julia - Julia 中的类型错误

Question

我创建了一个名为 HousingData 的新结构，还定义了迭代和长度等函数。但是，当我为 HousingData 对象使用函数 collect 时，我遇到了以下错误。

TypeError: in typeassert, expected Integer, got a value of type Float64

import Base: length, size, iterate
struct HousingData
    x
    y
    batchsize::Int
    shuffle::Bool
    num_instances::Int

    function HousingData(
        x, y; batchsize::Int=100, shuffle::Bool=false, dtype::Type=Array{Float64})
    
        new(convert(dtype,x),convert(dtype,y),batchsize,shuffle,size(y)[end])
    end
end

function length(d::HousingData)
    
    return ceil(d.num_instances/d.batchsize)
end



function iterate(d::HousingData, state=ifelse(
    d.shuffle, randperm(d.num_instances), collect(1:d.num_instances)))
   
     
    if(length(state)==0)
        return nothing
    end
    return ((d.x[:,state[1]],d.y[:,state[1]]),state[2:end])
end


x1 = randn(5, 100); y1 = rand(1, 100);
obj = HousingData(x1,y1; batchsize=20)


collect(obj)

Answer 1

您的代码中存在多个问题。第一个与 length 相关，不返回整数，而是返回 float 。 ceil 的行为解释了这一点:

julia> ceil(3.8)
4.0 # Notice: 4.0 (Float64) and not 4 (Int)

你可以很容易地解决这个问题:

function length(d::HousingData)
    return Int(ceil(d.num_instances/d.batchsize))
end

还有一个问题是你迭代函数的逻辑，和标榜的长度不一致。举一个比你的小的例子:

julia> x1 = [i+j/10 for i in 1:2, j in 1:6]
2×6 Array{Float64,2}:
 1.1  1.2  1.3  1.4  1.5  1.6
 2.1  2.2  2.3  2.4  2.5  2.6

# As an aside, unless you really want to work with 1xN matrices
# it is more idiomatic in Julia to use 1D Vectors in such situations
julia> y1 = [Float64(j) for i in 1:1, j in 1:6]
1×6 Array{Float64,2}:
 1.0  2.0  3.0  4.0  5.0  6.0

julia> obj = HousingData(x1,y1; batchsize=3)
HousingData([1.1 1.2 … 1.5 1.6; 2.1 2.2 … 2.5 2.6], [1.0 2.0 … 5.0 6.0], 3, false, 6)

julia> length(obj)
2

julia> for (i, e) in enumerate(obj)
           println("$i -> $e")
       end
1 -> ([1.1, 2.1], [1.0])
2 -> ([1.2, 2.2], [2.0])
3 -> ([1.3, 2.3], [3.0])
4 -> ([1.4, 2.4], [4.0])
5 -> ([1.5, 2.5], [5.0])
6 -> ([1.6, 2.6], [6.0])

迭代器产生 6 个元素，而此对象的 length 只有 2。这解释了为什么 collect 出错:

julia> collect(obj)
ERROR: ArgumentError: destination has fewer elements than required

了解您的代码，您可能是修复其逻辑的最佳人选。