如何用R生成指定长度、总和与数值范围的随机向量？

Here's a function that meets all your requirements: fixed vector length, elements within specified range, sum to the target total, and different results on each run:

rand_fixed_vector <- function(n, min_val, max_val, total_sum) {
  # Check if the total sum is feasible given constraints
  if (n * min_val > total_sum || n * max_val < total_sum) {
    stop("Error: Total sum is outside the feasible range for the given parameters.")
  }
  
  # Adjust values to work with non-negative constraints
  adjusted_total <- total_sum - n * min_val
  max_adjusted <- max_val - min_val
  
  # Generate initial vector of non-negative integers summing to adjusted_total
  adjusted_vector <- as.vector(rmultinom(1, adjusted_total, rep(1, n)))
  
  # Adjust any values that exceed the maximum allowed adjustment
  while (any(adjusted_vector > max_adjusted)) {
    # Calculate excess values beyond the max allowed
    excess <- sum(adjusted_vector[adjusted_vector > max_adjusted] - max_adjusted)
    # Cap those values at the maximum allowed adjustment
    adjusted_vector[adjusted_vector > max_adjusted] <- max_adjusted
    # Find positions where we can distribute the excess
    available_positions <- which(adjusted_vector < max_adjusted)
    # Distribute the excess to available positions randomly
    excess_additions <- as.vector(rmultinom(1, excess, rep(1, length(available_positions))))
    adjusted_vector[available_positions] <- adjusted_vector[available_positions] + excess_additions
  }
  
  # Convert back to the original value range
  final_vector <- adjusted_vector + min_val
  
  # Verify all constraints are met (optional, for robustness)
  stopifnot(
    length(final_vector) == n,
    sum(final_vector) == total_sum,
    all(final_vector >= min_val),
    all(final_vector <= max_val)
  )
  
  return(final_vector)
}

Why Your Original Function Failed #

Your original function produces a vector with length equal to max - min + 1 (since it iterates over min:max), not the fixed length you need. For example, with min=0 and max=100, it returns a vector of length 101 instead of 15. It's counting occurrences of each value in the sample range, not generating a fixed-length vector of values that sum to your target.

Usage Examples #

Example 1: Fixed length 15, values 0-100, sum 100 #

# Generate the vector
vec1 <- rand_fixed_vector(n=15, min_val=0, max_val=100, total_sum=100)

# Check constraints
length(vec1)          # 15
sum(vec1)             # 100
all(vec1 >= 0 & vec1 <= 100)  # TRUE

Example 2: Modified parameters (length 10, values 5-20, sum 100) #

vec2 <- rand_fixed_vector(n=10, min_val=5, max_val=20, total_sum=100)

sum(vec2)             # 100
length(vec2)          #10
all(vec2 >=5 & vec2 <=20) # TRUE

Example 3: Edge case (all values at minimum) #

vec3 <- rand_fixed_vector(n=5, min_val=3, max_val=10, total_sum=15)
# Output will be c(3,3,3,3,3) since 5*3=15

Each time you run the function without setting a seed, you'll get a different valid vector, as it uses R's random number generator.

内容的提问来源于stack exchange，提问作者LOS_2023