Fix measure with condition display in mpl and latex drawers

[enavarro51]

Summary

Fixes #6804

Details and comments

This PR fixes a problem where a measure with condition will sometimes display incorrectly in the 'mpl' and 'latex' drawers. The corrected displays are below.

qr = QuantumRegister(2, "qr")
cr1 = ClassicalRegister(2, "cr1")
cr2 = ClassicalRegister(2, "cr2")
qc = QuantumCircuit(qr, cr1, cr2)
qc.h(0)
qc.h(1)
qc.measure(0, cr1[0])
qc.measure(1, cr2[0]).c_if(cr1, 1)
qc.draw('mpl')

[TharrmashasthaPV]

Hi @enavarro51 . Nice one! I think supporting measure with condition in latex drawer has some issues. For instance, when cregbundle=False, then the same circuit given above returns as below:

You can notice here that it is hard to distinguish between the conditional bits and the bit on which the measurement is stored. I guess we would be able to overcome this issue if we had different notations for the conditional bits and the storage bit. I understand that it is a completely different issue that what this PR is for. But any opinions on that?

[enavarro51]

Good point. It would be nice if there were some kind of arrow so it mimics the mpl drawer. Not sure how that works in Latex.

[enavarro51]

@TharrmashasthaPV How does this look? Not perfect, but I think it solves your issue.

[TharrmashasthaPV]

@TharrmashasthaPV How does this look? Not perfect, but I think it solves your issue.

@enavarro51 This certainly is better than having just dots. But I was thinking we could do something like

I guess replacing \\cwx[#1] with \\ar @{<=} [#1,0] should do the trick if I am not wrong. Also in #6447 we were trying to get a minimal latex source code with minimal headers. So, I guess this way we can also be in line with its goal.😃 What do you think?

[enavarro51]

@TharrmashasthaPV Very nice! Thought you might have a trick to use. I'll drop it in. Seems to work in all the cases I've tried. I added it to the cregbundle True as well. Is there a way to add a bullet in addition to the '= 1' at the condition creg?

[TharrmashasthaPV]

@TharrmashasthaPV Very nice! Thought you might have a trick to use. I'll drop it in. Seems to work in all the cases I've tried. I added it to the cregbundle True as well. Is there a way to add a bullet in addition to the '= 1' at the condition creg?

Awesome. For having both the bullet and =1, I think currently we use \dstick{_{_{\hspace{1.5em}=1}}} \cw \cwx[-1] which just gives =1. But alternatively we can use something like \control \cw^(-0.25){^{=1}} \cwx[-1] to have both. The parameter that has been set to -0.25 can be adjusted as per the need I guess.

[enavarro51]

Thanks a lot, @TharrmashasthaPV. Sometimes it's great to have someone else write your code for you. 😊 Anyway, I've extended this so the mpl and latex drawers should now show the same information for measures and conditionals. I'm not sure why mpl uses hex for the condition and the other 2 drawers don't. Maybe a discussion for another time. You may need to incorporate these changes in #6248 and #6259 at some point.

[javabster]

Not a great reason. After a discussion with Luciano, I added the line onto the previous empty plot to make it more realistic. Not too important either way.

Ok, but how is it relevant to the work in this PR?

[enavarro51]

Uh, not really. Would you like it removed?

[javabster]

Also, could you add a release note please 😄

[javabster]

Uh, not really. Would you like it removed?

Ah ok. Generally it's preferable to keep the content of PRs as relevant as possible to the original issue, and open new issues to cover extra things, but in this case it's quite small and not worth opening a whole new PR for it (and there is already scope creep with the switching to hex conditionals as well anyway) so I think just leave it

[enavarro51]

Oh, 'scope creep'. I like that. 😊 Will do the release note.

[enavarro51]

Release note added in 056eb8b.

[javabster]

This lgtm! Thanks for working on this, just a few merge conflicts left and I think it'll be good to go

[enavarro51]

@javabster Merge conflicts fixed in 3bd0c3e.

[javabster]

LGTM 🚀

[TharrmashasthaPV]

The fix looks great! I was just curious to know if we could make a small aesthetic fix. In line 627, when we print the hex, it actually gets printed as a string inside a math environment. So it looks like

If we could just change this line to

f"{control}" + " \\cw^(%s){^{\\mathtt{%s}}} \\cwx[-%s]"

then we can get something like

which I personally think looks more like a hexadecimal. Any thoughts on this?

[enavarro51]

Agreed. That x was a bit too fancy, and the new form is more compact as well. Will make the change.

[TharrmashasthaPV]

Perfect. LGTM.

[1ucian0]

Thanks!