backend

This module contains the Backend class, which processes superconducting chip information into an undirected graph representation. It also supports the creation of custom undirected graphs to serve as virtual chips.

Classes:

Name	Description
Backend	A class to represent a quantum hardware backend as a nx.Graph.

Backend(chip: Literal['Baihua', 'Custom'] | dict)

A class to represent a quantum hardware backend as a nx.Graph.

Initialize a Backend object.

Parameters:

Name	Type	Description	Default
chip	str	Chip name, currently only 'Baihua' is avaliable.	required

Methods:

Name	Description
edge_filtered_graph	Create a subgraph by filtering out edges with fidelity below a specified threshold.
get_graph	Constructs and returns an undirected graph with nodes and weighted edges.
draw	Draw the chip layout.

Attributes:

Name	Type	Description
graph		Returns the graph representation of the object.

Source code in quark/circuit/backend.py

def __init__(self,chip: Literal['Baihua','Custom'] | dict):
    """Initialize a Backend object.

    Args:
        chip (str): Chip name, currently only 'Baihua' is avaliable.
    """
    if isinstance(chip,dict):
        self.chip_name = ' '
        self.chip_info = chip
        self.size = self.chip_info['size']
        self.priority_qubits = self.chip_info['priority_qubits']
        self.qubits_with_attributes = self._collect_qubits_with_attributes()
        self.couplers_with_attributes = self._collect_couplers_with_attributes()
        self.two_qubit_gate_basis = self.chip_info['global_info']['two_qubit_gate_basis'].lower()
    elif chip in ['Baihua','Dongling','Haituo','Yunmeng','Miaofeng','Yudu','Hongluo',]:
        self.chip_name = chip
        try:
            self.chip_info = load_chip_basic_info(chip)
        except:
            raise(ValueError(f'{chip} is under maintenance, configuration information is unavailable'))
        print('The last calibration time was',self.chip_info['calibration_time'])
        self.size = self.chip_info['size']
        self.priority_qubits = self.chip_info['priority_qubits']
        self.qubits_with_attributes = self._collect_qubits_with_attributes()
        self.couplers_with_attributes = self._collect_couplers_with_attributes()
        self.two_qubit_gate_basis = self.chip_info['global_info']['two_qubit_gate_basis'].lower()

    elif chip == 'Custom':
        self.chip_name = chip
        self.chip_info = dict()
        self.size = (0,0)
        self.qubits_with_attributes = list()
        self.couplers_with_attributes = list()
        self.priority_qubits = []
        self.two_qubit_gate_basis = 'cz'
    else:
        raise(ValueError(f'Wrong chip name! {chip}'))

graph property

Returns the graph representation of the object.

This property method calls self.get_graph() to generate and return the graph with nodes and edges.

Returns:

Type	Description
	networkx.Graph: The graph with nodes and weighted edges.

edge_filtered_graph(thres=0.6)

Create a subgraph by filtering out edges with fidelity below a specified threshold.

Parameters:

Name	Type	Description	Default
thres	float	The fidelity threshold. Defaults to 0.6.	0.6

Returns:

Type	Description
	networkx.Graph: A new NetworkX graph object containing only edges with fidelity greater than or equal to the threshold.

Source code in quark/circuit/backend.py

def edge_filtered_graph(self,thres = 0.6):
    """Create a subgraph by filtering out edges with fidelity below a specified threshold.

    Args:
        thres (float, optional): The fidelity threshold. Defaults to 0.6.

    Returns:
        networkx.Graph: A new NetworkX graph object containing only edges with fidelity greater than or equal to the threshold.
    """
    def edge_filter(u,v):
        return self.graph[u][v].get("fidelity") >= thres
    def node_filter(n):
        return self.graph.nodes[n].get("fidelity") >= thres
    subgraph_view = nx.subgraph_view(self.graph,filter_node=node_filter,filter_edge=edge_filter)
    return nx.Graph(subgraph_view)

get_graph()

Constructs and returns an undirected graph with nodes and weighted edges.

Returns:

Type	Description
	networkx.Graph: An undirected graph with nodes and weighted edges.

Source code in quark/circuit/backend.py

def get_graph(self):
    """Constructs and returns an undirected graph with nodes and weighted edges.

    Returns:
        networkx.Graph: An undirected graph with nodes and weighted edges.
    """
    G = nx.Graph()
    G.add_nodes_from(self.qubits_with_attributes)
    G.add_edges_from(self.couplers_with_attributes)
    return G

draw(show_couplers_fidelity: bool = False, show_qubits_attributes: Literal['T1', 'T2', 'fidelity', 'frequency', ''] = '', show_qubits_index: bool = False, show_couplers_index: bool = False, highlight_nodes: list = [], save_svg_fname: str | None = None, edge_fidelity_thres=0.0)

Draw the chip layout.

Parameters:

Name	Type	Description	Default
show_couplers_fidelity	bool	Whether to display the fidelity of couplers. Defaults to False.	False
show_qubits_attributes	Literal['T1', 'T2', 'fidelity', 'frequancy', '']	Specify which qubit attribute to display. Options include: - 'T1': Display the T1 time of qubits (in microseconds). - 'T2': Display the T2 time of qubits (in microseconds). - 'fidelity': Display the fidelity of qubits. - 'frequancy': Display the frequency of qubits (in GHz). - '': Display no attributes. Defaults to ''.	''
show_qubits_index	bool	Whether to display index for each qubit. Defaults to False.	False
show_couplers_index	bool	Whether to display index for each coupler. Defaults to False.	False
highlight_nodes	list	A list of qubits to highlight. Defaults to an empty list [].	[]
save_svg_fname	str | None	The filename for saving the drawing as an SVG. If None, the drawing will not be saved. The user is responsible for ensuring the validity of the provided file path. Defaults to None.	None

Returns:

Name	Type	Description
None		This function does not return a value but generates and optionally saves the chip layout.

Notes

• highlight_nodes specifies the qubits to be visually highlighted.
• show_couplers_fidelity and show_qubits_attributes can be enabled simultaneously without conflicts.
• When save_svg_fname is provided, the drawing will be saved as an SVG file.

Source code in quark/circuit/backend.py

def draw(self,show_couplers_fidelity:bool = False, show_qubits_attributes:Literal['T1','T2','fidelity','frequency','']='', 
         show_qubits_index:bool=False,show_couplers_index:bool=False,
         highlight_nodes:list = [], save_svg_fname: str|None = None,edge_fidelity_thres=0.0):
    """Draw the chip layout.

    Args:
        show_couplers_fidelity (bool, optional): Whether to display the fidelity of couplers. Defaults to False.
        show_qubits_attributes (Literal['T1', 'T2', 'fidelity', 'frequancy', ''], optional): 
            Specify which qubit attribute to display. Options include:
            - 'T1': Display the T1 time of qubits (in microseconds).
            - 'T2': Display the T2 time of qubits (in microseconds).
            - 'fidelity': Display the fidelity of qubits.
            - 'frequancy': Display the frequency of qubits (in GHz).
            - '': Display no attributes.
            Defaults to ''.
        show_qubits_index (bool, optional): Whether to display index for each qubit. Defaults to False.
        show_couplers_index (bool, optional): Whether to display index for each coupler. Defaults to False.
        highlight_nodes (list, optional): A list of qubits to highlight. Defaults to an empty list [].
        save_svg_fname (str | None, optional): 
            The filename for saving the drawing as an SVG. If None, the drawing will not be saved. 
            The user is responsible for ensuring the validity of the provided file path. Defaults to None.

    Returns:
        None: This function does not return a value but generates and optionally saves the chip layout.

    Notes:
        - `highlight_nodes` specifies the qubits to be visually highlighted.
        - `show_couplers_fidelity` and `show_qubits_attributes` can be enabled simultaneously without conflicts.
        - When `save_svg_fname` is provided, the drawing will be saved as an SVG file.
    """
    import matplotlib.pyplot as plt 
    from matplotlib.colors import Normalize,LinearSegmentedColormap
    from matplotlib.cm import ScalarMappable 

    graph_show = self.edge_filtered_graph(thres=edge_fidelity_thres)
    edge_fidelity = nx.get_edge_attributes(graph_show, 'fidelity') 
    node_attributes = nx.get_node_attributes(graph_show,show_qubits_attributes)
    if show_couplers_fidelity:
        if set(list(edge_fidelity.values())) == {0}:
            print('The two-qubit gate fidelity is N/A now.')
            is_edge_info_avaliable = False
        else:
            is_edge_info_avaliable = True
    if show_qubits_attributes:
        if set(list(node_attributes.values())) == {0}:
            print('The qubit attributes is N/A now.')
            is_node_info_avaliable = False
        else:
            is_node_info_avaliable = True            

    if show_couplers_fidelity is True and is_edge_info_avaliable is True:
        min_fidelity = sorted(list(edge_fidelity.values()))[0]
        max_fidelity = sorted(list(edge_fidelity.values()))[-1]
        edge_norm = Normalize(vmin = min_fidelity, vmax = max_fidelity)
        edge_cmap = LinearSegmentedColormap.from_list('truncated_blues', plt.get_cmap('Blues')(np.linspace(0.31, 1.0, 1000)))  #plt.get_cmap('Blues')
        edge_colors = [ScalarMappable(norm = edge_norm, cmap = edge_cmap).to_rgba(fidelity) for fidelity in edge_fidelity.values()]
    else:
        edge_colors = ['#083776' for edge in graph_show.edges()]

    edge_labels = {}
    for k,v in edge_fidelity.items():
        if v == None:
            fidelity = ''
        else:
            fidelity = np.round(v, 3)
        if show_couplers_fidelity is True and is_edge_info_avaliable is True:
            if show_couplers_index:
                edge_labels[k] = graph_show.edges[k].get('index')
            else:
                edge_labels[k] = fidelity
        else:
            edge_labels[k] = graph_show.edges[k].get('index')

    if show_qubits_attributes != '' and is_node_info_avaliable is True:
        min_attributes = sorted(list(node_attributes.values()))[0]
        max_attributes = sorted(list(node_attributes.values()))[-1]
        node_norm = Normalize(vmin = min_attributes, vmax = max_attributes)
        node_cmap = LinearSegmentedColormap.from_list('truncated_blues', plt.get_cmap('Blues')(np.linspace(0.31, 1.0, 1000))) #plt.get_cmap('Blues') 
        node_colors = [ScalarMappable(norm = node_norm, cmap = node_cmap).to_rgba(attribute) for attribute in node_attributes.values()]
        if show_qubits_attributes == 'fidelity':
            if show_qubits_index:
                node_labels = {node:node for node in graph_show.nodes()} 
            else:
                node_labels =  {node: np.round(attr, 3) for node, attr in node_attributes.items()} # 保留三位有效数字
        else:
            if show_qubits_index:
                node_labels = {node:node for node in graph_show.nodes()} 
            else:    
                node_labels =  {node: np.round(attr, 2) for node, attr in node_attributes.items()}  # 保留两位有效数字
        node_font_size = 8 
        if show_couplers_fidelity is True and is_edge_info_avaliable is True:
            figsize = (15, 15)
        else:
            figsize = (15, 13.5)
    else:
        node_colors = ['#083776' for node in graph_show.nodes()]
        node_labels = {node:node for node in graph_show.nodes()}   
        node_font_size = 10 
        if show_couplers_fidelity is True and is_edge_info_avaliable is True:
            figsize = (15, 13.5)
        else:
            figsize = (15, 13)

    if len(highlight_nodes) > 0:
        if isinstance(highlight_nodes, list) and any(isinstance(i, (list,tuple)) for i in highlight_nodes):
            highlight_nodes = list(set(e for sub in highlight_nodes for e in sub))
        node_colors_update = []
        node_labels_update = {}
        for idx, node in enumerate(graph_show.nodes()):
            if node in highlight_nodes:
                node_colors_update.append(node_colors[idx])
                node_labels_update[node] = node_labels[node]
            else:
                node_colors_update.append('#f3f3f3')
                node_labels_update[node] = ''
        node_colors = node_colors_update
        node_labels = node_labels_update

        subgraph = graph_show.subgraph(highlight_nodes)
        edge_colors_update = []
        edge_labels_update = {}
        for idx, edge in enumerate(graph_show.edges()):
            if edge in subgraph.edges():
                edge_colors_update.append(edge_colors[idx])
                edge_labels_update[edge] = edge_labels[edge]
            else:
                edge_colors_update.append('#f3f3f3')
                edge_labels_update[edge] = ''
        edge_colors = edge_colors_update
        edge_labels = edge_labels_update

    fig, ax = plt.subplots(figsize=figsize) #facecolor='none'
    pos = nx.get_node_attributes(graph_show, 'coordinate') 
    nx.draw(graph_show, pos,ax=ax, with_labels=False, node_color=node_colors, node_size=800, edgecolors='white',edge_color = edge_colors ,width = 18,) 
    nx.draw_networkx_labels(graph_show,pos,labels=node_labels,font_size=node_font_size, font_color='white')
    nx.draw_networkx_edge_labels(graph_show, pos, edge_labels=edge_labels,font_size=8, font_color='white',bbox=dict(facecolor='none', edgecolor='none'),rotate=False)

    if show_qubits_attributes != '' and is_node_info_avaliable is True:
        if show_qubits_attributes == 'T1':
            show_label = r'T1 ($\mu s$)'
        elif show_qubits_attributes == 'T2':
            show_label = r'T2 ($\mu s$)'
        elif show_qubits_attributes == 'fidelity':
            show_label = 'Single qubit gate fidelity'
        elif show_qubits_attributes == 'frequency':
            show_label = 'Frequency (GHz)'
        node_sm = ScalarMappable(cmap=node_cmap, norm=node_norm)
        node_sm.set_array(list(node_attributes.values()))  # 设置包含数据的数组
        if show_couplers_fidelity and is_edge_info_avaliable is True:
            node_cbar = fig.colorbar(node_sm, ax=ax, orientation='horizontal',pad=0.07, fraction=0.03, aspect=25)
        else:
            node_cbar = fig.colorbar(node_sm, ax=ax, orientation='horizontal',pad=0.001, fraction=0.0333, aspect=25)
        node_cbar.set_label(show_label)

    if show_couplers_fidelity is True and is_edge_info_avaliable is True:
        edge_sm = ScalarMappable(cmap=edge_cmap, norm=edge_norm)# 创建颜色条
        edge_sm.set_array(list(edge_fidelity.values()))  # 设置包含数据的数组
        edge_cbar = fig.colorbar(edge_sm, ax=ax, orientation='horizontal',  pad=0.001, fraction=0.0333, aspect=25)# 调整颜色条大小和位置，放置在底部
        edge_cbar.set_label('CZ fidelity')

    xlim = ax.get_xlim() 
    ylim = ax.get_ylim()
    xpos = (xlim[0] + xlim[1]) / 2
    ypos = ylim[0] - (ylim[1] - ylim[0]) * -0.03  # 计算标题位置
    ax.text(xpos, ypos, f'{self.chip_name}', va='center', ha='center', fontsize=24, fontweight='bold', color='k', family='serif')

    ax.invert_yaxis() # y 轴正半轴朝下

    if save_svg_fname:
        plt.savefig(save_svg_fname + '.svg', transparent=True, dpi=300, bbox_inches='tight') 
    plt.show()
    return None