Crash when look up table does not cover the range

[RanpengLi]

I add an option to make the model crash when the coordinate value is out of the range of the look-up table. I also manually set it for entropy model to make sure the look-up table always cover the entropy-pressure range.

• I have followed the instructions for indenting my code.

[bangerth]

Strictly speaking, this is an incompatible change. Do you want to document that the entropy reader now needs to be initialized with something that encompasses all possible pressures?

[bangerth]

Can you document what the parameter represents in the block above the function?

[bangerth]

Suggested change

	bool crash_if_not_in_range) const
	const bool crash_if_not_in_range) const

[bangerth]

This is expensive because you are copying the whole vector. But you can make it cheaper:

Suggested change

	const std::vector<double> x_coordinates = get_interpolation_point_coordinates(0);
	const std::vector<double> &x_coordinates = get_interpolation_point_coordinates(0);

Same for the place a few lines down.

[bangerth]

There are also a number of failing tests:

The following tests FAILED:
	312 - entropy_initial_lookup (Failed)
	313 - entropy_initial_lookup_wb (Failed)
	314 - entropy_plasticity (Failed)

What do we want to do about these tests?

[RanpengLi]

There are also a number of failing tests:

The following tests FAILED:
	312 - entropy_initial_lookup (Failed)
	313 - entropy_initial_lookup_wb (Failed)
	314 - entropy_plasticity (Failed)

What do we want to do about these tests?

Hi Wolfgang @bangerth , thank you for looking at this pull request! I have modified it and addressed your comments.

For the failed test, 314 failed because the entropy range in the lookup table (600-3000) doesn't cover the model range (up to 3021). I have decreased the maximum entropy in this test setup and will also need to update the result.

For 312 and 313, the model is trying to look up data with pressure = 0, while the look-up table has a pressure range start from 0.25 bar. @lhy11009 Haoyuan, I am mentioning you because you made these two tests. Would it be ok if I just simply change the 0.25 bar to 0 bar in your data table?

For the entropy look-up method, we definitely want the lookup table to cover the full entropy-pressure range of the model. So the three failed tests shows the exact situation we want to avoid with this new assert throw ;)

[lhy11009]

Hi Ranpeng, thanks for mentioning me in this. Sorry, I didn't see the question in your original post. It's great that you can adjust my tests to accommodate previous failures. I'll do a quick read-through of your commits and see what I can contribute.

[RanpengLi]

Hi Wolfgang, Haoyuan, and Rene @gassmoeller ,

Thank you for your help on this pull request! I have addressed Wolfgang's comments by:
-Adding a change log.
-Adding a description of the new parameter.
-Including some fixes in the code.

Also, I modified the test setup for the tests that failed with the following changes:

312 - entropy_initial_lookup (Failed) &
313 - entropy_initial_lookup_wb (Failed)

Both crashed because the pressure in the model starts from 0 Pa, but the lookup table starts from 25,000 Pa. To address this, I changed the surface pressure from 0 Pa to 25,000 Pa in the test setup. With the new surface pressure, I tested the results before and after the code modification on my local tester, and the results are identical.

314 - entropy_plasticity (Failed)

This test case crashed because the entropy in the model domain exceeded the range provided by the lookup table. To fix this, I decreased the maximum entropy in the setup. I think this could be more reasonable than including a large lookup table that covers the P-T range of the entire mantle.