Optimization Action Set

Tuning a Single Problem by Using the Default Set of Parameters

This section contains PROC CAS code.

Note: Input data must be accessible in your CAS session, either as a CAS table or as a transient-scope table. A CAS table has a two-level name: the first level is your CAS engine libref, and the second level is the table name. You refer to this table in the CAS procedure by specifying only the second level. For more information about two-level names, see Chapter 3, Shared Concepts (SAS Optimization: Mathematical Optimization Procedures). A transient-scope table is called directly from the action and exists in memory for the duration of the action. For more information about accessing data, see SAS Viya: System Programming Guide. For more information about PROC CAS and programming in CASL, see SAS Cloud Analytic Services: CASL Programmer’s Guide and SAS Cloud Analytic Services: CASL Reference.

This example illustrates how to use the tuner action. The standard set of MILP benchmark cases is called MIPLIB and can be found at http://miplib.zib.de/. Suppose you want to tune the solveMilp action’s parameters on the problem air05 from this set by using the tuner action’s default set of tuning parameters. The %mps2sasd macro in this example converts the air05.mps file stored in the current directory on your local machine to a CAS table, air05, that is accessible in your CAS session. In the %mps2sasd macro function, you need to specify both the path and the MPS filename. For more information about loading an MPS data set onto a CAS server, see Getting Started: OPTMILP Procedure, Getting Started: OPTMILP Procedure (SAS Optimization: Mathematical Optimization Procedures).

The following code calls the tuner action to tune the data table air05 by using the default tuning parameter set:

libname caslib cas;
filename air05 "path-to-air05.mps file";

%mps2sasd(mpsfile=air05, outdata=caslib.air05);

proc cas;
loadactionset 'optimization';
tuner /
   instances={{data='air05'}}
   milpParameters={maxtime=100, seed=73737}
   tunerParameters={maxconfigs=30, nthreads=4, logfreq=5}
;
run;
quit;

The maxtime subparameter in the milpparameters parameter sets the maximum run time that the action can use to solve one problem for each parameter configuration. This subparameter is required.

The maxconfigs subparameter in the tunerparameters parameter sets a limit on the maximum number of parameter configurations that the tuner action processes to solve the problems on the list. You can also use the maxtime subparameter in the tunerparameters parameter to set the tuner action’s maximum run time. The user must specify at least one of these two subparameters in the tunerparameters parameter.

The nthreads subparameter in the tunerparameters parameter specifies the number of threads that the tuner action can use for tuning. Each thread solves an independent MILP problem.

Output 2.15.1 shows the tuner’s iteration log. The first column shows the number of solved problems. The second column shows the number of parameter configurations tested. The third column shows the best run time found so far. The last column shows the tuner action’s run time. You can use the loglevel or logfreq subparameter in the tunerparameters parameter to control the printing or print frequency.

Output 2.15.1: Tuner Action Log


NOTE: Active Session now MYSESS.
NOTE: Added action set 'optimization'.
NOTE: Start to tune the MILP
SolveCalls Configurations BestTime Time
0 0 . 0.00
5 5 17.98 36.78
10 10 17.98 77.69
15 15 11.83 100.44
20 20 11.83 112.06
25 25 11.45 131.01
30 30 11.45 172.39
NOTE: The tuning time is 172.39 seconds.

Output 2.15.2 shows standard ODS output tables that are created by the tuner action, which include the performance information table, tuner information table, tuner summary table, and tuner results table.

Output 2.15.2: Tuner Action Output

Results from optimization.tuner

Performance Information
Execution Mode	Distributed
Number of Compute Nodes	3
Number of Threads per Node	4

Tuner Information
Target Solver	MILP
Number of Tuning Options	12
Number of Tuning Instances	1
Tuning Option Set	AUTOMATIC
Performance Goal	GEOMEAN
Tuner Time Limit	1.797693E308
Tuner Configurations Limit	30

Tuner Summary
Actual Tuning Time	172.38
Initial Run Time (geomean)	17.97
Initial Run Time (sum)	17.97
Best Run Time (geomean)	11.44
Best Run Time (sum)	11.44
Number of Tested Configurations	30
Number of Improved Configurations	2

Tuner Results
Configuration	conflictSearch	cutGomory	cutMiLifted	cutStrategy	cutZeroHalf	heuristics	nodelSel	presolver	probe	restarts	symmetry	varSel	Mean of Run Times	Sum of Run Times	Percentage Successful
0	automatic	automatic	automatic	automatic	automatic	automatic	automatic	automatic	automatic	automatic	automatic	automatic	17.97	17.97	100
1	automatic	automatic	automatic	automatic	automatic	automatic	automatic	automatic	automatic	automatic	none	automatic	11.44	11.44	100
2	moderate	aggressive	moderate	moderate	aggressive	moderate	bestBound	aggressive	basic	moderate	none	pseudo	11.82	11.82	100
3	aggressive	none	aggressive	none	moderate	automatic	bestBound	moderate	automatic	none	moderate	ryanFoster	25.93	25.93	100
4	moderate	none	automatic	automatic	none	automatic	automatic	none	none	basic	automatic	pseudo	29.78	29.78	100
5	moderate	moderate	moderate	moderate	aggressive	moderate	bestBound	basic	basic	moderate	none	pseudo	30.17	30.17	100
6	moderate	aggressive	moderate	moderate	aggressive	moderate	bestBound	basic	basic	moderate	moderate	pseudo	30.56	30.56	100
7	moderate	aggressive	moderate	moderate	aggressive	moderate	automatic	basic	basic	moderate	none	pseudo	32.01	32.01	100
8	moderate	aggressive	moderate	moderate	automatic	moderate	bestBound	basic	basic	moderate	none	pseudo	33.79	33.79	100
9	moderate	aggressive	moderate	moderate	aggressive	moderate	bestBound	basic	basic	moderate	none	automatic	34.27	34.27	100
10	aggressive	aggressive	moderate	moderate	aggressive	moderate	bestBound	basic	basic	moderate	none	pseudo	35.67	35.67	100
11	moderate	aggressive	moderate	moderate	aggressive	moderate	depth	basic	basic	moderate	none	pseudo	36.7	36.7	100
12	moderate	automatic	moderate	moderate	aggressive	moderate	bestBound	basic	basic	moderate	none	pseudo	36.77	36.77	100
13	moderate	aggressive	moderate	moderate	aggressive	moderate	bestBound	moderate	basic	moderate	none	pseudo	42.56	42.56	100
14	moderate	aggressive	moderate	none	aggressive	moderate	bestBound	basic	basic	moderate	none	pseudo	50.97	50.97	100
15	automatic	aggressive	moderate	moderate	aggressive	moderate	bestBound	basic	basic	moderate	none	pseudo	51.77	51.77	100
16	moderate	aggressive	moderate	moderate	aggressive	automatic	bestBound	basic	basic	moderate	none	pseudo	55.73	55.73	100
17	moderate	aggressive	moderate	moderate	aggressive	moderate	bestBound	basic	none	moderate	none	pseudo	59.7	59.7	100
18	moderate	none	automatic	moderate	aggressive	none	bestBound	moderate	automatic	none	automatic	ryanFoster	65.94	65.94	100
19	moderate	aggressive	moderate	moderate	aggressive	moderate	bestBound	basic	basic	none	none	pseudo	66.23	66.23	100
20	moderate	aggressive	moderate	moderate	aggressive	moderate	bestBound	basic	basic	moderate	none	pseudo	66.73	66.73	100
21	moderate	aggressive	moderate	moderate	moderate	moderate	bestBound	basic	basic	moderate	none	pseudo	68.92	68.92	100
22	moderate	aggressive	aggressive	moderate	aggressive	moderate	bestBound	basic	basic	moderate	none	pseudo	69.74	69.74	100
23	moderate	aggressive	moderate	moderate	aggressive	moderate	bestBound	basic	basic	moderate	aggressive	pseudo	71.64	71.64	100
24	moderate	aggressive	moderate	moderate	aggressive	moderate	bestBound	basic	basic	basic	none	pseudo	72.83	72.83	100
25	none	none	moderate	aggressive	none	automatic	bestEstimatedepth	moderate	none	basic	none	automatic	100.85	100.85	0
26	moderate	aggressive	moderate	moderate	aggressive	moderate	bestBound	basic	basic	moderate	none	maxInfeas	103.35	103.35	0
27	automatic	none	none	moderate	moderate	automatic	automatic	none	automatic	none	basic	minInfeas	100.43	100.43	0
28	none	moderate	moderate	none	aggressive	none	depth	automatic	automatic	automatic	none	minInfeas	100.06	100.06	0
29	aggressive	none	none	aggressive	moderate	none	depth	automatic	automatic	none	aggressive	minInfeas	100.72	100.72	0

Tuning a Single Problem by Using the Default Set of Parameters

This section contains Lua code for the analysis in the CASL version of this example, which contains details about the results.

Note: In order to run this code, the data that are described in the CASL version need to be loaded into CAS. One way to do this is to convert the example data to the comma-separated-value (CSV) file air05.csv and then use the following code to load the CSV file into CAS:

s:loadtable{casLib="casuser", path="air05.csv"}

For more information about coding in Lua, see Getting Started with SAS Viya for Lua and SAS Viya: System Programming Guide.

The following code calls the tuner action to tune the data table air05 by using the default tuning parameter set:

s:optimization_tuner{
   instances={{data={name="air05"}}},
   milpParameters={maxTime=30},
   tunerParameters={maxConfigs=30, nThreads=4, logFreq=5}
}

Tuning a Single Problem by Using the Default Set of Parameters

This section contains Python code for the analysis in the CASL version of this example, which contains details about the results.

Note: In order to run this code, the data that are described in the CASL version need to be loaded into CAS. One way to do this is to convert the example data to the comma-separated-value (CSV) file air05.csv and then use the following code to load the CSV file:

s.upload('air05.csv')

For more information about coding in Python, see Getting Started with SAS Viya for Python and SAS Viya: System Programming Guide.

The following code calls the tuner action to tune the data table air05 by using the default tuning parameter set:

s.optimization.tuner(
    instances=[{'data':{'name':'air05'}}],
    milpparameters={'maxtime':100},
    tunerparameters={'maxconfigs':30,'nthreads':4,'logfreq':5}
)

Tuning a Single Problem by Using the Default Set of Parameters

This example is not available for the R programming language.

Last updated: April 22, 2022