You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It is important that we can successfully run tket1 on arbitrary Hugrs, so that we can at least rebase.
For graphs with control flow, this means running tket1 on each of many non-intersecting regions.
An admissible region is
A set of input wires
A convex sibling subgraph of a dataflow parent such that all incoming wires to the subgraph are in the set of input wires
all ops in the subgraph are admissible ops
Certainly all quantum ops are admissible ops, one might also choose to include float and/or rotation and/or logic ops.
An isolated quantum op is itself an admissible region, so we can find a minimal(in number, i.e. merge any two mergable regions) set of admissible regions covering every quantum op.
Now rewrite each admissible region into a DFG with no non-local edges. By the definition of admissible ops, only qubits, bools, floats, rotations can be input to the DFG.
Run tket1 on the DFG, mapping non-quantum inputs to symbols.
Note that rewriting into a DFG is not strictly required, we can run on an admissible region in-place. I suggest that rewriting into a DFG will be simpler and more tractable to understand going forward.
It is important that we can successfully run tket1 on arbitrary Hugrs, so that we can at least rebase.
For graphs with control flow, this means running tket1 on each of many non-intersecting regions.
An admissible region is
Certainly all quantum ops are admissible ops, one might also choose to include float and/or rotation and/or logic ops.
An isolated quantum op is itself an admissible region, so we can find a minimal(in number, i.e. merge any two mergable regions) set of admissible regions covering every quantum op.
Now rewrite each admissible region into a DFG with no non-local edges. By the definition of admissible ops, only qubits, bools, floats, rotations can be input to the DFG.
Run tket1 on the DFG, mapping non-quantum inputs to symbols.
Note that rewriting into a DFG is not strictly required, we can run on an admissible region in-place. I suggest that rewriting into a DFG will be simpler and more tractable to understand going forward.
See #633
The text was updated successfully, but these errors were encountered: