ml-intern
File size: 4,763 Bytes
1e94df5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
"""
Explicit Formula Engine (from v3, enhanced for v_mix)
Uses the von Mangoldt explicit formula with 100,000 zeros.
"""

import numpy as np
from typing import List, Dict
from dataclasses import dataclass


@dataclass
class PrimePrediction:
    predicted_value: int
    confidence: float
    actual_is_prime: bool = False
    error: int = 0
    method: str = ""


class ExplicitFormulaEngine:
    def __init__(self, zeros: List[float]):
        self.zeros = np.array(zeros, dtype=np.float64)
        self._log = []

    def log(self, msg: str):
        self._log.append(msg)
        print(f"  [ExplicitFormula] {msg}")

    def compute_psi(self, x: float, n_zeros: int = None) -> float:
        if n_zeros is None:
            n_zeros = len(self.zeros)
        n_zeros = min(n_zeros, len(self.zeros))
        if x <= 1:
            return 0.0
        log_x = np.log(x)
        sqrt_x = np.sqrt(x)
        psi = x
        gamma_subset = self.zeros[:n_zeros]
        denom = 0.25 + gamma_subset ** 2
        cos_term = np.cos(gamma_subset * log_x) * 0.5
        sin_term = np.sin(gamma_subset * log_x) * gamma_subset
        psi -= np.sum(2 * sqrt_x * (cos_term + sin_term) / denom)
        psi -= np.log(2 * np.pi)
        if x > 1.01:
            psi -= 0.5 * np.log(1 - 1 / (x * x))
        return psi

    def compute_psi_vectorized(self, x_values: np.ndarray, n_zeros: int = None) -> np.ndarray:
        if n_zeros is None:
            n_zeros = min(10000, len(self.zeros))
        n_zeros = min(n_zeros, len(self.zeros))
        x_values = np.asarray(x_values, dtype=np.float64)
        log_x = np.log(np.maximum(x_values, 1.01))
        sqrt_x = np.sqrt(np.maximum(x_values, 0))
        psi = x_values.copy()
        gamma_arr = self.zeros[:n_zeros]
        for gamma in gamma_arr:
            denom = 0.25 + gamma * gamma
            cos_terms = np.cos(gamma * log_x) * 0.5
            sin_terms = np.sin(gamma * log_x) * gamma
            psi -= 2 * sqrt_x * (cos_terms + sin_terms) / denom
        psi -= np.log(2 * np.pi)
        mask = x_values > 1.01
        psi[mask] -= 0.5 * np.log(1 - 1 / (x_values[mask] ** 2))
        return psi

    def count_predicted_primes(self, x_max: int, n_zeros: int = None) -> int:
        if n_zeros is None:
            n_zeros = min(5000, len(self.zeros))
        x_range = np.arange(2, x_max + 1, dtype=np.float64)
        psi_prev = self.compute_psi_vectorized(x_range - 0.5, n_zeros=n_zeros)
        psi_next = self.compute_psi_vectorized(x_range + 0.5, n_zeros=n_zeros)
        jumps = psi_next - psi_prev
        expected = np.log(x_range)
        predicted = np.sum(jumps > expected * 0.5)
        return int(predicted)

    def minimum_zeros_for_primes(self, x_max: int, max_zeros: int = 100000) -> Dict:
        sieve = np.ones(x_max + 1, dtype=bool)
        sieve[:2] = False
        for i in range(2, int(x_max ** 0.5) + 1):
            if sieve[i]:
                sieve[i * i::i] = False
        actual_primes = np.sum(sieve)

        N_tests = [10, 50, 100, 200, 500, 1000, 2000, 5000,
                     10000, 20000, 50000, 100000]
        N_tests = [n for n in N_tests if n <= max_zeros]

        results = []
        for N in N_tests:
            predicted = self.count_predicted_primes(x_max, n_zeros=N)
            results.append({
                'N': N,
                'predicted_primes': predicted,
                'actual_primes': int(actual_primes),
                'accuracy': predicted / actual_primes if actual_primes > 0 else 0,
                'correct': predicted == actual_primes,
            })
            if predicted == actual_primes:
                break

        return {
            'x_max': x_max,
            'results': results,
            'minimum_N': next((r['N'] for r in results if r['correct']), None),
        }

    def prime_oscillation_contributions(self, x: float, n_zeros: int = 1000) -> Dict:
        n_zeros = min(n_zeros, len(self.zeros))
        log_x = np.log(x)
        sqrt_x = np.sqrt(x)
        contributions = []
        cumulative = 0.0
        for i in range(n_zeros):
            gamma = self.zeros[i]
            denom = 0.25 + gamma * gamma
            cos_term = np.cos(gamma * log_x) * 0.5
            sin_term = np.sin(gamma * log_x) * gamma
            actual = -2 * sqrt_x * (cos_term + sin_term) / denom
            cumulative += actual
            contributions.append({
                'index': i + 1,
                'gamma': gamma,
                'contribution': float(actual),
                'cumulative': float(cumulative),
            })
        return {
            'x': x,
            'total_contribution': float(cumulative),
            'contributions': contributions,
        }

    def get_log(self):
        return self._log