~fluidity-core/fluidity/adjoint

<?xml version="1.0" encoding="UTF-8" ?>

<!DOCTYPE testproblem SYSTEM "regressiontest.dtd">

<testproblem>

  <name>Burgers Equation MMS</name>

  <owner userid="pef"/>

  <tags>burgers adjoint revolve</tags>

  <problem_definition length="medium" nprocs="1">

    <command_line>../../bin/optimality op_A.oml; ../../bin/optimality op_B.oml; ../../bin/optimality op_C.oml; burgers_equation mms_A.bml; burgers_equation mms_B.bml; burgers_equation mms_C.bml; burgers_equation mms_D.bml; burgers_equation mms_E.bml</command_line>

  </problem_definition>

  <variables>

    <variable name="gradient_conv" language="python">

from fluidity_tools import stat_parser

import glob

gradient_conv = [stat_parser(x)["time_integral_ad_gradient_error"]["convergence"][-1] for x in sorted(glob.glob("op_?.stat"))]

    </variable>

    <variable name="functional_value_conv" language="python">

from fluidity_tools import stat_parser

import glob

import math

 

functional_errors = [abs(stat_parser(x)["time_integral_ad"]["value"][-1] - 10.0) for x in sorted(glob.glob("mms_adjoint_?.stat"))]

functional_value_conv = [math.log(functional_errors[i]/functional_errors[i+1], 2) for i in range(0, len(functional_errors)-1)]

    </variable>

  </variables>

  <pass_tests>

    <test name="functional_convergence" language="python">

assert min(functional_value_conv) &gt; 1.9

    </test>

    <test name="gradient_convergence" language="python">

assert min(gradient_conv) &gt; 1.8 # more tolerant because it's stochastic

    </test>

  </pass_tests>

  <warn_tests>

  </warn_tests>

</testproblem>