量子计算学习路径规划

In [4]:

import deepquantum as dq
import deepquantum.photonic as dqp
import torch
import torch.nn as nn

import deepquantum as dq import deepquantum.photonic as dqp import torch import torch.nn as nn

In [1]:

import deepquantum as dq
cir = dq.QubitCircuit(2)
cir.h(0)
cir.cnot(0, 1)
cir.rx(1, 0.2)
cir.observable(0)
print(cir())
print(cir.expectation())

import deepquantum as dq cir = dq.QubitCircuit(2) cir.h(0) cir.cnot(0, 1) cir.rx(1, 0.2) cir.observable(0) print(cir()) print(cir.expectation())

tensor([[0.7036+0.0000j],
        [0.0000-0.0706j],
        [0.0000-0.0706j],
        [0.7036+0.0000j]])
tensor([0.])

In [2]:

cir = dq.QubitCircuit(2, mps=True, chi=4)
cir.h(0)
cir.cnot(0, 1)
cir.rx(1, 0.2)
cir.observable(0)
print(cir())
print(cir.expectation())

cir = dq.QubitCircuit(2, mps=True, chi=4) cir.h(0) cir.cnot(0, 1) cir.rx(1, 0.2) cir.observable(0) print(cir()) print(cir.expectation())

[tensor([[[1.+0.j, 0.+0.j],
         [0.-0.j, 1.+0.j]]]), tensor([[[0.7036+0.0000j],
         [0.0000-0.0706j]],

        [[0.0000-0.0706j],
         [0.7036+0.0000j]]])]
tensor([2.6822e-07])

In [3]:

cir = dq.QumodeCircuit(2, [1,1])
cir.dc([0,1])
cir.ps(0, 0.1)
cir.bs([0,1], [0.2,0.3])
print(cir())
print(cir.measure())

cir = dq.QumodeCircuit(2, [1,1]) cir.dc([0,1]) cir.ps(0, 0.1) cir.bs([0,1], [0.2,0.3]) print(cir()) print(cir.measure())

tensor([[ 0.6480-0.0650j, -0.2034+0.7311j],
        [ 0.1294+0.7477j,  0.6383+0.1294j]])
{|20>: 487, |02>: 516, |11>: 21}

In [5]:

class Hybrid(nn.Module):
    def __init__(self, dim_in, nqubit):
        super().__init__()
        self.fc1 = nn.Linear(dim_in, nqubit)
        # 构建好的线路本身就是nn.Module
        self.cir = self.circuit(nqubit)
        self.fc2 = nn.Linear(nqubit, 1)

    def circuit(self, nqubit):
        cir = dq.QubitCircuit(nqubit)
        cir.hlayer()
        # 准备将经典数据编码到量子线路中
        cir.rylayer(encode=True)
        cir.rxlayer()
        cir.cnot_ring()
        for i in range(nqubit):
            cir.observable(i)
        return cir

    def forward(self, x):
        x = torch.arctan(self.fc1(x))
        # 前向计算的第一个参数对应于要编码的数据
        self.cir(x)
        exp = self.cir.expectation()
        out = self.fc2(exp)
        return out

nqubit = 4
batch = 2
nfeat = 8
x = torch.sin(torch.tensor(list(range(batch * nfeat)))).reshape(batch, nfeat)
net = Hybrid(nfeat, nqubit)
y = net(x)
print(y)
for i in net.named_parameters():
    print(i)

class Hybrid(nn.Module): def __init__(self, dim_in, nqubit): super().__init__() self.fc1 = nn.Linear(dim_in, nqubit) # 构建好的线路本身就是nn.Module self.cir = self.circuit(nqubit) self.fc2 = nn.Linear(nqubit, 1) def circuit(self, nqubit): cir = dq.QubitCircuit(nqubit) cir.hlayer() # 准备将经典数据编码到量子线路中 cir.rylayer(encode=True) cir.rxlayer() cir.cnot_ring() for i in range(nqubit): cir.observable(i) return cir def forward(self, x): x = torch.arctan(self.fc1(x)) # 前向计算的第一个参数对应于要编码的数据 self.cir(x) exp = self.cir.expectation() out = self.fc2(exp) return out nqubit = 4 batch = 2 nfeat = 8 x = torch.sin(torch.tensor(list(range(batch * nfeat)))).reshape(batch, nfeat) net = Hybrid(nfeat, nqubit) y = net(x) print(y) for i in net.named_parameters(): print(i)

tensor([[-0.4744],
        [-0.4817]], grad_fn=<AddmmBackward0>)
('fc1.weight', Parameter containing:
tensor([[-0.0689,  0.2601,  0.3227,  0.3401,  0.0606,  0.2179,  0.3464,  0.2867],
        [ 0.2584, -0.0296,  0.0859, -0.1323, -0.3210,  0.1817,  0.3025, -0.2655],
        [-0.0068, -0.1469, -0.0710, -0.1834,  0.2303,  0.2150,  0.0892, -0.1840],
        [ 0.3451,  0.3216, -0.1844, -0.2062, -0.0359, -0.2329,  0.3050,  0.0050]],
       requires_grad=True))
('fc1.bias', Parameter containing:
tensor([-0.2667,  0.1115, -0.1669, -0.2236], requires_grad=True))
('cir.operators.8.theta', Parameter containing:
tensor(2.5965, requires_grad=True))
('cir.operators.9.theta', Parameter containing:
tensor(2.3473, requires_grad=True))
('cir.operators.10.theta', Parameter containing:
tensor(10.3887, requires_grad=True))
('cir.operators.11.theta', Parameter containing:
tensor(5.4013, requires_grad=True))
('fc2.weight', Parameter containing:
tensor([[0.2741, 0.1773, 0.2788, 0.1564]], requires_grad=True))
('fc2.bias', Parameter containing:
tensor([-0.4741], requires_grad=True))

In [ ]: