Differences

This shows you the differences between two versions of the page.

--- svmranker:usage [2020/11/17 22:20] – [Usage] merlin
+++ svmranker:usage [2020/11/20 14:10] – merlin
@@ Line 1: / Line 1: @@
- **[[svmranker:main| Main ]]** ||  **[[svmranker:main| Tool ]]** ||  **[[svmranker:usage| Usage & Case Study ]] **
+{{:svmranker:main.png?nolink|}}**[[svmranker:main| Main ]]{{:svmranker:tool.png?nolink|}}[[svmranker:tool| Tool ]]{{:svmranker:usage.png?nolink|}}[[svmranker:usage| Usage & Case Study ]] **
 ====== Usage ======
 In the following, we assume that current directory is SVMRanker.
@@ Line 9: / Line 9: @@
   Usage: CLIMain.py [OPTIONS] COMMAND [ARGS]...
     "python3 CLIMain.py COMMAND --help" for more details
   Options:
     --help  Show this message and exit.
   Commands:
     lmulti
     lnested
-    parseboogie
+   parseboogie
-    parsectoboogie
+   parsectoboogie
-    parsectopy
+   parsectopy
-As we can see, SVMRanker provides five commands. The first two commands allow for proving termination of a given program while the remaining three can be used for parsing the input file and translate it to a different format. In the remaining part of the section we focus on the details for the use of the lmulti and lnested commands.
+As we can see, SVMRanker provides five commands. The first two commands allow for proving termination of a given program while the remaining three can be used for parsing the input file and translate it to a different format. In the remaining part of the section we focus on the details for the use of the **lmulti** and **lnested** commands.
-lmulti, short for learning multiphase ranking function, instructs SVMRanker to learn a multiphase ranking function for the given program. To get the detailed usage information for this command, one can use the following command.
+**lmulti**, short for learning multiphase ranking function, instructs SVMRanker to learn a multiphase ranking function for the given program. To get the detailed usage information for this command, one can use the following command.
   python3 ./CLIMain.py lmulti --help
 The output is the following.
-  SVMRanker --- Version 1.0
+<code>
+SVMRanker --- Version 1.0
 Usage: CLIMain.py lmulti [OPTIONS] SOURCE
 Options:
+ --depth_bound INTEGER           depth bound default set to 2
- <code>
+ --filetype [C|BOOGIE]           --file C: input is c file.
+                                 --file BOOGIE:input is boogie file. default set to BOOGIE
+ --sample_strategy [ENLARGE|CONSTRAINT]
+                                 --sample_strategy ENLARGE: enlarge the sample zone when sample num not enough.
+                                 --sample_strategy CONSTRAINT: find feasible points by constraint if sample
+                                                               num not enough default set to ENLARGE
+ --cutting_strategy [NEG|MINI|POS]
+                                 use f(x) < b to cut
+                                 --cutting_strategy POS: b is a postive number
+                                 --cutting_strategy NEG: b is a negative number
+                                 --cutting_strategy MINI: b is the minimum value of sampled points
+                                                          default set to MINI
+ --template_strategy [SINGLEFULL|FULL]
+                                 templates used for learning
+                                 --template_strategy SINGLEFULL: templates are either single variable
+                                                                 or combination of all variables
+                                 --template_strategy FULL: template is combination of all variables
+                                                           default set to SINGLEFULL
+ --print_level [DEBUG|INFO|NONE]
+                                 --print_level DEBUG: print all the information of the learning and debugging
+                                 --print_level INFO: print the information of the learning
+                                 --print_level NONE: only print the result information of the learning
+                                                     default set to NONE
+ --help                          Show this message and exit
+ </code>
+As the help shows, there are several options available to tune the execution of lmulti; we present their usage by means of a couple of examples. \\
+<code c Example1.c>
+ int main() {
+   int x, y;
+   while(x > 0 || y > 0) {
+     x = x + y - 1;
+     y = y - 1;
+   }
+ }
+</code>
+The is the first C program we consider here; see the file src/example/**Example1.c**. This program can be shown to be terminating by means of a 2-multiphase ranking function, as we get by running SVMRanker to learn its multiphase ranking function as follows.
+  python3 ./CLIMain.py lmulti **--filetype** C example/**Example1.c**
+SVMRanker completes the analysis by returning a 2-multiphase ranking function for the **Example1.c** program, as shown below.
+<file>
 SVMRanker --- Version 1.0
-Usage: CLIMain.py lmulti [OPTIONS] SOURCE
+example/**Example1.c**
+--------------------LEARNING MULTIPHASE SUMMARY-------------------
+MULTIPHASE DEPTH:  2
+LEARNING RESULT:  TERMINATE
+-----------RANKING FUNCTIONS----------
+.0 * 1 + 1.0 *  y^1 + 5.0 * 1
+.0796 *  x^1 + 0.482 * 1 + 0.482 * 1
+</file>
+Notice that we used the option **--filetype** to specify the type of the input program, given that SVMRanker supports both Boogie programs and C programs as input file, with the former being the default format. Furthermore, we can also provide the option ****--depth_bound**** to set the maximal number of phases SVMRanker can use when learning a multiphase ranking function. The default value of this option is 2, and this has been enough in the analysis of **Example1.c**, since **Example1.c** can be proved to terminate by a 2-multiphase ranking function. Such a default value is suitable for several programs, but it can be increased as needed, as we will see with the second example src/example/**Example2.c**, as shown below.
-Options:
+<code c Example2.c>
-  --depth_bound INTEGER           depth bound default set to 2
+ int main() {
-  --filetype [C|BOOGIE]           --file C: input is c file.
+   int x, y;
-                                  --file BOOGIE:input is boogie file. default set to BOOGIE
+   while(x > 0) {
+     x = x + y;
-  --sample_strategy [ENLARGE|CONSTRAINT]
+     y = y + z;
-                                  --sample_strategy ENLARGE: enlarge the sample zone when sample num not enough.
+     z = z - 1;
-                                  --sample_strategy CONSTRAINT: find feasible points by constraint if sample
+   }
-                                                                num not enough default set to ENLARGE
+ }
+</code>
+If we run SVMRanker on **Example2.c** with the default value of 2 for **--depth_bound**, we obtain **Unknown** as result. In order to get an appropriate multiphase ranking function for **Example2.c**, one needs to call lmulti with the option ****--depth_bound**** set to at least 3.
+ python3 ./CLIMain.py lmulti **--filetype** C **--depth_bound** 3 example/**Example2.c**
+With the help of **--depth_bound**, SVMRanker produces the result shown below.
+<file>
+SVMRanker --- Version 1.0
+example/Example2.c
+--------------------LEARNING MULTIPHASE SUMMARY-------------------
+MULTIPHASE DEPTH:  3
+LEARNING RESULT:  TERMINATE
+-----------RANKING FUNCTIONS----------
+.0 * 1 + 2.0 * 1 + 1.0 *  z^1 + 2.0 * 1
+.0 * 1 + 0.2154 *  y^1 + 1.0 * 1 + 1.0 * 1
+.0911 *  x^1 + 0.3226 * 1 + 0.3226 * 1 + 0.3226 * 1
+</file>
+We now present the other options that let SVMRanker use different strategies in the process of learning a multiphase ranking function; different strategies regarding how program data points are sampled, how the state space is cut, and what templates are used, influence the running time of SVMRanker and possibly the final result.
-  --cutting_strategy [NEG|MINI|POS]
+ * - -sample_strategy This option controls the strategy SVMRanker uses to sample program data points. Possible values are CONSTRAINT and ENLARGE (the default): CONSTRAINT samples randomly the points satisfying the loop condition/guard; This strategy can be slow as we need to get the assignments of variables satisfying the loop guard. ENLARGE samples uniformly the points in a predefined space, which can be very efficient; However, if all the sampled points cannot satisfy the loop guard, we will then try to enlarge the sampling space until seeing some point satisfying the loop guard.
-                                  use f(x) < b to cut
-                                  --cutting_strategy POS: b is a postive number
-                                  --cutting_strategy NEG: b is a negative number
-                                  --cutting_strategy MINI: b is the minimum value of sampled points
-                                                           default set to MINI
-  --template_strategy [SINGLEFULL|FULL]
+ * - -cutting_strategy This option controls the bound $b$ of the constraint f(x) < b that is used to cut the program state space in two parts for the current phase's decreasing function f. Possible values are NEG, POS, and MINI (the default): NEG chooses randomly a negative value for b; POS chooses randomly a positive value for b; MINI selects the minimum value of f on the sampled points.
-                                  templates used for learning
-                                  --template_strategy SINGLEFULL: templates are either single variable
-                                                                  or combination of all variables
-                                  --template_strategy FULL: template is combination of all variables
-                                                            default set to SINGLEFULL
-  --print_level [DEBUG|INFO|NONE]
+ * - -template_strategy This option controls what templates are used in the learning procedure. Possible values are FULL and SINGLEFULL (the default): FULL uses as templates the linear combinations of all program variables; SINGLEFULL extends FULL with templates using only one variable at a time.
-                                  --print_level DEBUG: print all the information of the learning and debugging
-                                  --print_level INFO: print the information of the learning --print_level
-                                                      NONE: only print the result information of the learning
-                                                      default set to NONE
-  --help                          Show this message and exit
+ * - -print_level This option controls the verbosity of the SVMRanker output.
-  </code>
-As the help shows, there are several options available to tune the execution of lmulti; we present their usage by means of a couple of examples.
+The SVMRanker command **lnested**, short for learning nested ranking function, is used for learning a nested ranking function for a given program. The usage information of **lnested** can be obtained by the following command, with the output below.
+  python3 ./CLIMain.py **lnested** --help
+The output is the following.
+<file>
+SVMRanker --- Version 1.0
+Usage: CLIMain.py **lnested** [OPTIONS] SOURCE
-  <file java>
+Options:
+ --depth_bound INTEGER           depth bound default set to 2
-	int main() {
+ --filetype** [C|BOOGIE]           --file C: input is c file.
-    	int x, y;
+                                 --file BOOGIE: input is boogie file. default set to BOOGIE
-    	while(x > 0 || y > 0) {
-    		x = x + y - 1;
-    		y = y - 1;
-    	}
-	}
-  </file>
+ --sample_strategy [ENLARGE|CONSTRAINT]
+                                 --sample_strategy ENLARGE: enlarge the sample zone when sample num not enough.
+                                 --sample_strategy CONSTRAINT: find feasible points by constraint
+                                                  if sample num not enough default set to ENLARGE
+ --print_level [DEBUG|INFO|NONE]
+                                 --print_level DEBUG: print all the information of the learning and debugging
+                                 --print_level INFO: print the information of the learning
+                                 --print_level NONE: only print the result information of the learning
+                                                     default set to NONE
+ --help                          Show this message and exit.
+</file>
+As we can see, the options of ****lnested**** are also the ones of lmulti; also the use of **lnested** is similar to the one of **lmulti**, just the outcome can be different.
+For instance, we can prove termination of ****Example2.c**** by means of a learned nested ranking function by running SVMRanker as follows.
+  python3 ./CLIMain.py **lnested** **--filetype** C ****--depth_bound**** 3 example/**Example2.c**
+The output is shown below.
+<file>
+SVMRanker --- Version 1.0
+example/Example2.c
+--------------------LEARNING NESTED SUMMARY-------------------
+NESTED DEPTH:  3
+LEARNING RESULT:  TERMINATE
+-----------RANKING FUNCTIONS----------
+.0 *  z^1.0 + 0.9 * 1; 1.0 *  y^1.0 + 0.9 * 1; 1.0 *  x^1.0 + 0.7 * 1
+</file>