Hi, I'm working on an android application for the analysis of measurements (e.g, blood pressure) using a Bayesian network model (PEModelOR is the class in my android application), which is embedded in a smart phone. Since the measurements are to be taken at sequential times, the model contains a number of variables referring to different time controls. With the initial version with 2 controls the model run without a problem but when I extended the model to 4 controls it crashed producing the following error:
10-12 16:50:15.291: WARN/dalvikvm(4848): VFY: arbitrarily rejecting large method (regs=134 count=17618) 10-12 16:50:15.291: WARN/dalvikvm(4848): VFY: rejected Lcom/example/ pregmonit/PEModelOR;.<init> ()V 10-12 16:50:15.291: WARN/dalvikvm(4848): Verifier rejected class Lcom/ example/pregmonit/PEModelOR; 10-12 16:50:15.311: DEBUG/AndroidRuntime(4848): Shutting down VM 10-12 16:50:15.311: WARN/dalvikvm(4848): threadid=1: thread exiting with uncaught exception (group=0x40027e08) 10-12 16:50:15.341: ERROR/AndroidRuntime(4848): FATAL EXCEPTION: main 10-12 16:50:15.341: ERROR/AndroidRuntime(4848): java.lang.VerifyError: com.example.pregmonit.PEModelOR 10-12 16:50:15.341: ERROR/AndroidRuntime(4848): at com.example.pregmonit.EBayes.<init>(EBayes.java:20) 10-12 16:50:15.341: ERROR/AndroidRuntime(4848): at com.example.pregmonit.EBayes.android_main(EBayes.java:78) 10-12 16:50:15.341: ERROR/AndroidRuntime(4848): at com.example.pregmonit.Status.onCreate(Status.java:54) I saw that in a recent discussion http://groups.google.com/group/android-developers/browse_thread/thread/2126a73ba0cf7ade Dirk has encountered the same problem but unfortunately I could not see any solution to that. The extended PEModelOR method runs without problems in Java but it fails in the android application due to the dalvikvm verifier as far as I understood. The PEModelOR method is a description of a Bayesian Network in java and it simply contains definition of 62 variables and 62 probability distribution tables (in total 700 lines and the java file is 41 KB); below is a sample of the code (note that BayesNet is a class loaded via external jar): import BayesianNetworks.*; public class PEModelOR extends BayesNet { public PEModelOR() { name = "PEModelOR"; ...... final DiscreteVariable DB = new DiscreteVariable ("DB", DiscreteVariable.CHANCE, new String[] { "yes","no" }); ....... DiscreteFunction p62= new DiscreteFunction( new DiscreteVariable[] {DB}, new DiscreteVariable[] {FHDiab}, new double[] {0.0052, 0.2, 0.9948, 0.8 }); ....... Does the size/definition of PEModelOR cause the problem? I'm not sure how I can split it as it needs to be loaded at once for creating the network and running the model with available measurements. The thing is that this is even not the final version of the model, where the goal is to have 11 time controls, ending up with 112 variables and probability tables, and 1245 lines. Is there a way out or alternative of implementing PEModelOR so that it does not fail at verification? I will highly appreciate any help and suggestions on solving this problem. Thanks a lot in advance. Cheers, MarVel -- You received this message because you are subscribed to the Google Groups "Android Developers" group. To post to this group, send email to android-developers@googlegroups.com To unsubscribe from this group, send email to android-developers+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/android-developers?hl=en
