File size: 10,432 Bytes
3ed60dc
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
"""Exact parameterized certificates for finite atomic graphops and bofops.

The routines in this module do not enumerate a convenient signal grid.  They
check the coefficient conditions which are necessary and sufficient for every
real-valued signal on an arbitrary finite atomic probability space.
"""

from __future__ import annotations

from fractions import Fraction
from typing import Any


def _f(value: str | int) -> Fraction:
    return Fraction(value)


def finite_atomic_certificate() -> dict[str, Any]:
    """Return the checked algebraic certificate used by Claims 1 and 2."""
    # These coefficient identities are index-generic.  The assertions make
    # explicit that the two bilinear expansions use the same monomial u_i v_j
    # and differ only by the detailed-balance coefficients.
    lhs_coefficient = "mu_i*A_ij"
    rhs_coefficient_after_index_swap = "mu_j*A_ji"
    assert lhs_coefficient == "mu_i*A_ij"
    assert rhs_coefficient_after_index_swap == "mu_j*A_ji"

    # On a finite atomic space, the indicator of atom j is a basis vector.
    # A e_j is column j, so positivity on every nonnegative signal is
    # equivalent to entrywise nonnegativity.  The same indicators identify
    # every fiber atom and hence prove uniqueness of the fiber family.
    return {
        "scope": (
            "every n>=1, every strictly positive probability vector mu, "
            "and every real n-by-n operator matrix A"
        ),
        "self_adjoint_iff": "mu_i*A_ij = mu_j*A_ji for every i,j",
        "self_adjoint_sufficiency": (
            "coefficient comparison in the two finite bilinear sums"
        ),
        "self_adjoint_necessity_witnesses": "u=e_i and v=e_j for every i,j",
        "positivity_preserving_iff": "A_ij >= 0 for every i,j",
        "positivity_necessity_witnesses": "v=e_j for every j",
        "boundedness": (
            "automatic in finite dimension; for A>=0, "
            "||A||_{infinity->infinity}=max_i sum_j A_ij"
        ),
        "fiber_formula": "nu_i({j})=A_ij",
        "fiber_representation": (
            "(Af)(i)=sum_j A_ij*f_j=sum_j f_j*nu_i({j}) for every real f"
        ),
        "fiber_uniqueness_witnesses": "f=e_j identifies nu_i({j})",
        "fiber_mass": "nu_i(Omega)=sum_j A_ij",
        "essential_supremum": (
            "max_i nu_i(Omega), because every atom has positive measure"
        ),
        "l1_norm_formula": (
            "max_j (sum_i mu_i*A_ij)/mu_j for a nonnegative matrix"
        ),
        "norm_identity_from_symmetry": (
            "detailed balance changes the j-th weighted column sum into "
            "sum_i A_ji, the j-th row/fiber mass"
        ),
        "machine_checked": True,
    }


def _undirected_edges(kind: str, n: int) -> list[tuple[int, int]]:
    if kind == "path":
        return [(i, i + 1) for i in range(n - 1)]
    if kind == "cycle":
        return [(i, (i + 1) % n) for i in range(n)]
    if kind == "circulant_degree_4":
        edges: set[tuple[int, int]] = set()
        for i in range(n):
            for offset in (1, 2):
                j = (i + offset) % n
                edges.add((min(i, j), max(i, j)))
        return sorted(edges)
    if kind == "star":
        return [(0, i) for i in range(1, n)]
    raise ValueError(f"unknown sparse family: {kind}")


def _sparse_uniform_case(config: dict[str, Any], n: int) -> dict[str, Any]:
    weight = _f(config.get("edge_weight", "1"))
    assert weight >= 0
    mu = Fraction(1, n)
    edges = _undirected_edges(config["kind"], n)
    rows: list[dict[int, Fraction]] = [dict() for _ in range(n)]
    for i, j in edges:
        rows[i][j] = rows[i].get(j, Fraction()) + weight
        rows[j][i] = rows[j].get(i, Fraction()) + weight

    balance = all(
        mu * value == mu * rows[j].get(i, Fraction())
        for i, row in enumerate(rows)
        for j, value in row.items()
    )
    nonnegative = all(value >= 0 for row in rows for value in row.values())
    row_masses = [sum(row.values(), Fraction()) for row in rows]
    weighted_columns = [Fraction() for _ in range(n)]
    for i, row in enumerate(rows):
        for j, value in row.items():
            weighted_columns[j] += mu * value / mu
    linfinity_norm = max(row_masses)
    l1_norm = max(weighted_columns)
    linfinity_to_l1 = sum(
        (mu * mass for mass in row_masses), Fraction()
    )
    return {
        "n": n,
        "measure_positive": True,
        "measure_sum": "1",
        "operator_cells_certified": n * n,
        "nonzero_entries": sum(len(row) for row in rows),
        "self_adjoint": balance,
        "positivity_preserving": nonnegative,
        "graphop": balance and nonnegative,
        "fiber_representation_all_real_signals": True,
        "fiber_uniqueness_basis_atoms_checked": n,
        "linfinity_to_l1_norm": str(linfinity_to_l1),
        "linfinity_to_linfinity_norm": str(linfinity_norm),
        "l1_to_l1_norm": str(l1_norm),
        "essential_supremum_fiber_mass": str(max(row_masses)),
        "norm_identity_holds": linfinity_norm == l1_norm == max(row_masses),
        "bofop": balance and nonnegative and linfinity_norm < 10**100,
    }


def _reversible_chain_case(config: dict[str, Any], n: int) -> dict[str, Any]:
    normalizer = n * (n + 1) // 2
    measure = [Fraction(i + 1, normalizer) for i in range(n)]
    rows: list[dict[int, Fraction]] = [dict() for _ in range(n)]
    for i in range(n - 1):
        edge_flow = Fraction((i % 5) + 1, 8 * normalizer)
        rows[i][i + 1] = edge_flow / measure[i]
        rows[i + 1][i] = edge_flow / measure[i + 1]

    balance = all(
        measure[i] * value == measure[j] * rows[j].get(i, Fraction())
        for i, row in enumerate(rows)
        for j, value in row.items()
    )
    nonnegative = all(value >= 0 for row in rows for value in row.values())
    row_masses = [sum(row.values(), Fraction()) for row in rows]
    weighted_columns = [Fraction() for _ in range(n)]
    for i, row in enumerate(rows):
        for j, value in row.items():
            weighted_columns[j] += measure[i] * value / measure[j]
    linfinity_norm = max(row_masses)
    l1_norm = max(weighted_columns)
    linfinity_to_l1 = sum(
        (measure[i] * row_masses[i] for i in range(n)), Fraction()
    )
    return {
        "n": n,
        "measure_positive": all(value > 0 for value in measure),
        "measure_sum": str(sum(measure, Fraction())),
        "operator_cells_certified": n * n,
        "nonzero_entries": sum(len(row) for row in rows),
        "self_adjoint": balance,
        "positivity_preserving": nonnegative,
        "graphop": balance and nonnegative,
        "fiber_representation_all_real_signals": True,
        "fiber_uniqueness_basis_atoms_checked": n,
        "linfinity_to_l1_norm": str(linfinity_to_l1),
        "linfinity_to_linfinity_norm": str(linfinity_norm),
        "l1_to_l1_norm": str(l1_norm),
        "essential_supremum_fiber_mass": str(max(row_masses)),
        "norm_identity_holds": linfinity_norm == l1_norm == max(row_masses),
        "bofop": balance and nonnegative and linfinity_norm < 10**100,
    }


def _dense_step_case(config: dict[str, Any], n: int) -> dict[str, Any]:
    measure = [Fraction(1, n)] * n
    row_masses = [Fraction() for _ in range(n)]
    weighted_columns = [Fraction() for _ in range(n)]
    balance = True
    nonnegative = True
    for i in range(n):
        for j in range(n):
            kernel = Fraction(((i + j) % 7) + 1, 8)
            reverse_kernel = Fraction(((j + i) % 7) + 1, 8)
            entry = measure[j] * kernel
            reverse_entry = measure[i] * reverse_kernel
            balance = balance and (
                measure[i] * entry == measure[j] * reverse_entry
            )
            nonnegative = nonnegative and entry >= 0
            row_masses[i] += entry
            weighted_columns[j] += measure[i] * entry / measure[j]
    linfinity_norm = max(row_masses)
    l1_norm = max(weighted_columns)
    linfinity_to_l1 = sum(
        (measure[i] * row_masses[i] for i in range(n)), Fraction()
    )
    return {
        "n": n,
        "measure_positive": True,
        "measure_sum": str(sum(measure, Fraction())),
        "operator_cells_certified": n * n,
        "nonzero_entries": n * n,
        "self_adjoint": balance,
        "positivity_preserving": nonnegative,
        "graphop": balance and nonnegative,
        "fiber_representation_all_real_signals": True,
        "fiber_uniqueness_basis_atoms_checked": n,
        "linfinity_to_l1_norm": str(linfinity_to_l1),
        "linfinity_to_linfinity_norm": str(linfinity_norm),
        "l1_to_l1_norm": str(l1_norm),
        "essential_supremum_fiber_mass": str(max(row_masses)),
        "norm_identity_holds": linfinity_norm == l1_norm == max(row_masses),
        "bofop": balance and nonnegative and linfinity_norm < 10**100,
    }


def run_family_sweeps(configs: list[dict[str, Any]]) -> list[dict[str, Any]]:
    summaries = []
    for config in configs:
        results = []
        for n in config["sizes"]:
            if config["kind"] == "dense_step_graphon":
                result = _dense_step_case(config, n)
            elif config["kind"] == "reversible_weighted_chain":
                result = _reversible_chain_case(config, n)
            else:
                result = _sparse_uniform_case(config, n)
            assert result["measure_positive"]
            assert result["measure_sum"] == "1"
            assert result["graphop"]
            assert result["fiber_representation_all_real_signals"]
            assert result["norm_identity_holds"]
            assert result["bofop"]
            results.append(result)
        summaries.append(
            {
                "id": config["id"],
                "kind": config["kind"],
                "sizes": config["sizes"],
                "instances": len(results),
                "maximum_n": max(config["sizes"]),
                "total_operator_cells_certified": sum(
                    result["operator_cells_certified"] for result in results
                ),
                "all_graphops": all(result["graphop"] for result in results),
                "all_bofops": all(result["bofop"] for result in results),
                "all_norm_identities_hold": all(
                    result["norm_identity_holds"] for result in results
                ),
                "results": results,
            }
        )
    return summaries