Pre-optimized filters in research 4 | 四期科研预优化过滤器

[LmeSzinc]

Added 3 pre-optimized research filter string, optimized by dev_tools/research_optimizer.py. See research drops rates and spawn rates in azur-stats.lyoko.io. People without a basic knowledge of these are not recommended to write costum filter, but use pre-optimized presets in Alas: Series 4 Blueprints+Tenrai, Series 4 Blueprints Only, Series 4 Tenrai Only. You could also optimize a filter to meet your own needs using the optimise tool.

增加了3个预优化科研过滤器，根据 dev_tools/research_optimizer.py 优化，科研掉落和项目刷新详见 azur-stats.lyoko.io. 对这些没有基本了解的人，不建议编写自定义过滤器，建议使用预优化科研过滤器：四期 蓝图+天雷，四期 仅蓝图，四期 仅天雷。你也可以使用优化器优化出满足自己需求的过滤器。

Applicable Scene | 适用场景

Filters are simulated with these settings, no B/T/E researches, no H researches but H-0.5, run Alas 7x24, with an interval of 0 minute (feature of Alas). 过滤器由一下设置模拟而来，不做 B/T/E 科研，不做除 H-0.5 以外的 H 科研，7x24 运行 Alas，收菜间隔 0 分钟（Alas特性）

ResearchPool.remove_projects = 'B > T > E > H1 > H2 > H4'
FilterSimulator.active = 24 / 24
FilterSimulator.interval = 0 / 60 / 24

Simulated Results | 模拟结果

Series 4 Blueprints+Tenrai | 四期 蓝图+天雷

Target

# Agir, Hakuryuu, Anchorage, August, Marcopolo, Tenrai
FilterSimulator.target = np.array([513, 513, 343, 343, 343, 100])

Optimized filter

S4-DR0.5 > S4-PRY0.5 > S4-H0.5 > S4-Q0.5 > S4-DR2.5 > 0.5 > S4-G1.5
> S4-Q1 > S4-DR5 > S4-DR8 > S4-G4 > S4-PRY2.5 > 1 > S4-Q2 > reset
> S4-G2.5 > S4-PRY5 > S4-PRY8 > 1.5 > 2 > S4-Q4 > 2.5 > 4 > 5 > S4-C6
> S4-C8 > 6 > 8 > S4-C12 > 12

Simulated results

Average day costs: 157.83350708333364
Average rewards: [517.47387289 516.87603132 396.54676909 396.44919517 396.37545139 103.24530705]

Series 4 Blueprints Only | 四期 仅蓝图

Target

# Agir, Hakuryuu, Anchorage, August, Marcopolo, Tenrai
FilterSimulator.target = np.array([513, 513, 343, 343, 343, 0])

Optimized filter

S4-DR0.5 > S4-PRY0.5 > S4-H0.5 > S4-DR8 > S4-DR2.5 > S4-DR5 > S4-G1.5
> S4-PRY2.5 > S4-Q0.5 > 0.5 > S4-G2.5 > S4-Q1 > 1 > reset > S4-G4
> S4-PRY5 > 1.5 > S4-Q2 > 2 > S4-PRY8 > 2.5 > S4-Q4 > 4 > 5 > S4-C6
> 6 > S4-C8 > 8 > S4-C12 > 12

Simulated results

Average day costs: 145.0297189583339
Average rewards: [514.41146437 514.23722459 363.05105517 362.85409698 362.78996925 61.6554753 ]

Series 4 Tenrai Only | 四期 仅天雷

Target

# Agir, Hakuryuu, Anchorage, August, Marcopolo, Tenrai
FilterSimulator.target = np.array([0, 0, 0, 0, 0, 150])

Optimized filter

S4-Q0.5 > S4-PRY0.5 > S4-DR0.5 > S4-Q4 > S4-Q1 > S4-Q2 > S4-H0.5 > 0.5
> S4-G4 > S4-G1.5 > 1 > S4-DR2.5 > S4-PRY2.5 > reset > S4-G2.5 > 1.5
> 2 > 2.5 > S4-DR5 > S4-PRY5 > 4 > 5 > S4-C6 > S4-DR8 > S4-PRY8 > S4-C8
> 6 > 8 > S4-C12 > 12

Simulated results

Average day costs: 167.3758158333346
Average rewards: [216.81089944 216.309424 485.40791076 483.33995047 482.16563555 150.08036127]

[SF-monkey]

目前我需要4期天雷+4期彩船圖紙，偶尔兼顾下2、3期的彩炮。在用这个filter Q0.5 > S4-0.5 > S2-0.5 > S3-0.5 > S4-Q4 > S4-Q1 > S4-Q2 > S4-DR-2.5 > S4-G4 > S4-G1.5 > S2-Q1 > 1 > S4-D2.5 > S4-G2.5 > S4-G2.5 > S2-Q4 > reset > shortest

能否帮忙看下有无改进的地方？

[qq787951904]

为什么G项目排这么前，收益不是很低的么？

[ste-one]

请问，我如果只需要彩蓝图+天雷，这个排序可以吗，需要把G的优先级往前调整吗？（魔改的蓝图+天雷） S4-DR0.5 > S4-PRY0.5 > S4-H0.5 > S4-Q0.5 > S4-DR2.5 > 0.5

S4-Q1 > S4-DR5 > S4-DR8 > > S4-Q2 > S4-G1.5 > S4-G4 > 1 > reset S4-G2.5 > S4-PRY2.5 > S4-PRY5 > S4-PRY8 > 1.5 > 2 > S4-Q4 > 2.5 > 4 > 5 S4-C6 > S4-C8 > 6 > 8 > S4-C12 > 12