Closed jlperla closed 7 years ago
I have fixed most errors. But there is still one that comes from HJBE_discretized_univariate_test/bigger_value_function_test. I saved the new csv file generated by this test and compared it with the old one. I noticed that some entries between two files have a difference about 1e^-10, and the aggregate difference over 10000 entries is 4.8654e^-6. I guess the error is caused by different ways we used to do the division. Do you think this difference is acceptable?
Totally acceptable. Feel free to update the test csv files if that is the only problem, and close the ticket if the test suite runs
Now all tests can be successfully completed.
After your modification and tests in
discretize_univariate_diffusion_test.m
some of the other files are broken because of the change to multiplying $A$ by $\Delta$.Can you fix this by modifying the following functions:
/matlab/lib/simple_HJBE_discretized_univariate.m
/matlab/lib/simple_joint_HJBE_stationary_distribution_univariate.m
/matlab/lib/simple_optimal_stopping_diffusion.m
/matlab/lib/stationary_distribution_discretized_univariate.m
/matlab/examples/optimal_stopping/LCP_example.m
The modifications will be pretty simple. In particular, mostly multiplying by $\Delta$ where currently there is an $A$ to follow the notation and equation numbers in
operator_discretization_finite_differences.pdf
andoptimal_stopping.pdf
You will know you are successful when you can run all of the tests in the matlab folder (i.e.
run_tests.m
) with no errors. Also, the/matlab/examples/optimal_stopping/LCP_example.m
will require running manually.