Matrix dimensions must agree

Question

b.m.n n 2016-11-28

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/314426-matrix-dimensions-must-agree

编辑： b.m.n n 2017-2-23

Hello... I use {air_sea ToolBox } So when I Run it , I get this error (Matrix dimensions must agree.),How can I fix them . (error message is >>> Error using * Inner matrix dimensions must agree.)..

3 个评论
显示 1更早的评论隐藏 1更早的评论

KSSV 2016-11-29

This error occurs, mostly when you are trying to multiply two matrices which don't obey multiplication condition. Go to the specific line where error pops out, and check the dimensions of the matrices.

Jan 2016-12-5

@b.m.n n: As James has said before: We need the complete message. There is also a part, which tells the failing line. It is not efficient, if we guess, where the error occurres.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Guillaume 2016-11-29

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/314426-matrix-dimensions-must-agree#answer_245151

在 MATLAB Online 中打开

There are a few things that are confusing in your code:

lwest = ...
...; clear all

So you wipe out all the variables including the lwest. What was the point in calculating it in the first place, then. Is there actually a need for the clc; close all; clear all or is it just something you always write regardless of if it's needed or not?

Tsea = A(:,2);     %so Tsea is a column vector, size Nx1
TE = Tsea(E,:);    %since Tsea has only one column, : is just that one column.

So why the :? Was Tsea expected to contain more than one column or is it just redundant and there to confuse the reader?

TE = Tsea(E); %would work just the same and is less confusing

It's not clear where your error is coming from since you haven't told us which line is causing the error, and there does not appear to be any matrix multiplication in your code. However, the following line is suspicious

e_a=e_a.';%795*1

The comment is wrong. Since e_a is a function of TE it starts as a column vector, and the above line transposes it to a row vector, so it'll be size 1xN, not Nx1.

Moreover, in your last line you have:

something = f(TE) .* f(e_a) .* Fc
something = 795x1 .* 1x795 .* Fc (12x1? according to earlier comment)

The first part of the multiplication will only work in R2016b, in earlier version it will produce an error. In R2016, it will create a 795 x 795 matrix. Whichever version, if Fc is indeed 12x1, it can't be multiplied with a 795 x 795 matrix.

2 个评论
显示无隐藏无

b.m.n n 2016-12-3

Thanks for your concern (So why the :)in Tsea = A(:,2);% defined Due to (A )the original data and (TSea) part of them. Fc>>> is The Monthly average for one year.

Guillaume 2016-12-5

在 MATLAB Online 中打开

My point about the : wasn't with the line Tsea = A(:, 2), that is fine and makes Tsea a column vector. It was with the line TE = Tsea(E, :). The : selects all column, but by definition Tsea has only one column, so the colon is redundant and the line is equivalent to TE = Tsea(E). This raises the question: is there a bug? Did the author meant to select more than one column of something but used the wrong variable?

As stated, the line

lwest = -emiss_lw*sigmaSB.*TE.^4.*(0.39 - 0.05*sqrt(e_a)).*Fc;

is never going to work if FC is 12x1 and TE is 795x1, as you can't multiply matrices of incompatible size. In versions prior to R2016b, the line would also fail because you're doing a memberwise multiplication of TE, a 795x1 matrix, with e_a, a 1x795 matrix. It will work in R2016b because of the new implicit expansion, but will result in a 795x795 matrix which may not be the intent.

请先登录，再进行评论。