New multiqubit gate

[GiggleLiu]

julia> c1 = chain(5, [ContinuousBlockMarker(put(5, (2,3)=>SWAP), "SWAP"), 
  ContinuousBlockMarker(put(5, 2=>I2), "id"), put(5, 2=>ContinuousBlockMarker(X, "X")),
  control(5, (5,3), (2,4,1)=>ContinuousBlockMarker(put(3, (1,3)=>SWAP), "SWAP"))])

julia> vizcircuit(c1)

On the other side, if you try to visualize the following gate, you will get an error.

    c2 = chain(5, [ContinuousBlockMarker(put(5, (2,3)=>SWAP), "SWAP"), ContinuousBlockMarker(put(5, 2=>I2), "id"), put(5, 2=>ContinuousBlockMarker(X, "X")), control(5, (5,3), (2,4,1)=>ContinuousBlockMarker(put(3, (1,2)=>SWAP), "SWAP"))])

[VarLad]

Great！I hope this gets simplified in the future！Thank you taking the time @GiggleLiu 😄

[GiggleLiu]

Yeah, but we need a proposal about how to make it simpler. I do not want to insert a name field to the ChainBlock. If you have any interesting idea, feel free to open an issue.

[Roger-luo]

Can you explain why we need an extra block node? It seems to me a method like nameof is sufficient for determining the name of a block.

[VarLad]

@Roger-luo Sounds great to me！🙂

[GiggleLiu]

Can you explain why we need an extra block node? It seems to me a method like nameof is sufficient for determining the name of a block.

But the name is not static.

[Roger-luo]

I think there at least should be a default fallback name for primitive gates, or why ContinuousBlockMarker(put(5, 2=>I2), "id") is not displayed if the user specified it manually? I don't think we should hide any block that user has specified explicitly. we can let ContinuousBlockMarker to change the default name, so the code will look less noisy, e.g

c1 = chain(5, [ContinuousBlockMarker(put(5, (2,3)=>SWAP), "my custom name for SWAP"), 
  ContinuousBlockMarker(put(5, 2=>I2), "explicit id"), put(5, 2=>X),
  control(5, (5,3), put(3, (1,3)=>SWAP))])

where the X gate will be using a default name X, and the block I2 should be displayed explicitly since the user has specified a name for it explicitly (or why should we allow the user to do so?)

on the other hand, what does the word Continuous stands for in ContinuousBlockMarker? can we make it shorter to be just BlockName or something like that?

[GiggleLiu]

I2 is not shown because we use occpied_locs does not count it. Maybe instead of

ContinuousBlockMarker(put(5, (2,3)=>SWAP))

We should use

put(5, (2,3)=>put(5, (2,3)=>LabeledBlock(SWAP, "SWAP"))

[VarLad]

Whats the problem with,

​put​(​5​, (​2​,​3​)​=>​LabeledBlock​(SWAP, ​"​SWAP​"​))

[VarLad]

Also, if you could shorten it or make a "lazy" version of LabeledBlock

​put​(​5​, (​2​,​3​)​=>​label(SWAP, ​"​SWAP​"​))

This looks perfect！
Can this be done？

[GiggleLiu]

Whats the problem with,

​put​(​5​, (​2​,​3​)​=>​LabeledBlock​(SWAP, ​"​SWAP​"​))

It can not handle the following gate

ContinuousBlockMarker(chain(5, put(5, 2=>X), put(5, 3=>X)), "XX")

Instead, one has to change the circuit to

chain(5, put(5, (2, 3)=>label(kron(X, X), "XX")))

It is not a big issue, the later looks more intuitive.

[VarLad]

Instead, one has to change the circuit to

chain(5, put(5, (2, 3)=>label(kron(X, X), "XX")))

It is not a big issue, the later looks more intuitive.

What about

​circuit = chain​(​5​, ​ put​(​2​=>​X), ​put​(​3​=>​X))
XX = chain(5, put(1:5 => label(circuit, "XX"))

[VarLad]

Just one more thing！
If a given DaggeredBlock has a name/label assigned to it, can you plot it as name/label† in a rectangular box？
This might be the best solution for plotting Daggered Block, other than '

[GiggleLiu]

You mean this?

chain(5, [put(5, (2,3)=>label(SWAP, "SWAP")'), put(5, 2=>label(I2, "id")), put(5, 2=>label(X, "X")), control(5, (5,3), (2,4,1)=>put(3, (1,3)=>label(SWAP, "SWAP")))]) |> vizcircuit

[GiggleLiu]

@Roger-luo good now?

[Roger-luo]

LGTM

[GiggleLiu]

We can also use ishermitian to detect whether the content in a labeled block is hermitian, so that we can save a dagger in this case (because the SWAP gate is hermitian).

Which is better?

[VarLad]

Great！Sounds good to me！
Thanks a lot for taking the time again😄

[GiggleLiu]

Great！Sounds good to me！
Thanks a lot for taking the time again

Thank you too for providing practical using cases to guide the design of visualization.

@Roger-luo , I think this PR is ready to merge.

[Roger-luo]

I think we should just follow whatever user is written - we shouldn't perform optimization at this level.