Update BM7 to use papermill workflow

[pfhub]

@wd15, the new version of the PFHub website is available at https://random-cat-1505.surge.sh

[wd15]

@tkphd, ready for review, see https://random-cat-1505.surge.sh/results/benchmark7a.0.ipynb/

[pfhub]

@wd15, the new version of the PFHub website is available at https://random-cat-1505.surge.sh

[tkphd]

Looks good! It would be better if the color for each dataset matched between the two graphs. Is that possible?

[pfhub]

@wd15, the new version of the PFHub website is available at https://random-cat-1505.surge.sh

[pfhub]

@wd15, the new version of the PFHub website is available at https://random-cat-1505.surge.sh

[wd15]

@tkphd, please review again. Colors are consistent across line plots now.

[wd15]

@tkphd, this is ready for review now

[tkphd]

Much better, @wd15.

The accuracy plot grid lines are unconventional: equal spacing leads one to expect linear scales, but these are log-log plots. Especially since the values are less important than the slope of the trendlines, can these plots be regenerated using more standard log-scaled semiology, i.e., grid lines at the powers of 10 with minor ticks along the axes showing log-spacing?

[tkphd]

(Implicitly, in the trendline labels, please use $\mathcal{O}(•)$)

[wd15]

* _Optional:_

  * For both accuracy plots, the grid lines are unconventional: equal spacing leads one to expect
    linear scales, but these are log-log plots. Especially since the values are less important

What? What's unconventional? No idea what you mean.

    than the slope of the trendlines, can these plots be regenerated using more standard
    log-scaled semiology? I.e., grid lines at the powers of 10 with minor ticks along the axes
    showing log-spacing.

No idea about this. It looks like a standard log-log plot to me.

[wd15]

@tkphd: all the titles and legend names should be better now.

[pfhub]

@wd15, the new version of the PFHub website is available at https://random-cat-1505.surge.sh

[wd15]

@tkphd: can you review again? I didn't do the optional part as it seemed onerous.

[tkphd]

Looks much better, thanks. A minor quibble: in the first plot, $\mathcal{O}(k)$ is dashed and $\mathcal{O}(k^2)$ is dotted; in the second, $\mathcal{O}(h)$ is dotted and $\mathcal{O}(h^2)$ is dashed. Switching the order of one or the other for consistency would be preferable.

[pfhub]

@wd15, the new version of the PFHub website is available at https://random-cat-1505.surge.sh

[wd15]

@tkphd, should be better now. Please take a look again.

[pfhub]

@wd15, the new version of the PFHub website is available at https://random-cat-1505.surge.sh