Opened 4 years ago

Last modified 3 years ago

#6312 new defect

dgesv failing in windows when computing Ax=b

Reported by: arunkumar palanisamy Owned by: Andreas Heuermann
Priority: high Milestone:
Component: Run-time Version: 1.16.0
Keywords: Cc:

Description

When computing numerical computation for data Reconciliation, the dgesv solver fails to compute Ax=b in windows, where as for the same problem the solution is successfully computed in linux. An example to reproduce the problem is available in the testsuite in the following location

testsuite/openmodelica/dataReconciliation/TSP_FourFlows11.mos
testsuite/openmodelica/dataReconciliation/TSP_FourFlows10.mos

use the old FrontEnd to run the tests

omc -d=-newInst TSPFourFlows11.mos

I am attaching the numerical computation log for TSP_FourFlows11.mos ran from linux and windows, look for the heading in the log file

Calculations of Matrix (F*Sx*Ft) f* = c(x,y) 
============================================

From the log file we can see that the calculation made for (F*Sx*Ft) and c(x,y) is the same for both in windows and linux, but when computing f*, dgesv cannot compute the solution in windows.

Attachments (2)

NewDataReconciliationSimpleTests.TSP_FourFlows11_debug.txt (17.8 KB ) - added by arunkumar palanisamy 4 years ago.
debug log from windows
NewDataReconciliationSimpleTests.TSP_FourFlows11_debug_linux.txt (90.1 KB ) - added by arunkumar palanisamy 4 years ago.
debug log from linux

Download all attachments as: .zip

Change History (6)

by arunkumar palanisamy, 4 years ago

debug log from windows

by arunkumar palanisamy, 4 years ago

debug log from linux

comment:1 by Andreas Heuermann, 4 years ago

I tried testsuite/openmodelica/dataReconciliation/TSP_FourFlows11.mos with todays master and for me this fails on Linux and on Windows.

record SimulationResult
    resultFile = "",
    simulationOptions = "startTime = 0.0, stopTime = 1.0, numberOfIntervals = 500, tolerance = 1e-06, method = 'dassl', fileNamePrefix = 'NewDataReconciliationSimpleTests.TSP_FourFlows11', options = '', outputFormat = 'mat', variableFilter = '.*', cflags = '', simflags = '-reconcile -sx=./NewDataReconciliationSimpleTests/resources/NewDataReconciliationSimpleTests.TSP_FourFlows9_Inputs.csv -eps=0.0023 -lv=LOG_JAC'",
    messages = "Simulation execution failed for model: NewDataReconciliationSimpleTests.TSP_FourFlows11
LOG_SUCCESS       | info    | The initialization finished successfully without homotopy method.
LOG_SUCCESS       | info    | The simulation finished successfully.
stdout            | info    | DataReconciliation Starting!
stdout            | info    | NewDataReconciliationSimpleTests.TSP_FourFlows11
stdout            | error   | solveSystemFstar() Failed !, The solution could not be computed, The info satus is 13
",
    timeFrontend = 0.6608404,
    timeBackend = 0.1515216,
    timeSimCode = 0.011052,
    timeTemplates = 0.09209050000000001,
    timeCompile = 0.900345,
    timeSimulation = 0.06415620000000001,
    timeTotal = 1.8801648
end SimulationResult;

The error message is identical as well and suggest that U(13,13) is 0, so the matrix A is singular.

So I would say it is not a problem depending on Windows or Linux.

comment:2 by arunkumar palanisamy, 4 years ago

@Andreas, then the test (TSP_FourFlows11.mos) should fail in jenkins as well, but it is working there and also I tested in dev.openmodelica.org server which is linux Linux 4.15.0-51-generic x86_64, The baseline of TSP_FourFlows11.mos is updated using dev server

comment:3 by Francesco Casella, 4 years ago

Milestone: 1.17.01.18.0

Retargeted to 1.18.0 because of 1.17.0 timed release.

comment:4 by Francesco Casella, 3 years ago

Milestone: 1.18.0

Ticket retargeted after milestone closed

Note: See TracTickets for help on using tickets.