Clarify the Conventions in Pulser

[HGSilveri]

• Adds a "Conventions" page to the docs, establishing all the conventions in Pulser
• Modifies the simulation in XY mode to comply with this convention

Closes #571 .

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[a-corni]

Great documentation, as far as I am concerned it is very clear and it will for sure help users :) Just found a minor typo and have two suggestions

[madagra]

@HGSilveri I have only a couple of small comments but it all looks good to me:

• I would add a note about the endianness to which the chosen basis state ordering corresponds when expressing bitstrings. I think it is little endian, if I am not mistaken. The "Measurements samples ordering" is probably the right place for it.
• In the XY basis, however, the state ordering looks flipped to me, now you get big endian when you sample. I guess the explanation given to @a-corni is valid also in this case but I want to point out this since it can be pretty dangerous in some cases.

[HGSilveri]

@HGSilveri I have only a couple of small comments but it all looks good to me:

• I would add a note about the endianness to which the chosen basis state ordering corresponds when expressing bitstrings. I think it is little endian, if I am not mistaken. The "Measurements samples ordering" is probably the right place for it.
• In the XY basis, however, the state ordering looks flipped to me, now you get big endian when you sample. I guess the explanation given to @a-corni is valid also in this case but I want to point out this since it can be pretty dangerous in some cases.

Thanks for the review @madagra ! I'm not sure I understand your comments. From my understanding, endianness refers to how a binary number is stored in a register and displayed, so it is only relevant when you are reading your samples as a binary number. I think it's preferable to leave this up to the user and stick solely to matching the samples to the order of the qubit IDs. Let me give you an example:

Take a register with qubit IDs ("fish", "cat", "dog"). The state vector is going to be |fish, cat, dog> and let's say that upon measurement in the ground-rydberg basis it is projected onto |rgg>. Then, we will register a count in sample 100. That's all. The user can then decide whether "dog" or "fish" is the smallest-memory site in their Register and this will determine whether they read the sample as "4" or "1".

Also, can you point me to where ordering is flipped in the XY basis? This shouldn't be the case.

[lvignoli]

I had a quick read without checking the validity of the maths.
As a user that comes back to these computations regularly, I would say most of the useful information is there. Nice job!

[lvignoli]

I in fact had this small comment

[a-corni]

Nit: There is a reference to conventions page in the notebook about SLM

according to Pulser's conventions for XY basis states

Perhaps we could link both pages ?

[a-corni]

This looks good to me !

[HGSilveri]

This looks good to me !

I was still waiting for a reply from @madagra, but I guess I can merge and then make modifications later if necessary