some matplotlib figures not showing up

[fkazemian]

[fkazemian]

aiyagari.html

[mmcky]

thanks @fkazemian I think this is an issue with the code which is not showing the matplotlib figure in the compiled notebook. It may need the matplotlib cell magics. @natashawatkins when you removed the output from the files (such as aiyagari) did you run the resultant notebook to check if images show up in the notebook?

[natashawatkins]

Yes.

This is a bug with either matplotlib or the jupyter notebook - if you re-run the cell, the image will show up. For some reason it seems that matplotlib needs to be initialised once before it displays images inline.

[natashawatkins]

Ah I see this is in the generated notebook... um yes ok this seems to be a recent bug with Jupyter notebooks.

Here is a discussion of the issue: holoviz/holoviews#68

Can we change the settings for matplotlib?

[mmcky]

@natashawatkins any changes are just required to be in a code-block for it to be contained in the notebook. Are suggesting we update the lecture to have these changes to make showing images more robust?

[natashawatkins]

No I was wondering whether it would be possible to edit the config file for matplotlib as suggested in the thread. The location should be ~/.matplotlibrc. They suggest changing the default backend could fix the issue.

[fkazemian]

In amss.html, it seems codes that creat matplotlib figures, are not running.

[fkazemian]

amss2.html

[natashawatkins]

That's weird, maybe it's timing out...

We should definitely remove those print statements, on another note

[mmcky]

@fkazemian is this happening locally for you also?

When I built the html via IPYNB for the compare-lectures site -- it might have timed out the execution for this specific lecture. Can you confirm if the amss2 lecture works locally?

[fkazemian]

arma:

Hi @mmcky I don't have this problem when I'm running it locally:

[fkazemian]

(career)

but it works fine locally

[fkazemian]

(debugging)

but it works fine locally:

[fkazemian]

discrete_dp:

[mmcky]

hey @fkazemian when I run arma locally -- the first time I get the text output and the second run I get the figure. That is a bit strange. Do you get the same?

[fkazemian]

Hi @mmcky , no actually in the first time run I got the figure, and also I just checked arma locally again and I got the figure in my first run. I am using the computer in the lab ( Number1).

[fkazemian]

(estspec)

[fkazemian]

(finite_markov)

[fkazemian]

(getting_started)

[fkazemian]

(hist_dep_policies)

[fkazemian]

(ifp)

[fkazemian]

(jv)

[fkazemian]

(kalman)

[fkazemian]

(lake_model)

[fkazemian]

(linear_algebra)

[fkazemian]

(lln_clt)

[fkazemian]

(lqcontrol)

[fkazemian]

(lu_tricks)

[fkazemian]

(lucas_model)

[fkazemian]

(markov_asset)

[fkazemian]

mccall_model_with_separation

[fkazemian]

(mle)

[fkazemian]

(numba)

[fkazemian]

odu

[fkazemian]

ols:

[fkazemian]

opt_tax_recur:

[fkazemian]

optgrowth:

[fkazemian]

pandas:

[fkazemian]

perm_income

[fkazemian]

python_by_example

[fkazemian]

python_oop

[fkazemian]

samuelson

[fkazemian]

schelling

[fkazemian]

scipy

[fkazemian]

smoothing

[fkazemian]

stationary_densities

[fkazemian]

wald_friedman

[fkazemian]

writing_good_code

[mmcky]

I have edited the builder to prepend %matplotlib inline prior to executing a notebook to ensure matplotlib is setup correctly. The prepended line is then removed prior to writing the executed notebook to disk.

[mmcky]

@fkazemian I have pushed the latest to compare-lectures.quantecon.org. Would you mind checking if this is fixed. There may still be a few missing if execution time takes longer than 1200s.

[fkazemian]

Hi @mmcky sure, I'll check all of them

[fkazemian]

writing_good_code

wald_friedman

stationary_densities

smoothing

scipy

schelling

samuelson

python_oop

python_by_example

perm_income

[fkazemian]

pandas

optgrowth

opt_tax_recur

ols

odu

numba

mle

mccall_model_with_separation

matplotlib

[fkazemian]

Hi @mmcky I checked about 20 of them (as you can see above) and all of them are fixed, so it seems this bug is fixed and I change the label to Fixed, Thanks

[mmcky]

thanks @fkazemian that's a big help. Good news.