浮点范围搜索异常：已知ratio与X组合无法被Python代码捕获

核心问题 #

问题根源在于浮点数累加的精度误差和低效的双重循环逻辑：

• 浮点数（如float类型）累加0.000001时，多次循环后会出现微小偏差，导致X或ratio无法精确命中目标值21.278393和2.252316。
• 双重循环遍历所有X和ratio组合，不仅效率极低，还放大了浮点误差的影响。

修正方案 #

方案1：用整数计数替代浮点数累加 #

通过整数步数计算实际的X和ratio值，避免浮点数累加带来的误差：

from tqdm import tqdm

start_ratio = 2.199164
end_ratio = 2.277869
target = 47.92567055
start_X = 21.072100
end_X = 21.373130
step = 0.000001
EPSILON = 1e-6 

found = []

# 计算总步数，用整数避免浮点误差
total_X_steps = int((end_X - start_X) / step) + 1

for i in tqdm(range(total_X_steps), desc="Scanning X values"):
    current_X = start_X + i * step
    # 直接计算所需的ratio，避免遍历所有ratio
    required_ratio = target / current_X
    # 检查ratio是否在范围内，且乘积误差符合要求
    if (start_ratio <= required_ratio <= end_ratio) and abs(required_ratio * current_X - target) < EPSILON:
        # 保留6位小数，匹配目标值的精度
        rounded_ratio = round(required_ratio, 6)
        rounded_X = round(current_X, 6)
        print(f"[FOUND] ratio={rounded_ratio:.6f}, X={rounded_X:.6f}, value={rounded_ratio*rounded_X:.6f}")
        with open("./data/phase_2.txt", "a") as file:
            file.write(f"[FOUND] ratio={rounded_ratio:.6f}, X={rounded_X:.6f}, value={rounded_ratio*rounded_X:.6f}\n")
        found.append((rounded_ratio, rounded_X, rounded_ratio*rounded_X))

print(f"Matches found: {len(found)}")

方案2：使用Decimal确保高精度计算 #

如果必须用Decimal，需确保所有数值、步长和运算都基于Decimal类型，避免混合float：

from tqdm import tqdm
from decimal import Decimal, getcontext

# 设置足够高的精度
getcontext().prec = 12

start_ratio = Decimal('2.199164')
end_ratio = Decimal('2.277869')
target = Decimal('47.92567055')
start_X = Decimal('21.072100')
end_X = Decimal('21.373130')
step = Decimal('0.000001')
EPSILON = Decimal('1e-6')

found = []

total_X_steps = int((end_X - start_X) / step) + 1

for i in tqdm(range(total_X_steps), desc="Scanning X values"):
    current_X = start_X + i * step
    required_ratio = target / current_X
    if (start_ratio <= required_ratio <= end_ratio) and abs(required_ratio * current_X - target) < EPSILON:
        rounded_ratio = round(required_ratio, 6)
        rounded_X = round(current_X, 6)
        print(f"[FOUND] ratio={rounded_ratio}, X={rounded_X}, value={rounded_ratio*rounded_X}")
        with open("./data/phase_2.txt", "a") as file:
            file.write(f"[FOUND] ratio={rounded_ratio}, X={rounded_X}, value={rounded_ratio*rounded_X}\n")
        found.append((rounded_ratio, rounded_X, rounded_ratio*rounded_X))

print(f"Matches found: {len(found)}")

关键改进点 #

• 避免双重循环：通过ratio = target / X直接计算所需ratio，减少不必要的遍历，同时消除ratio遍历的误差。
• 整数计数计算值：用start_X + i * step代替X += step，彻底避免浮点数累加的精度漂移。
• 匹配精度控制：对计算出的ratio和X保留6位小数，确保与目标组合的精度一致。

内容的提问来源于stack exchange，提问作者Gor Barseghyan