r-ompr:当我向模型添加一些约束时出现错误消息

Question

当我向OMPR模型添加约束时，我收到一条错误消息(它像这样正常工作(

n = dim(note_mpg)[1]
nb_joueurs = 18
perf = scale(note_mpg$performance_beta)
cote = note_mpg$cote_alpha
poste = note_mpg$Poste
note_mpg$Buts[is.na(note_mpg$Buts)] <- 0
buts = scale(note_mpg$Buts)
results = MIPModel() %>%
add_variable(z[i], i = 1:n, type = "binary") %>%
set_objective(sum_expr((perf[i] + buts[i]) * z[i], i = 1:n), "max") %>%
add_constraint(sum_expr(z[i], i = 1:n) == nb_joueurs) %>%
# add_constraint(sum_expr( (poste[i] == "G") * z[i], i = 1:n) == 2) %>%
# add_constraint(sum_expr( (poste[i] == "D") * z[i], i = 1:n) == 6) %>%
# add_constraint(sum_expr( (poste[i] == "M") * z[i], i = 1:n) == 6) %>%
# add_constraint(sum_expr( (poste[i] == "A") * z[i], i = 1:n) == 4) %>%
add_constraint(sum_expr(cote[i] * z[i], i = 1:n) <= 500) %>%
solve_model(with_ROI(solver = "glpk")) %>% 
get_solution(z[i]) %>% 
filter(value > 0)

如果我在poste上添加了一个/一些约束(我删除了注释上的#(，我会收到消息

Error in check_for_unknown_vars_impl(model, the_ast) : 
The expression contains a variable that is not part of the model.

非常感谢：(

Answer 1

我最近遇到了类似的问题。我能够在索引中使用过滤函数来修复它，而不是使用sum_expr中的比较。

# Example to replicate your poste variable
poste = rep(LETTERS[1:5],2)
print(poste)
#  [1] "A" "B" "C" "D" "E" "A" "B" "C" "D" "E"

# function that accepts the indices and the letter you want to filter poste to
# returns a vector of T/F (one for each index in i_indices)
filter_function <- function(i_indices,letter){
# A list of indices that align to each of the letters in poste
# Change this for your actual data
index_list = lapply(unique(poste),function(letter) which(poste==letter))
names(index_list) = unique(poste)

# Get the T/F value for each index in i_indices
# T if poste[index] == the provided letter
# F otherwise
return(sapply(i_indices,function(index) index %in% index_list[[letter]]))
}

# Build the model
m = MIPModel() %>%
add_variable(z[i],i=1:10,type='binary') %>%
# Call the filter function after your indices
# Passing the index and the letter you want to limit the indices to
add_constraint(sum_expr(z[i], i = 1:10,
filter_function(i,'B')) == 2)
m$constraints
# Only sums the indices of z where poste == 'B'
# (i = 2 and i = 7)
# [[1]]
# $lhs
# expression(z[2L] + z[7L])
# 
# $sense
# [1] "=="
# 
# $rhs
# expression(2)
# 
# attr(,"class")
# [1] "model_constraint"

Answer 2

感谢@cookesd的回复，并对延误表示歉意。

我终于找到了办法，但不是很干净。。。

results= MIPModel() %>%
add_variable(z[i], i = 1:n, type = "binary") %>%
set_objective(sum_expr((perf[i] + buts[i]) * z[i], i = 1:n), "max") %>%
add_constraint(sum_expr(z[i], i = 1:n) == nb_joueurs) %>%
add_constraint(sum_expr(cote[i] * z[i], i = 1:n) <= 500)  %>%
add_constraint( sum_expr(z[i], i = 1:n, poste[i] == "G") == as.numeric(input$gardiens)) %>%
add_constraint( sum_expr(z[i], i = 1:n, poste[i] == "D") == as.numeric(input$def)) %>%
add_constraint( sum_expr(z[i], i = 1:n, poste[i] == "M") == as.numeric(input$mil)) %>%
add_constraint( sum_expr(z[i], i = 1:n, poste[i] == "A") == as.numeric(input$att))

contraint3 = as.expression(sum_expr(z[i], i = 1:n, poste[i] == "G"))
contraint4 = as.expression(sum_expr(z[i], i = 1:n, poste[i] == "D"))
contraint5 = as.expression(sum_expr(z[i], i = 1:n, poste[i] == "M"))
contraint6 = as.expression(sum_expr(z[i], i = 1:n, poste[i] == "A"))

results$constraints[[3]]$lhs =contraint3
results$constraints[[4]]$lhs =contraint4
results$constraints[[5]]$lhs =contraint5
results$constraints[[6]]$lhs =contraint6

我手动添加results$constraints[[k]]$lhs的值

对于您的问题，我检查了一下，打印值时一切正常。。。我不理解这个bug，如果你有其他想法，请不要犹豫。