tqec icon indicating copy to clipboard operation
tqec copied to clipboard

Block Implementation: Task Checklist

Open inmzhang opened this issue 8 months ago • 6 comments

This is a comprehensive checklist of the blocks planned or needed to be implemented in TQEC. Please add in the comments or edit here directly if I missed anything.

Convention Definitions

Fixed-boundary(referred as fixed-parity before ): the weight 2 stabilizers are always in the same place, every second location as you go clockwise around the boundary.

Fixed-bulk: the top left weight 4 stabilizer in each 2kx2k square is always blue (Z).

Fixed-bulk Convention Checklist

  • [x] Regular cubes (ZXCubes except for ZZX and XXZ spatial cubes)
  • [x] Spatial cubes
  • [x] Regular pipes (spatial and temporal pipes without Hadamard transition)
  • [x] Temporal Hadamard pipe
  • [ ] Spatial Hadamard pipe
  • [ ] Y basis initialization and measurement
  • [ ] Walking/sliding code
  • [ ] Patch Rotation(no open issue)
  • [ ] Magic state cultivation(no open issue)

Linked Issues

  • #561
  • #548
  • #491

Linked PRs

  • #558
  • #559
  • #567

Fixed-boundary Convention Checklist

  • [x] Regular cubes (ZXCubes except for ZZX and XXZ spatial cubes)
  • [x] Spatial cubes
  • [x] Regular pipes (spatial and temporal pipes without Hadamard transition)
  • [x] Temporal Hadamard pipe
  • [ ] Spatial Hadamard pipe
  • [ ] Y basis initialization and measurement
  • [ ] Walking/sliding code
  • [ ] Patch Rotation(no open issue)
  • [ ] Magic state cultivation(no open issue)

Linked Issues

  • #631
  • #561
  • #548
  • #491

Linked PRs

  • #562

inmzhang avatar Apr 19 '25 03:04 inmzhang

Hi, I am trying to pick up a new task to work on. It seems that all subtasks are being tracked and taken by someone except for Patch Rotation and Magic state cultivation? Do we have other issues that needs more hands except for these two? If not, do we have a high-level goal/plan for Patch Rotation and Magic state cultivation? Maybe I can do some investigations and open new issues to work on them.

Zhaoyilunnn avatar Jul 20 '25 05:07 Zhaoyilunnn

Hi, I am trying to pick up a new task to work on. It seems that all subtasks are being tracked and taken by someone except for Patch Rotation and Magic state cultivation? Do we have other issues that needs more hands except for these two? If not, do we have a high-level goal/plan for Patch Rotation and Magic state cultivation? Maybe I can do some investigations and open new issues to work on them.

To implement new blocks, these are the only two at the moment. Maybe also ask @KabirDubey if he wants help on sliding / Y-basis if you want. If you are interested in the two issues, exploring them and starting to open granular issues would be perfect yes.

nelimee avatar Jul 20 '25 08:07 nelimee

HI @Zhaoyilunnn, I have started to think about magic state cultivation. I wouldn't mind working on this together, if you are interested.

purva-thakre avatar Jul 20 '25 16:07 purva-thakre

HI @Zhaoyilunnn, I have started to think about magic state cultivation. I wouldn't mind working on this together, if you are interested.

@purva-thakre thanks. Yes I am interested in magic state cultivation. I have looked through the paper and the code but didn't dive deep. While I will likely be working on some other issues before MSC, I will be always available if you would like to discuss or explore any specific part of MSC regarding its principle or implementation.

Zhaoyilunnn avatar Jul 21 '25 01:07 Zhaoyilunnn

Good to know @Zhaoyilunnn!

MSC requires quite a lot of work. Classical feedforward is required, but this functionality is not yet available in Stim.

Quoting Yiming from an email:

For cultivation, the classical feedback is needed for "growing the color code", "discard and retry upon detection events" and "decide whether to keep the state upon the value of complementary gap". But for simulation purposes, I do think these can all be worked around without classical feedback. However, when we want to do the T gate with gate teleportation, fast feedforward operation is not avoidable.

Adrien (during a Wednesday meeting, quoting Austin) said that we might have to define our own language. Although I am not too sure about this, as the functionality in JAX does offer dynamic feedback. Need to explore more. We could start with defining what classical feedforward might look like for tqec? Then, check if we can use JAX for our purposes...

purva-thakre avatar Jul 21 '25 15:07 purva-thakre

Probably worth looking into a new paper by Kwok Ho Wan when we get to the stage of implementing magic state cultivation. https://arxiv.org/abs/2509.01224

purva-thakre avatar Sep 05 '25 02:09 purva-thakre