310f3251 test is unsolvable, wrap around

[neoneye]

Problem

ARC-Interactive

I was unable to solve this task.

My reasoning.

• Maintain same repeating tile.
• For every non-black pixel, insert a red pixel at relative position -1, -1.

My predicted solution

The expected solution

Difference

My mistake was that I assumed maintain same repeating tile, thus assuming that the red pixels would wrap around the edge.

Solution

Add a train pair, that demonstrates that the tiles are non-repeating.

[neoneye]

It's better now than before. However there are no priors in the training pairs that demonstrates no-wrap-around. IMHO it's somewhat counter intuitive that it's only the test pair.

Idea for improvement, tweak one of the train pairs to demonstrate that there is no wrapping around when drawing the red pixels.

This task was updated in commit: b7fd42c

The task now looks like this in ARC-Interactive

[CerebralAdvantage]

Agreed. It's like this particular puzzle was created and is being maintained by an intern who is unclear on the concept of "train". No intelligent thing would assume "no local wrapping", because of example 1 and ESPECIALLY example 2. You need at least one train example that clearly shows this detail. Otherwise you might was well be shooting for a score of 99. If there is any evidence to the contrary, please present it. Without that, our only evidence is that the Test input was changed after the first issue was reported.

[fchollet]

Thanks for the reports. I issued a fix for this task (live at HEAD). The fix is to add wraparound and add a task to demonstrate it.