eigen-trust 0.1.0

A library for managing trust in a distributed network with zero-knowledge features.
Documentation 
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

'''
Ref: http://ilpubs.stanford.edu:8090/562/1/2002-56.pdf
'''

import secrets
import random

from typing import List, Tuple, NewType, Dict


TMatrix = List[List[float]]
TVector = List[float]


def eigentrust_central(num_peers: int, local_trust_matrix: TMatrix, initial_trust_scores: TVector):
    """
    Algorithm 1 in section 4.3.
    """
    delta = 0.01

    def transpose(t: TMatrix) -> TMatrix:
        return [
            [ t[j][i] for j in range(len(row))]
            for i, row in enumerate(t)
        ]

    def get_matrix_size(a: TMatrix) -> Tuple[int, int]:
        m = len(a)
        if m == 0:
            raise ValueError
        n = len(a[0])
        if n == 0:
            raise ValueError
        return (m, n)

    def matrix_mul(a: TMatrix, b: TMatrix) -> TMatrix:
        assert is_matrix_multipiable(a, b)
        m, na = get_matrix_size(a)
        _, l = get_matrix_size(b)
        return [
            [sum([a[i][k] * b[k][j] for k in range(na)]) for j in range(l)]
            for i in range(m)
        ]

    def is_matrix_multipiable(a: TMatrix, b: TMatrix):
        _, na = get_matrix_size(a)
        nb, _ = get_matrix_size(b)
        return na == nb

    def get_vector_distance(a: TMatrix, b: TMatrix):
        if len(b) != len(a):
            raise ValueError
        s = 0
        for index, row_a in enumerate(a):
            row_b = b[index]
            if len(row_a) != 1 or len(row_b) != 1:
                raise ValueError
            s += pow(row_a[0] - row_b[0], 2)
        return pow(s, 0.5)

    ct = transpose(local_trust_matrix)
    t_matrix: TMatrix = [[i] for i in initial_trust_scores]
    # FIXME: If something goes wrong here we get a infinite loop XD
    while 1:
        new_t = matrix_mul(ct, t_matrix)
        if get_vector_distance(t_matrix, new_t) < delta:
            break
        t_matrix = new_t
    return [row[0] for row in new_t]


def eigentrust_distributed(num_peers: int, local_trust_matrix: TMatrix, initial_trust_scores: TVector):
    """
    A simple version of the algorithm 3, in section 4.6.
    """
    delta = 0.001
    TPeerScore = NewType('TPeerScore', float)
    TPeerIndex = NewType('TPeerIndex', int)

    class Peer:
        index: TPeerIndex
        neighbors: Dict[TPeerIndex, 'Peer']
        local_trust_values: Dict[TPeerIndex, TPeerScore]
        # Global trust value for myself
        ti: TPeerScore
        last_cij_ti: Dict[TPeerIndex, TPeerScore]
        is_converged: bool

        def __init__(self, index: TPeerIndex) -> None:
            self.index = index
            self.neighbors = {}
            self.local_trust_values = {}
            self.last_cij_ti = {}
            self.ti = 0
            self.is_converged = False

        def add_neighbor(self, peer: 'Peer', local_trust_value: TPeerScore, pretrust_score: TPeerScore) -> None:
            self.neighbors[peer.index] = peer
            self.local_trust_values[peer.index] = local_trust_value
            self.last_cij_ti[peer.index] = pretrust_score

        def heartbeat(self) -> None:
            if self.is_converged:
                return

            new_ti = TPeerScore(0.0)
            for j, neighbor_j in self.neighbors.items():
                # Compute `t_i(k+1) = (1 - a)*(c_1i*t_1(k) + c_ji*t_z(k) + ... + c_ni*t_n(k)) + a*p_i`
                # We haven't considered `a` here.
                if self.index not in neighbor_j.last_cij_ti:
                    continue
                new_ti += TPeerScore(neighbor_j.last_cij_ti[self.index])
            # Send c_ij * t_i(k+1)to all peers j
            for j in self.neighbors.keys():
                self.last_cij_ti[j] = TPeerScore(self.local_trust_values[j] * new_ti)
            if abs(new_ti - self.ti) <= delta:
                self.is_converged = True
            self.ti = new_ti

    class Network:
        size: TPeerIndex
        peers: Tuple['Peer', ...]
        is_converged: bool

        def __init__(self, size: TPeerIndex) -> None:
            self.size = size
            self.peers = self._gen_peers(size)
            self.is_converged = False

        def _gen_peers(self, size: TPeerIndex) -> Tuple['Peer', ...]:
            return tuple(
                Peer(TPeerIndex(i)) for i in range(size)
            )

        def _connect_peers(self, local_trust_matrix: TMatrix) -> None:
            for i, c_i in enumerate(local_trust_matrix):
                for j, c_ij in enumerate(c_i):
                    if i == j:
                        continue
                    self.peers[i].add_neighbor(self.peers[j], c_ij, initial_trust_scores[j])

        def tick(self):
            # Randomly choose an order to perform heartbeats
            peer_list = list(self.peers)
            # random.shuffle(peer_list)
            # Become `True` when all peers are converged
            is_not_all_converged = True
            for peer in peer_list:
                peer.heartbeat()
                is_not_all_converged = is_not_all_converged and peer.is_converged
            self.is_converged = is_not_all_converged

        def get_global_trust_scores(self) -> TVector:
            t = sum([peer.ti for peer in self.peers])
            return [peer.ti / t for peer in self.peers]

    network = Network(num_peers)
    network._connect_peers(local_trust_matrix)

    while not network.is_converged:
        network.tick()

    return network.get_global_trust_scores()


def gen_trust_matrix(num_peers: int) -> TMatrix:
    # i, j = 0
    # sum(row) = 1
    # For each row, get `num_peers - 1` values, and normalize them.
    mat = []
    for i in range(num_peers):
        random_ints = [secrets.randbelow(32) for _ in range(num_peers - 1)]
        s = sum(random_ints)
        normalized = [j / s for j in random_ints]
        normalized.insert(i, 0)
        mat.append(normalized)
    return mat


def print_matrix(c: TMatrix):
    print('=' * (len(c) * 20))
    for row in c:
        print(',\t'.join([str(i) for i in row]))
    print('=' * (len(c) * 20))


def main():
    num_peers_str = input('Enter number of peers\n')
    num_peers = int(num_peers_str)
    initial_trust_scores = [1/num_peers for i in range(num_peers)]
    print('initial_trust_scores = ', initial_trust_scores)
    # c = gen_trust_matrix(num_peers)
    c = [
        [0.,	0.,	0.9333333333333333,	0.06666666666666667],
        [0.4727272727272727,	0.,	0.03636363636363636,	0.4909090909090909],
        [0.3695652173913043,	0.3695652173913043,	0.,	0.2608695652173913],
        [0.15625,	0.4375,	0.40625,	0.],
    ]
    print('number of peers = ', num_peers)
    print('local reputation matrix is generated randomly:')
    print_matrix(c)

    res_central = eigentrust_central(num_peers, c, initial_trust_scores)
    res_distributed = eigentrust_distributed(num_peers, c, initial_trust_scores)

    print(f'res_central\t= {res_central}')
    print(f'res_distributed\t= {res_distributed}')

main()