Added ediff1d to numpy frontend

[Aneesh02]

Close #13356

[Aneesh02]

@Infrared1029 Request you to look at my PR

[Infrared1029]

Hello @Aneesh02, the test appears to be failing, can you make sure it passes so I can take a close look? Thanks!

[Aneesh02]

Hello @Infrared1029, Thanks for reviewing I have fixed the error in test_run_1. The linting error is not from the files I have changed.

[Infrared1029]

Hello @Aneesh02, the implementation looks good to me, just a few comments, make sure to follow the original function signature i.e numpy.ediff1d(ary, to_end=None, to_begin=None) instead of numpy.ediff1d(x, to_end=None, to_begin=None), the test appears to generate values for x only, this is sub-optimal, the test has to generate values for all possible inputs the function takes, i.e you should generate values for to_begin and to_end as well, the tests fails with paddle backend, it has to do with the dtypes (paddle.diff does not support uint8), you might need to add a with_unsupported_dtypes decorator to the backend diff implementation for paddlepaddle. Thanks!

[Aneesh02]

@Infrared1029 Thanks for the review. I have incorporated the original function signature. Can you please help me as to where can I find paddle.diff in the backend files?

[Aneesh02]

@Infrared1029 please help me resolve this error

[Aneesh02]

UPDATE: All the tests are passing for now. Just the test file needs to fixed to incorporate values for to_begin and to_end.

[Infrared1029]

UPDATE: All the tests are passing for now. Just the test file needs to fixed to incorporate values for to_begin and to_end.

let me know once you get that done:)

[Aneesh02]

@Infrared1029 I have fixed the issues. Please have a look. Thanks!

[Aneesh02]

@AnnaTz @Infrared1029 I have tried a lot I cannot find out why the error is coming. Please help me resolve it. The example generated is fine.

[AnnaTz]

@AnnaTz @Infrared1029 I have tried a lot I cannot find out why the error is coming. Please help me resolve it. The example generated is fine.

Just found ediff1 in our jax numpy frontend. You could take a look at how that was implemented and tested. It's practically the same function.

[Aneesh02]

@AnnaTz Finally it worked. Looks like the way I wrote the function was wrong. It was a good learning experience for me. Thank you @AnnaTz @Infrared1029 for reviewing the PR multiple times.

[Aneesh02]

Thanks @AnnaTz @Infrared1029