category

language_learning/python/2021.05.10_parallel_calculations_using_sh_files/a.py

@@ -0,0 +1,29 @@
+import numpy as np
+
+# 设置
+cpus = 7  # 使用的CPU个数（等于提交任务的个数）
+parameter_array_all = np.arange(0, 10, 0.1) # 需要计算的参数
+
+
+# 通过.sh脚本文件修改的任务指标。job_index从0开始，最大值为cpus-1
+job_index = -1
+
+
+# 预处理
+len_of_parameter_all = len(parameter_array_all)  # 需要计算参数的个数
+if len_of_parameter_all%cpus == 0:
+	len_parameter = int(len_of_parameter_all/cpus) # 一个CPU/任务需要计算参数的个数
+	parameter_array = parameter_array_all[job_index*len_parameter:(job_index+1)*len_parameter]
+else:
+	len_parameter = int(len_of_parameter_all/(cpus-1)) # 一个CPU/任务需要计算参数的个数
+	if job_index != cpus-1:
+		parameter_array = parameter_array_all[job_index*len_parameter:(job_index+1)*len_parameter]
+	else:
+		parameter_array = parameter_array_all[job_index*len_parameter:len_of_parameter_all]
+
+
+# 任务
+with open('a'+str(job_index)+'.txt', 'w') as f:
+	for parameter in parameter_array:
+		result = parameter**2
+		f.write(str(parameter)+'            '+str(result)+'\n')

language_learning/python/2021.05.10_parallel_calculations_using_sh_files/a.sh

@@ -0,0 +1,5 @@
+#!/bin/sh
+#PBS -N task
+#PBS -l nodes=1:ppn=1
+export OMP_NUM_THREADS=1
+python a.py

language_learning/python/2021.05.10_parallel_calculations_using_sh_files/combine.py

@@ -0,0 +1,6 @@
+f = open('combine.txt', 'w')
+for job_index in range(7):
+    with open('a'+str(job_index)+'.txt', 'r') as f0:
+        text = f0.read()
+    f.write(text)
+f.close()

language_learning/python/2021.05.10_parallel_calculations_using_sh_files/task.sh

@@ -0,0 +1,14 @@
+#!/bin/sh
+
+for ((job_index=0; job_index<7; job_index++))
+do
+
+cp a.py a${job_index}.py
+sed -i "s/job_index = -1/job_index = ${job_index}/" a${job_index}.py
+
+
+cp a.sh a${job_index}.sh
+sed -i "s/python a.py/python a${job_index}.py/" a${job_index}.sh
+qsub a${job_index}.sh
+
+done