Fisher two period

[AMonninger]

This PR adjusts the notebook in the following ways:

1. Updating Seaborn import to avoid warning.
2. Changing numbers of decimal places shown in widgets in line with step size.
3. Writing all labels in Latex notation with lower values for time periods.
4. Replace notation from "RHi" and "RLo" to "RInit" and "RNew".
5. Figure 4 works now (Problem was that PermGroFac cannot be exactly zero).

[AMonninger]

@llorracc can you have a look at it?

[llorracc]

Updating Seaborn import to avoid warning.

I wanted you to try to fix things so that the warning was no longer presented, regardless of the log level.
That is, fix the bug, don't hide it. This presumably involves updating the version of seaborn that is installed in the HARK/binder/requirements.txt file; when you do that, you should trigger a test of whether that breaks any of the other DemARKs. If you are not sure how to do all of this, please ask Seb or Matt.

(Also, when I run your notebook, I am still getting the error message, so I'm puzzled you are not).

Figure 4 works now (Problem was that PermGroFac cannot be exactly zero).

Please post it as an issue in HARK that we should be able to accommodate PermGroFac of zero.

Writing all labels in Latex notation with lower values for time periods.

That was not the task. The task was to have identical variable names in the python code as in the plot labels. So the VARIABLE in python should not be B1, it should be b_1. (Please match the case in the lecture notes, where variables are lower case).

PS. I've turned this into a "draft" PR. Am trying to learn how draft PR's work.

[mnwhite]

A permanent income growth factor of zero will break both HARK and our models in various ways, at least for models that normalize by permanent income. All of the representations will fail completely. Everything will work properly if someone approximates zero as PermGroFac = 0.0001.

…

On Tue, Sep 12, 2023 at 9:46 PM Christopher Llorracc Carroll < ***@***.***> wrote: Updating Seaborn import to avoid warning. I wanted you to try to fix things so that the warning was no longer presented, regardless of the log level. That is, fix the bug, don't hide it. This presumably involves updating the version of seaborn that is installed in the HARK/binder/requirements.txt file; when you do that, you should trigger a test of whether that breaks any of the other DemARKs. If you are not sure how to do all of this, please ask Seb or Matt. (Also, when I run your notebook, I am still getting the error message, so I'm puzzled you are not). Figure 4 works now (Problem was that PermGroFac cannot be exactly zero). Please post it as an issue in HARK that we should be able to accommodate PermGroFac of zero. Writing all labels in Latex notation with lower values for time periods. That was not the task. The task was to have identical variable names in the python code as in the plot labels. So the VARIABLE in python should not be B1, it should be b_1. (Please match the case in the lecture notes, where variables are lower case). PS. I've turned this into a "draft" PR. Am trying to learn how draft PR's work. — Reply to this email directly, view it on GitHub <#206 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADKRAFNDLNNBGDR4BVEYLN3X2EGBFANCNFSM6AAAAAA4VJBU7E> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[AMonninger]

Seaborn:

• styles in matplotlib and seaborn have gone out of sync recently. https://matplotlib.org/stable/api/prev_api_changes/api_changes_3.6.0.html#seaborn-styles-renamed
• Option 1: We keep using the old matplotlib styles:

plt.style.use("seaborn-v0_8-darkgrid")

• Option 2: We use the style from seaborn directly

import seaborn as sns
sns.set_style("darkgrid")

• In the latest push I used option 2

Variables in Latex:

• I changed variable names now accordingly (B1 changed to b_1)
• I changed many variable names to lower cases and kept only aggregate variables with an upper case (CRRA, R)

PermGroFac:

• In my earlier push, I changed PermGroFac from 0 to 0.0001 to avoid a break in the code (as @mnwhite suggested)
• If you have very strong feelings that a user should be able to declare PermGroFac of 0, we could add two lines of code replacing any input of PermGroFac = 0.0 to PermGroFac = 0.0001

[AMonninger]

@alanlujan91 Any idea why the checks are failing? It seems to fail because of something unrelated to the notebook I changed. I'm still using an old python version 3.8.8. Do I need to update to 3.11?

I also cannot fetch the latest version of DemArk... https://github.com/AMonninger/DemARK

[alanlujan91]

I think the error is because @mnwhite refactored the check_conditions part of HARK, so when the tests pull the latest version of HARK, it makes a different notebook fail. You are correct that this is not related to your notebook, but it might have to be fixed (in a different PR) before yours can be merged.

[alanlujan91]

@sbenthall you might be interested in this PR

[llorracc]

@ashishk87,

It seems that I asked Adrian for some more extensive changes last year. Could you look over the thread above and his PR and revise your PR so that it incorporates whatever is worthwhile about Adrian's edits in addition to your own?

[llorracc]

@AMonninger, @ashishk87 is working on merging his PR for this year with yours from last year to incorporate the improvements from both.

Yours is generating conflicts. Could you help Ashish resolve the conflicts and get your PR working, so he can meld them together?

[Copilot]

Copilot wasn't able to review any files in this pull request.

[llorracc]

@ashishk87,

could you review this whole conversation and see if it contains anything we should incorporate in the notebook? And if not, could you see if you have permission to close the issue, and if not, confer with Adrian and if you both agree the issue should be closed email me?