Qualtran v0.7

We're happy to announce the release of Qualtran v0.7 containing a new rotation synthesis module, support for classical data in quantum programs, quantum standard library improvements, and more.

Rotation synthesis

The rotation synthesis module provides a robust, high-performance toolkit for compiling $SU(2)$ unitaries into the Clifford+T gate set. By implementing the current state-of-the-art synthesis protocols from Kliuchnikov et al. (2023), this module allows users to reduce the T-count of quantum algorithms, which is often the primary bottleneck in fault-tolerant quantum computing.

Key Features:

1. Implementation of four key synthesis protocols: diagonal (ancilla-free), mixed diagonal, fallback ("repeat-until-success"), and mixed fallback.
2. Arbitrary precision: the module is backed by mpmath and supports arbitrarily small target diamond norm distances. See the docs for notes on choosing the number of digits of precision.
3. The resulting circuits can be exported as Cirq circuits and Quirk links.

Implemented by @NoureldinYosri in #1749 #1750 #1751.

Classical data and measurement-based uncomputation

Qualtran now supports classical data as part of the compute graph.

• Measurement operations take a QBit to a CBit.
• There are classical versions of each quantum data type, e.g. CInt, CFxp, etc. #1584 #1717
• Programs containing measurement and classical data can be simulated using the new superoperator tensor simulator. #1596
• A new "Phased Classical" simulation protocol supports simulation of circuits consisting of classical-reversible logic (X, CNOT, Toffoli) as well as diagonal gates (T, Z, CZ). Combined with the MeasureX gate, this simulator can be used to fuzz-test measurement-based uncomputation. #1590 #1682 #1698

Breaking changes

• BloqAsCirqGate is now only a cirq.Gate and CirqGateAsBloq is now only a Bloq for conceptual consistency. #1603.

Quantum Standard Library Fixes

We continue to aspire to have a tested, accurate standard library of quantum algorithms. The following changes were made to improve the correctness of quantum algorithms included in qualtran.bloqs.

• Fix bug in classical simulation of Add with mixed signed/unsigned integers by @mpharrigan in #1711
• Fix additional bug in ECAdd by @fpapa250 in #1666
• Consistent Toffoli.controlled() by @mpharrigan in #1771 #1712
• Fix RowColumnOracle Signature by @petim0 in #1780
• Fix Identity's ctrl_adder by @petim0 in #1781
• LinearCombinaison adjoint is more general and checks are correct by @petim0 in #1784
• Override adjoint for ControlledAddOrSubtract by @anurudhp in #1735
• Update GF2Inverse to use the decomposition from "Concrete quantum cryptanalysisof binary elliptic curves" by @NoureldinYosri in #1728
• replace synthesize_reduction_matrix_q with GF2ShiftRight in GF2Multipication by @NoureldinYosri in #1686

Resource estimation changes: The default factor is now four T gates per CSwap #1642. Qubit counting now uses greedy_topological_sort to reduce idle qubits in allocating circuits #1641.

New Bloqs in the Standard Library

We continue to expand the available bloqs in the standard library of quantum algorithms and subroutines.

• [QGFPoly] Add QGFPoly, a new data type to represent polynomials over galois fields by @tanujkhattar in #1620
• [QGFPoly] Add GF2PolyAddK, GFPolySplit and GFPolyJoin bloqs by @tanujkhattar in #1622
• [QGFPoly] Add GF2PolyAdd bloq for quantum-quantum addition of polynomials over GF(2^m) by @tanujkhattar in #1624 #1681 #1705 #1727
• Add GF2MulMBUC Bloq by @tanujkhattar in #1718 #1726
• Implement Kikuchi graph oracles and block-encoding by @anurudhp in #1555
• Add bloq for the Guided Hamiltonian Problem by @anurudhp in #1695
• Add Always metabloq to simplify decompositions with compute-uncompute pairs by @anurudhp in #1605
• Add Split2/Join2 for simpler partitioning by @mpharrigan in #1730

Other improvements

• Support for Numpy 2 #1761
• New reference docs system #1684 and new tutorial notebooks #1694

New Contributors

• @pqvr made their first contribution in #1643
• @ahkatlio made their first contribution in #1651
• @micpap25 made their first contribution in #1652
• @maxglick made their first contribution in #1739
• @amandashoe made their first contribution in #1753
• @petim0 made their first contribution in #1780

Full Changelog: v0.6.0...v0.7.0