现在开始开发 deploy server / docker (fine) most basic one use ubuntu / linux. windows cost a lot of resources
Minimum Viable Product
logo绘制 color scheme 设计亮色和暗色 a11y里面应该提到过
before we call it a day
使用greedy算法来修补offspring
penalty function: what degree should I use, like I use tournement selection to choose 2 parents, then use crossover and mutation to generate 2 offspring, these offspring may not cover the
spp or set cover
benchmark reference
repair
**********standard BGA use stotistic ranking?or set a fixed coefficient in panalty function,
hyper parameters setting, generation, population setting
stotistic ranking: smallest cost function item finaly stiill have some results violate the constrains
start with a feasible one for SA? not feasible, how to solve? 可以是set cover
start with a feasible one? means that we need to start with a solution that cover all of the elements and only cover ones?