tensorcircuit-nightly 1.0.2.dev20250108__py3-none-any.whl → 1.4.0.dev20251103__py3-none-any.whl

tests/test_results.py

@@ -1,347 +0,0 @@
-import pytest
-import numpy as np
-
-
-import tensorcircuit as tc
-from tensorcircuit.results import counts
-from tensorcircuit.results.readout_mitigation import ReadoutMit
-
-d = {"000": 2, "101": 3, "100": 4}
-
-
-def test_marginal_count():
-    assert counts.marginal_count(d, [1, 2])["00"] == 6
-    assert counts.marginal_count(d, [1])["0"] == 9
-    assert counts.marginal_count(d, [2, 1, 0])["001"] == 4
-
-
-def test_count2vec():
-    assert counts.vec2count(counts.count2vec(d, normalization=False), prune=True) == d
-
-
-def test_kl():
-    a = {"00": 512, "11": 512}
-    assert counts.kl_divergence(a, a) == 0
-
-
-def test_expectation():
-    assert counts.expectation(d, [0, 1]) == -5 / 9
-    assert counts.expectation(d, None, [[1, -1], [1, 0], [1, 1]]) == -5 / 9
-
-
-def test_plot_histogram():
-    d = {"00": 10, "01": 2, "11": 8}
-    d1 = {"00": 11, "11": 9}
-    print(counts.plot_histogram([d, d1]))
-
-
-def test_readout():
-    nqubit = 4
-    shots = 4096
-    c = tc.Circuit(nqubit)
-    c.H(0)
-    c.cnot(0, 1)
-    c.x(3)
-
-    idea_count = c.sample(batch=shots, allow_state=True, format="count_dict_bin")
-    raw_count = run([c], shots)[0]
-
-    # test "inverse",  "constrained_least_square", "M3"
-    mit = ReadoutMit(execute=run)
-    mit.cals_from_system([0, 1, 2, 3, 6], shots=10000, method="local")
-
-    mit_count1 = mit.apply_correction(
-        raw_count, [1, 3, 2], method="inverse"
-    )  #  direct(Max2),iterative(Max3), inverse,square
-    mit_count2 = mit.apply_correction(
-        raw_count, [1, 3, 2], method="constrained_least_square"
-    )
-    idea_count2 = counts.marginal_count(idea_count, [1, 3, 2])
-
-    assert counts.kl_divergence(idea_count2, mit_count1) < 0.05
-    assert counts.kl_divergence(idea_count2, mit_count2) < 0.05
-
-    # test "global" and "equal"
-    mit = ReadoutMit(execute=run)
-    mit.cals_from_system([0, 1, 2, 3], shots=100000, method="global")
-    A_global = mit.get_matrix([1, 3, 2])
-    mit_countg = mit.apply_correction(
-        raw_count, [1, 3, 2], method="constrained_least_square"
-    )
-
-    mit = ReadoutMit(execute=run)
-    mit.cals_from_system([0, 1, 2, 3], shots=100000, method="local")
-    A_local = mit.get_matrix([1, 3, 2])
-    mit_countl = mit.apply_correction(
-        raw_count, [1, 3, 2], method="constrained_least_square"
-    )
-
-    np.testing.assert_allclose(A_global, A_local, atol=1e-2)
-    assert counts.kl_divergence(mit_countg, mit_countl) < 0.05
-
-
-def test_readout_masks():
-    mit = ReadoutMit(execute=run)
-    mit.cals_from_system(
-        [1, 2, 4], shots=8192, method="local", masks=["01010", "10101", "11111"]
-    )
-    np.testing.assert_allclose(
-        mit.single_qubit_cals[1][0, 0], 0.02 * np.sin(2) + 0.978, atol=1e-2
-    )
-
-
-def test_readout_expv():
-    nqubit = 4
-    c = tc.Circuit(nqubit)
-    c.H(0)
-    c.cnot(0, 1)
-    c.x(3)
-
-    idea_count = c.sample(batch=100000, allow_state=True, format="count_dict_bin")
-    raw_count = run([c], 100000)[0]
-
-    cal_qubits = [0, 1, 2, 3]
-    use_qubits = [0, 1]
-
-    # idea_value = c.expectation_ps(z=[0,1])
-    idea_count2 = counts.marginal_count(idea_count, use_qubits)
-    idea_value = counts.expectation(idea_count2, z=[0, 1])
-
-    mit = ReadoutMit(execute=run)
-    mit.cals_from_system(cal_qubits, shots=100000, method="local")
-    mit_count = mit.apply_correction(raw_count, use_qubits, method="inverse")
-    mit_value = counts.expectation(mit_count, z=[0, 1])
-
-    mit = ReadoutMit(execute=run)
-    mit.cals_from_system(cal_qubits, shots=100000, method="local")
-    mit_value1 = mit.expectation(raw_count, z=[0, 1], method="inverse")
-
-    mit = ReadoutMit(execute=run)
-    mit.cals_from_system(cal_qubits, shots=100000, method="global")
-    mit_value2 = mit.expectation(raw_count, z=[0, 1], method="square")
-
-    np.testing.assert_allclose(idea_value, mit_value, atol=1e-2)
-    np.testing.assert_allclose(idea_value, mit_value1, atol=1e-2)
-    np.testing.assert_allclose(idea_value, mit_value2, atol=1e-2)
-
-    # test large size
-    nqubit = 20
-    c = tc.Circuit(nqubit)
-    c.H(0)
-    for i in range(nqubit - 1):
-        c.cnot(i, i + 1)
-    c.rx(1, theta=0.9)
-
-    idea_count = c.sample(batch=100000, allow_state=True, format="count_dict_bin")
-    raw_count = run([c], 100000)[0]
-
-    cal_qubits = list(range(nqubit))
-    use_qubits = list(range(nqubit))
-
-    # idea_value = c.expectation_ps(z=[0,1])
-    idea_count2 = counts.marginal_count(idea_count, use_qubits)
-    idea_value = counts.expectation(idea_count2, z=list(range(nqubit)))
-
-    mit = ReadoutMit(execute=run)
-    mit.cals_from_system(cal_qubits, shots=100000, method="local")
-    mit_value1 = mit.expectation(raw_count, z=list(range(nqubit)), method="inverse")
-
-    np.testing.assert_allclose(idea_value, mit_value1, atol=1e-1)
-
-
-def test_M3():
-    try:
-        import mthree  # pylint: disable=unused-import
-    except ImportError:
-        pytest.skip("****** No mthree, skipping test suit *******")
-
-    nqubit = 20
-    c = tc.Circuit(nqubit)
-    c.H(0)
-    for i in range(nqubit - 1):
-        c.cnot(i, i + 1)
-    c.rx(1, theta=0.9)
-
-    idea_count = c.sample(batch=100000, allow_state=True, format="count_dict_bin")
-    raw_count = run([c], 100000)[0]
-
-    cal_qubits = list(range(nqubit))
-    use_qubits = list(range(nqubit))
-
-    idea_count2 = counts.marginal_count(idea_count, use_qubits)
-    idea_value = counts.expectation(idea_count2, z=list(range(nqubit)))
-
-    mit = ReadoutMit(execute=run)
-    mit.cals_from_system(cal_qubits, shots=100000, method="local")
-    mit_count = mit.apply_correction(raw_count, use_qubits, method="M3_auto")
-    mit_value = counts.expectation(mit_count, z=list(range(nqubit)))
-    np.testing.assert_allclose(idea_value, mit_value, atol=1e-1)
-
-    nqubit = 4
-    shots = 4096
-    c = tc.Circuit(nqubit)
-    c.H(0)
-    c.cnot(0, 1)
-    c.x(3)
-
-    idea_count = c.sample(batch=shots, allow_state=True, format="count_dict_bin")
-    raw_count = run([c], shots)[0]
-
-    mit_count3 = mit.apply_correction(raw_count, [1, 3, 2], method="M3_direct")
-    mit_count4 = mit.apply_correction(raw_count, [1, 3, 2], method="M3_iterative")
-    idea_count2 = counts.marginal_count(idea_count, [1, 3, 2])
-    assert counts.kl_divergence(idea_count2, mit_count3) < 0.05
-    assert counts.kl_divergence(idea_count2, mit_count4) < 0.05
-
-
-def partial_sample(c, batch, readout_error=None):
-    measure_index = []
-    for inst in c._extra_qir:
-        if inst["name"] == "measure":
-            measure_index.append(inst["index"][0])
-    if len(measure_index) == 0:
-        measure_index = list(range(c._nqubits))
-
-    ct = c.sample(
-        allow_state=True,
-        batch=batch,
-        readout_error=readout_error,
-        format="count_dict_bin",
-    )
-    return tc.results.counts.marginal_count(ct, measure_index)
-
-
-def run(cs, shots):
-    # customized backend for mitigation test
-    ts = []
-    for c in cs:
-        count = simulator(c, shots)
-        ts.append(count)
-    return ts
-
-
-def simulator(c, shots, logical_physical_mapping=None):
-    # with readout_error noise
-    nqubit = c._nqubits
-    if logical_physical_mapping is None:
-        logical_physical_mapping = {i: i for i in range(nqubit)}
-
-    gg = []
-    for i in range(200):
-        gg.append(np.sin(i) * 0.02 + 0.978)
-        # gg.append(0.98 - i * 0.01)
-    readout_error = np.reshape(gg[0 : nqubit * 2], (nqubit, 2))
-    mapped_readout_error = [[1, 1]] * nqubit
-    for lq, phyq in logical_physical_mapping.items():
-        mapped_readout_error[lq] = readout_error[phyq]
-    return partial_sample(c, shots, mapped_readout_error)
-
-
-def test_mapping():
-    nqubit = 15
-    shots = 100000
-    c = tc.Circuit(nqubit)
-    c.H(4)
-    c.cnot(4, 5)
-    c.cnot(5, 6)
-    c.cnot(6, 7)
-    c.rx(4, theta=0.8)
-    c.rx(7, theta=1.8)
-    c.measure_instruction(4)
-    c.measure_instruction(5)
-    c.measure_instruction(6)
-    c.measure_instruction(7)
-
-    mit = ReadoutMit(execute=run)
-    mit.cals_from_system(list(range(15)), shots=100000, method="local")
-
-    show_qubits = [6, 7, 5]
-
-    idea_count = c.sample(batch=shots, allow_state=True, format="count_dict_bin")
-    idea_count1 = counts.marginal_count(idea_count, show_qubits)
-
-    def miti_kl_mean(logical_physical_mapping):
-        ls = []
-        for _ in range(10):
-            raw_count = simulator(c, shots, logical_physical_mapping)
-            mit_count1 = mit.apply_correction(
-                raw_count,
-                qubits=show_qubits,
-                positional_logical_mapping={1: 5, 0: 4, 2: 6, 3: 7},
-                logical_physical_mapping=logical_physical_mapping,
-                method="square",
-            )
-            ls.append(counts.kl_divergence(idea_count1, mit_count1))
-        # print("std", np.std(listtt), np.mean(listtt)) # smaller error rate and larger shots, better mititation.
-        np.testing.assert_allclose(np.mean(ls), 0.01, atol=1e-2)
-
-    logical_physical_mapping = {4: 0, 6: 2, 7: 3, 5: 1}
-    miti_kl_mean(logical_physical_mapping)
-
-    logical_physical_mapping = {4: 4, 5: 5, 6: 6, 7: 7}
-    miti_kl_mean(logical_physical_mapping)
-
-    logical_physical_mapping = {4: 8, 5: 9, 6: 10, 7: 11}
-    miti_kl_mean(logical_physical_mapping)
-
-
-def test_readout_expv_map():
-    shots = 100000
-    nqubit = 7
-    c = tc.Circuit(nqubit)
-    c.H(3)
-    c.cnot(3, 4)
-    c.cnot(4, 5)
-    c.rx(3, theta=0.8)
-    c.rx(4, theta=1.2)
-    c.measure_instruction(3)
-    c.measure_instruction(4)
-    c.measure_instruction(5)
-
-    idea_count = c.sample(batch=100000, allow_state=True, format="count_dict_bin")
-    idea_value = counts.expectation(idea_count, z=[4, 5])
-
-    # logical_physical_mapping = {3: 3, 4: 4, 5: 5}
-    logical_physical_mapping = {3: 1, 5: 3, 4: 6}
-    positional_logical_mapping = {1: 4, 0: 3, 2: 5}
-
-    raw_count = simulator(c, shots, logical_physical_mapping)
-
-    cal_qubits = list(range(nqubit))
-
-    mit = ReadoutMit(execute=run)
-    mit.cals_from_system(cal_qubits, shots=100000, method="local")
-    mit_value1 = mit.expectation(
-        raw_count,
-        z=[4, 5],
-        positional_logical_mapping=positional_logical_mapping,
-        logical_physical_mapping=logical_physical_mapping,
-        method="inverse",
-    )
-
-    mit_count = mit.apply_correction(
-        raw_count,
-        qubits=[3, 4, 5],
-        positional_logical_mapping=positional_logical_mapping,
-        logical_physical_mapping=logical_physical_mapping,
-        method="inverse",
-    )
-    mit_value = counts.expectation(mit_count, z=[1, 2])
-    # print("idea", idea_value)
-    # print("mit", mit_value)
-
-    mit = ReadoutMit(execute=run)
-    mit.cals_from_system(cal_qubits, shots=100000, method="global")
-    mit_value2 = mit.expectation(
-        raw_count,
-        z=[4, 5],
-        positional_logical_mapping=positional_logical_mapping,
-        logical_physical_mapping=logical_physical_mapping,
-        method="inverse",
-    )
-
-    # print("mit1", mit_value1)
-    # print("mit2", mit_value2)
-    np.testing.assert_allclose(idea_value, mit_value, atol=3 * 1e-2)
-    np.testing.assert_allclose(idea_value, mit_value1, atol=3 * 1e-2)
-    np.testing.assert_allclose(idea_value, mit_value2, atol=3 * 1e-2)

tests/test_shadows.py

@@ -1,160 +0,0 @@
-import pytest
-from pytest_lazyfixture import lazy_fixture as lf
-import numpy as np
-import tensorcircuit as tc
-from tensorcircuit.shadows import (
-    shadow_bound,
-    shadow_snapshots,
-    global_shadow_state,
-    entropy_shadow,
-    renyi_entropy_2,
-    expectation_ps_shadow,
-    global_shadow_state1,
-)
-
-
-@pytest.mark.parametrize("backend", [lf("tfb"), lf("jaxb")])
-def test_jit(backend):
-    nq, repeat = 8, 5
-    ps = [1, 0, 0, 0, 2, 0, 0, 0]
-    sub = (1, 3, 6, 7)
-    error = 0.1
-    ns, k = shadow_bound(ps, error)
-    ns //= repeat
-
-    thetas = 2 * np.random.rand(2, nq) - 1
-
-    c = tc.Circuit(nq)
-    for i in range(nq):
-        c.H(i)
-    for i in range(2):
-        for j in range(nq):
-            c.cnot(j, (j + 1) % nq)
-        for j in range(nq):
-            c.rz(j, theta=thetas[i, j] * np.pi)
-
-    psi = c.state()
-    pauli_strings = tc.backend.convert_to_tensor(np.random.randint(1, 4, size=(ns, nq)))
-    status = tc.backend.convert_to_tensor(np.random.rand(ns, repeat))
-
-    def classical_shadow(psi, pauli_strings, status):
-        lss_states = shadow_snapshots(psi, pauli_strings, status)
-        expc = expectation_ps_shadow(lss_states, ps=ps, k=k)
-        ent = entropy_shadow(lss_states, sub=sub, alpha=2)
-        return expc, ent
-
-    csjit = tc.backend.jit(classical_shadow)
-
-    exact_expc = c.expectation_ps(ps=ps)
-    exact_rdm = tc.quantum.reduced_density_matrix(psi, cut=[0, 2, 4, 5])
-    exact_ent = tc.quantum.renyi_entropy(exact_rdm, k=2)
-    expc, ent = csjit(psi, pauli_strings, status)
-    expc = np.median(expc)
-
-    np.testing.assert_allclose(expc, exact_expc, atol=error)
-    np.testing.assert_allclose(ent, exact_ent, atol=5 * error)
-
-
-@pytest.mark.parametrize("backend", [lf("tfb"), lf("jaxb")])
-def test_state(backend):
-    nq, ns = 2, 10000
-
-    c = tc.Circuit(nq)
-    c.H(0)
-    c.cnot(0, 1)
-
-    psi = c.state()
-    bell_state = psi[:, None] @ psi[None, :]
-
-    pauli_strings = tc.backend.convert_to_tensor(np.random.randint(1, 4, size=(ns, nq)))
-    status = tc.backend.convert_to_tensor(np.random.rand(ns, 5))
-    lss_states = shadow_snapshots(c.state(), pauli_strings, status)
-    sdw_state = global_shadow_state(lss_states)
-    sdw_state1 = global_shadow_state1(lss_states)
-
-    np.testing.assert_allclose(sdw_state, bell_state, atol=0.1)
-    np.testing.assert_allclose(sdw_state1, bell_state, atol=0.1)
-
-
-@pytest.mark.parametrize("backend", [lf("tfb"), lf("jaxb")])
-def test_ent(backend):
-    nq, ns, repeat = 6, 1000, 500
-
-    thetas = 2 * np.random.rand(2, nq) - 1
-
-    c = tc.Circuit(nq)
-    for i in range(nq):
-        c.H(i)
-    for i in range(2):
-        for j in range(nq):
-            c.cnot(j, (j + 1) % nq)
-        for j in range(nq):
-            c.rz(j, theta=thetas[i, j] * np.pi)
-
-    sub = [1, 4]
-    psi = c.state()
-
-    pauli_strings = tc.backend.convert_to_tensor(np.random.randint(1, 4, size=(ns, nq)))
-    status = tc.backend.convert_to_tensor(np.random.rand(ns, repeat))
-    snapshots = shadow_snapshots(psi, pauli_strings, status, measurement_only=True)
-
-    exact_rdm = tc.quantum.reduced_density_matrix(
-        psi, cut=[i for i in range(nq) if i not in sub]
-    )
-    exact_ent = tc.quantum.renyi_entropy(exact_rdm, k=2)
-    ent = entropy_shadow(snapshots, pauli_strings, sub, alpha=2)
-    ent2 = renyi_entropy_2(snapshots, sub)
-
-    np.testing.assert_allclose(ent, exact_ent, atol=0.1)
-    np.testing.assert_allclose(ent2, exact_ent, atol=0.1)
-
-
-# @pytest.mark.parametrize("backend", [lf("tfb"), lf("jaxb")])
-# def test_expc(backend):
-#     import pennylane as qml
-#
-#     nq, ns, repeat = 6, 2000, 1000
-#
-#     thetas = 2 * np.random.rand(2, nq) - 1
-#
-#     c = tc.Circuit(nq)
-#     for i in range(nq):
-#         c.H(i)
-#     for i in range(2):
-#         for j in range(nq):
-#             c.cnot(j, (j + 1) % nq)
-#         for j in range(nq):
-#             c.rz(j, theta=thetas[i, j] * np.pi)
-#
-#     ps = [1, 0, 0, 0, 0, 3]
-#     sub = [1, 4]
-#     psi = c.state()
-#
-#     pauli_strings = tc.backend.convert_to_tensor(np.random.randint(1, 4, size=(ns, nq)))
-#     status = tc.backend.convert_to_tensor(np.random.rand(ns, repeat))
-#     snapshots = shadow_snapshots(psi, pauli_strings, status, measurement_only=True)
-#
-#     exact_expc = c.expectation_ps(ps=ps)
-#     exact_rdm = tc.quantum.reduced_density_matrix(
-#         psi, cut=[i for i in range(nq) if i not in sub]
-#     )
-#     exact_ent = tc.quantum.renyi_entropy(exact_rdm, k=2)
-#     print(exact_expc, exact_ent)
-#
-#     expc = np.median(expection_ps_shadow(snapshots, pauli_strings, ps=ps, k=9))
-#     ent = entropy_shadow(snapshots, pauli_strings, sub, alpha=2)
-#     ent2 = renyi_entropy_2(snapshots, sub)
-#     print(expc, ent, ent2)
-#
-#     pl_snapshots = np.asarray(snapshots).reshape(ns * repeat, nq)
-#     pl_ps = np.tile(np.asarray(pauli_strings - 1)[:, None, :], (1, repeat, 1)).reshape(
-#         ns * repeat, nq
-#     )
-#     shadow = qml.ClassicalShadow(pl_snapshots, pl_ps)
-#     H = qml.PauliX(0) @ qml.PauliZ(5)
-#     pl_expc = shadow.expval(H, k=9)
-#     pl_ent = shadow.entropy(sub, alpha=2)
-#     print(pl_expc, pl_ent)
-#
-#     assert np.isclose(expc, pl_expc)
-#     assert np.isclose(ent, pl_ent)

tests/test_simplify.py

@@ -1,46 +0,0 @@
-import os
-import sys
-
-thisfile = os.path.abspath(__file__)
-modulepath = os.path.dirname(os.path.dirname(thisfile))
-
-sys.path.insert(0, modulepath)
-import numpy as np
-import tensornetwork as tn
-from tensorcircuit import simplify
-
-
-def test_infer_shape():
-    a = tn.Node(np.ones([2, 3, 5]))
-    b = tn.Node(np.ones([3, 5, 7]))
-    a[1] ^ b[0]
-    a[2] ^ b[1]
-    assert simplify.infer_new_shape(a, b) == ((2, 7), (2, 3, 5), (3, 5, 7))
-
-
-def test_rank_simplify():
-    a = tn.Node(np.ones([2, 2]), name="a")
-    b = tn.Node(np.ones([2, 2]), name="b")
-    c = tn.Node(np.ones([2, 2, 2, 2]), name="c")
-    d = tn.Node(np.ones([2, 2, 2, 2, 2, 2]), name="d")
-    e = tn.Node(np.ones([2, 2]), name="e")
-
-    a[1] ^ c[0]
-    b[1] ^ c[1]
-    c[2] ^ d[0]
-    c[3] ^ d[1]
-    d[4] ^ e[0]
-
-    nodes = simplify._full_rank_simplify([a, b, c, d, e])
-    assert nodes[0].shape == tuple([2 for _ in range(6)])
-    assert len(nodes) == 1
-
-    f = tn.Node(np.ones([2, 2]), name="f")
-    g = tn.Node(np.ones([2, 2, 2, 2]), name="g")
-    h = tn.Node(np.ones([2, 2, 2, 2]), name="h")
-
-    f[1] ^ g[0]
-    g[2] ^ h[1]
-
-    nodes = simplify._full_rank_simplify([f, g, h])
-    assert len(nodes) == 2