thulc

    # phase_real_mean 是已知的算力消耗平均值，拿到了这个均值之后，进行 lambda_vector 的一次的更新
    def update_lambda_vector(self, lambda_vector, phase_real_mean, target_real):
        
        # 从字符串到1D tensor一条龙服务
        auto_lambda_alpha = [float(s) for s in self.global_params["auto_lambda_alpha"].split("_")]
        auto_lambda_alpha = [tf.reshape(tf.convert_to_tensor(x), [1]) for x in auto_lambda_alpha]
        auto_lambda_alpha = tf.concat(auto_lambda_alpha, axis=0)
        
        delta_lambda = auto_lambda_alpha*((phase_real_mean - target_real)/target_real)

thulc

http://web.stanford.edu/class/cme241/

Overview of the Course
Theory of Markov Decision Processes (MDPs)
Dynamic Programming (DP) Algorithms
Backward Induction (BI) and Approximate DP (ADP) Algorithms
Reinforcement Learning (RL) Algorithms
Plenty of Python implementations of models and algorithms
We apply these algorithms to 5 Financial/Trading problems:
(Dynamic) Asset-Allocation to maximize Utility of Consumption

thulc

from ctypes.util import find_library
print(find_library("gs"))
import camelot

# pdf path
path3 = '/Users/paul/Code/2015年中册_small.pdf'

# data save path
data_path='/Users/paul/PycharmProjects/pythonProject/pdf2excel/data/'

thulc

这里有强大的搜索引擎
https://www.projectpro.io/article/100-machine-learning-datasets-curated-for-you/407

谷歌数据集搜索
https://datasetsearch.research.google.com/

thulc

# how ?
# just read paper and write code