time Dt1 {increment | interval} {=} Dt2

Summary

Write a restart file at a specific wall time since the start of the run. Time string format allows s, m, h, d for seconds, minutes, hours, days

Parameter	Value	Default
Dt1	string	–
Option	{increment | interval}	–
{=}	{= | are | is}	–
Dt2	string	–

Component Separator Character

Syntax

Component Separator Character {=} Separator

Summary

The separator is the single character used to separate the output variable basename (e.g. “stress”) from the suffices (e.g. “xx”, “yy”) when displaying the names of the individual variable components. For example, the default separator is “_”, which results in names similar to “stress_xx”, “stress_yy”, … “stress_zx”. To eliminate the separator, specify an empty string (“”) or NONE.

Parameter	Value	Default
{=}	{= | is}	–
Separator	string	–

Cycle Count

Syntax

Cycle Count {=} Count

Summary

Specify the number of restart steps which will be written to the restart database before previously written steps are overwritten. For example, if the cycle count is 5 and restart is written every 0.1 seconds, the restart system will write 0.1, 0.2, 0.3, 0.4, 0.5 to the database. It will then overwrite the first step with data from time 0.6, the second with time 0.7. At time 0.8, the database would contain data at times 0.6, 0.7, 0.8, 0.4, 0.5. Note that time will not necessarily be monotonically increasing on a database that specifies the cycle count.

Parameter	Value	Default
{=}	{= | are | is}	–
Count	integer	–

Database Name

Syntax

Database Name {=} StreamName

Summary

The database containing the input and/or output restart data. If this analysis is being restarted, restart data will be read from this file. If the analysis is writing restart data, the data will be written to this file. It will be overwritten if it exists (after being read if applicable). If the filename begins with the ‘/’ character, it is an absolute path; otherwise, the path to the current directory will be prepended to the name. See also the ‘Input Database’ and ‘Output Database’ commands.

Parameter	Value	Default
{=}	{= | are | is}	–
StreamName	string	–

Database Type

Syntax

Database Type {=} {catalyst | catalyst_exodus | cgns | dof | dof_exodus | exodus | exodusii | exonull | generated | genesis | null | parallel_exodus | textmesh}

Summary

The database type/format used for the restart file.

Parameter	Value	Default
{=}	{= | are | is}	–
DatabaseTypes	{catalyst | catalyst_exodus | cgns | dof | dof_exodus | exodus | exodusii | exonull | generated | genesis | null | parallel_exodus | textmesh}	–

Debug Dump

Syntax

Debug Dump

Summary

Specify whether the restart system will write the restart data immediately after reading the restart data if the run is restarting. The output data can be compared with the restart input data to determine whether they match.

Decomposition Method

Syntax

Decomposition Method {=} {block | cyclic | external | geom_kway | hsfc | kway | kway_geom | linear | map | metis_sfc | random | rcb | rib | variable}

Summary

The decomposition algorithm to be used to partition elements to each processor in a parallel run.

Parameter	Value	Default
{=}	{= | are | is}	–
Method	{block | cyclic | external | geom_kway | hsfc | kway | kway_geom | linear | map | metis_sfc | random | rcb | rib | variable}	–

Exists

Syntax

Exists {=} {abort | add_suffix | append | overwrite}

Summary

Specify the behavior when creating this database and there is an existing file with the same name. The default behavior is “OVERWRITE” which deletes the existing file and creates a new file of the same name. “APPEND” will (if possible) append the new data to the end of the existing file. “ABORT” will print an error message and end the analysis. “ADD_SUFFIX” will add a suffix to the file name and output to that file.

Parameter	Value	Default
{=}	{= | is}	–
Option2	{abort | add_suffix | append | overwrite}	–

File Cycle Count

Syntax

File Cycle Count {=} Count

Summary

Each restart dump will be written to a separate file suffixed with A,B, … The count specifies how many separate files are used before the cycle repeats. For example, if “FILE CYCLE COUNT = 3” is specified, the restart dumps would be written to file-A.rs, file-B.rs, file-C.rs, file-A.rs, … The maximum value for the cycle count is 26.

Parameter	Value	Default
{=}	{= | are | is}	–
Count	integer	–

Input Database Name

Syntax

Input Database Name {=} StreamName

Summary

The database containing the input restart data. If this analysis is being restarted, restart data will be read from this file. See also the ‘Database’ and ‘Output Database’ commands.

Parameter	Value	Default
{=}	{= | are | is}	–
StreamName	string	–

Optional

Syntax

Optional

Summary

The database will be read if it exists, but it is not an error if there is no restart database to read for this region during a restarted analysis.

Output Database Name

Syntax

Output Database Name {=} StreamName

Summary

The database containing the output restart data. If the analysis is writing restart data, the data will be written to this file. It will be overwritten if it exists. See also the ‘Database’ and ‘Input Database’ commands.

Parameter	Value	Default
{=}	{= | are | is}	–
StreamName	string	–

Output On Signal

Syntax

Output On Signal {=} {sigabrt | sigalrm | sigfpe | sighup | sigill | sigint | sigkill | sigpipe | sigquit | sigsegv | sigterm | sigusr1 | sigusr2}

Summary

When the specified signal is raised, the output stream associated with this block will be output.

Parameter	Value	Default
{=}	{= | are | is}	–
Signals	{sigabrt | sigalrm | sigfpe | sighup | sigill | sigint | sigkill | sigpipe | sigquit | sigsegv | sigterm | sigusr1 | sigusr2}	–

Output Restart State

Syntax

Output Restart State {=} {off | on}

Summary

Outputs the restarted state to the new restarted results file

Description

NOTE: This command must be placed at the Sierra scope of the input file. Allows the analyst to visualize the restarted state for debugging

Parameter	Value	Default
{=}	{= | are | is}	–
Option	{off | on}	–

Overlay Count

Syntax

Overlay Count {=} Count

Summary

Specify the number of restart outputs which will be overlaid on top of the last written step. For example, if restarts are being output every 0.1 seconds and the overlay count is specified as 2, then restart will write times 0.1 to step 1 of the database. It will then write 0.2 and 0.3 also to step 1. It will then increment the database step and write 0.4 to step 2; overlay 0.5 and 0.6 on step 2… At the end of the analysis, assuming it runs to completion, the database would have times 0.3, 0.6, 0.9, … However, if there were a problem during the analysis, the last step on the database would contain an intermediate step.

Parameter	Value	Default
{=}	{= | are | is}	–
Count	integer	–

Overwrite

Syntax

Overwrite {=} {false | no | off | on | true | yes}

Summary

(DEPRECATED, Use EXISTS) Specify whether the restart database should be overwritten if it exists. The default behavior is to overwrite unless this command is specified in the restart block and either off, false, or no is specified.

Parameter	Value	Default
{=}	{= | is}	–
Option2	{false | no | off | on | true | yes}	–

Property

Syntax

Property PropertyName {=} PropertyValue

Summary

Define a database property named “PropertyName” with the value “PropertyValue”. If PropertyValue consists of all digits, it will define an integer property. If PropertyValue is “true” or “yes” or “false” or “no”, it will define a logical property; otherwise it will define a string property. If PropertyName consists of multiple strings, they will be concatenated together with “_” separating the individual words. Supported properties are typically database dependent; Current properties are:

• COMPRESSION_LEVEL = [0..9]

• COMPRESSION_SHUFFLE = true|false|on|off

• FILE_TYPE = netcdf4 (forces use of netcdf-4 hdf5-based file)

• INTEGER_SIZE_DB = 4|8

• INTEGER_SIZE_API = 4|8

• LOGGING = true|false|on|off

• MAX_NAME_LENGTH = value

Parameter	Value	Default
PropertyName	string	–
{=}	{= | are | is}	–
PropertyValue	string	–

Restart

Syntax

Restart {=} {auto}

Summary

Specify automatic restart file read.

Description

NOTE: This command must be placed at the Sierra scope of the input file.

Specify that the analysis should be restarted from the last common time on all restart databases for each Region in the analysis. In addition to this line command, each Region in the analysis (strictly, only the region(s) that will be restarted) must have a restart block specifying the database to read the restart state data.

By default, use of this command will not cause output files (e.g., results, history, heartbeat, restart) to be overwritten. Instead output files will be written with the same basename and the suffix -s000*. Common visualization packages are written to handle this file organization gracefully in order for the user to view all results seamlessly.

Parameter	Value	Default
{=}	{= | are | is}	–
{auto}	{auto | automatic}	–

Restart Time

Syntax

Restart Time {=} Time

Summary

Specify restart file read at a specified time.

Description

NOTE: This command must be placed at the Sierra scope of the input file.

Specify the time that the analysis will be restarted. In addition to this line command, each Region in the analysis (strictly, only the region(s) that will be restarted) must have a restart block specifying the database to read the restart state data. The restart ‘time’ must be greater than zero and less than or equal to the termination time.

By default, use of this command will cause previous output files (e.g., results, history, heartbeat, restart) to be overwritten. If this command is chosen, the onus is placed on the user to ensure that previous output files are not overwritten.

Parameter	Value	Default
{=}	{= | are | is}	–
Time	real	–

Start Time

Syntax

Start Time {=} Start_time

Summary

Specify the time to start outputting results from this output request block. This time overrides all ‘at time’ and ‘at step’ specifications.

Parameter	Value	Default
{=}	{= | are | is}	–
Start_time	real	–

Synchronize Output

Syntax

Synchronize Output

Summary

In an analysis with multiple regions, it is sometimes desirable to synchronize the output of results data between the regions. This can be done by adding the SYNCHRONIZE OUTPUT command line to the results output block. If a results block has this set, then it will write output whenever a previous region writes output. The ordering of regions is based on the order in the input file, algorithmic considerations, or by solution control specifications.

Although the USE OUTPUT SCHEDULER command line can also synchronize output between regions, the SYNCHRONIZE OUTPUT command line will synchronize the output with regions where the output frequency is not under the direct control of the Sierra IO system. Examples of this are typically coupled applications where one or more of the codes are not Sierra-based applications such as Alegra and CTH. A results block with SYNCHRONIZE OUTPUT specified will also synchronize its output with the output of the external code.

The SYNCHRONIZE OUTPUT command can be used with other output scheduling commands such as time-based or step-based output specifications.

Shift To Start Time

Syntax

Shift To Start Time

Summary

The shift to start time option allows a user to shift the restart time to the start time of the current region. An example use case would be if a restart time of 0.5 is specified, but the user would like to start the simulation at time 1.0.

Termination Time

Syntax

Termination Time {=} Final_time

Summary

Specify the time to stop outputting results from this output request block.

Parameter	Value	Default
{=}	{= | are | is}	–
Final_time	real	–

Timestep Adjustment Interval

Syntax

Timestep Adjustment Interval {=} Nsteps

Summary

Specify the number of steps to ‘look ahead’ and adjust the timestep to ensure that the specified output times or simulation end time will be hit ‘exactly’.

Parameter	Value	Default
{=}	{= | are | is}	–
Nsteps	integer	–

Use Dynamic Topology Io

Syntax

Use Dynamic Topology Io

Summary

Specify that the app use IO for dynamic topology modifications where the output files are stored in a single database. Legacy file format for dynamically changing topology results in the creation of multiple files for each output on a mesh modification. This option leverages the ability of netCDF to create mesh groups within a single database and concatenate all mesh files into one. The names of each mesh group are of the form IOSS_MESH_GROUP-??? where ??? is the 1-based output index 1, 2, …, 10, …., 100, … Please note that netCDF has a current limit of 65,536 groups

Use External Control

Syntax

Use External Control

Summary

Do not schedule any results output but instead depend on an externally coupled application to explicitly call for results output.

Use Output Scheduler

Syntax

Use Output Scheduler Timer_name

Summary

Associates a predefined output scheduler with this output block (results, restart, heartbeat, or history).

Parameter	Value	Default
Timer_name	string	–

7.38.2. Restart Time

Syntax

Restart Time {=} Time

Summary

Specify restart file read at a specified time.

Description

NOTE: This command must be placed at the Sierra scope of the input file.

Specify the time that the analysis will be restarted. In addition to this line command, each Region in the analysis (strictly, only the region(s) that will be restarted) must have a restart block specifying the database to read the restart state data. The restart ‘time’ must be greater than zero and less than or equal to the termination time.

By default, use of this command will cause previous output files (e.g., results, history, heartbeat, restart) to be overwritten. If this command is chosen, the onus is placed on the user to ensure that previous output files are not overwritten.

Parameter	Value	Default
{=}	{= | are | is}	–
Time	real	–