pattern_fragments.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "%matplotlib inline"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "\n# Optimized pattern fragments\n\nGraphix offers several pattern fragments for quantum gates, pre-optimized by reducing the Pauli measurements,\nwith fewer resource requirement than the standard pattern.\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "First, for Toffoli gate, here is the pattern based on the decomposition of CCX gate with CNOT and single-qubit rotations,\nturned into a measurement pattern:\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "import numpy as np\nfrom graphix import Circuit\n\ncircuit = Circuit(3)\ncircuit.ccx(0, 1, 2)\npattern = circuit.transpile().pattern\npattern.draw_graph(flow_from_pattern=False)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Using :code:`opt=True` option for :code:`transpile` method, we switch to patterns with non-XY plane measurements allowed,\nwhich has gflow (not flow). For CCX gate, the number of ancilla qubits required is nearly halved:\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "pattern = circuit.transpile(opt=True).pattern\npattern.draw_graph(node_distance=(1.2, 0.8))\n# sphinx_gallery_thumbnail_number = 2"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Now let us add z-rotation gates, requiring two ancillas in the original pattern fragment,\nwhich becomes one for patterns with :code:`opt=True`.\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "circuit = Circuit(3)\ncircuit.ccx(0, 1, 2)\nfor i in range(3):\n    circuit.rz(i, np.pi / 4)\npattern = circuit.transpile(opt=True).pattern\npattern.draw_graph(flow_from_pattern=True, node_distance=(1, 0.5))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Swap gate is just a swap of node indices during compilation, requiring no ancillas.\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "circuit = Circuit(3)\ncircuit.ccx(0, 1, 2)\ncircuit.swap(1, 2)\ncircuit.swap(2, 0)\nfor i in range(3):\n    circuit.rz(i, np.pi / 4)\npattern = circuit.transpile(opt=True).pattern\npattern.draw_graph(flow_from_pattern=False, node_distance=(1, 0.4))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "using :code:`opt=True` and with either CCX, Rzz or Rz gates, the graph will have gflow.\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "circuit = Circuit(4)\ncircuit.cnot(1, 2)\ncircuit.cnot(0, 3)\ncircuit.ccx(2, 1, 0)\ncircuit.rx(0, np.pi / 3)\ncircuit.cnot(0, 3)\ncircuit.rzz(0, 3, np.pi / 3)\ncircuit.rx(2, np.pi / 3)\ncircuit.ccx(3, 1, 2)\ncircuit.rx(0, np.pi / 3)\ncircuit.rx(3, np.pi / 3)\npattern = circuit.transpile(opt=True).pattern\npattern.draw_graph(flow_from_pattern=False, node_distance=(1, 0.4))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "reducing the size further\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "pattern.perform_pauli_measurements()\npattern.draw_graph(flow_from_pattern=False, node_distance=(1, 0.6))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "For linear optical QPUs with single photons, such a resource state can be generated by fusing the follwoing microcluster states:\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "from graphix.extraction import get_fusion_network_from_graph\n\nnodes, edges = pattern.get_graph()\nfrom graphix import GraphState\n\ngs = GraphState(nodes=nodes, edges=edges)\nget_fusion_network_from_graph(gs, max_ghz=4, max_lin=4)"
      ]
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.10.13"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}