Simple question. Using 'solve' with vectors

Hi, I am a matlab rookie. Since I am doing simulation work recently, I start to learn matlab from A, again.

Here is my question. I want to deliver 'x' from the following equation such as log(x)=2/(A-B)

Here, A and B are [10,1] vectors with numbers.

So, my result x should get some matrix like x[10,1].

For this, I programmed like,

CC=zeros(10,1);
for i=1:10
syms x
solve(log(x(i))-(A(i)-B(i))/0.67)
CC(i)=CC(i)+x
end

but the command window keeps saying "??? The following error occurred converting from sym to double: Error using ==> mupadmex Error in MuPAD command: DOUBLE cannot convert the input expression into a double array.

If the input expression contains a symbolic variable, use the VPA function instead."

1. What should I change?

2. If I obtain the result, matlab usually shows in expoenetial terms. How can I order matlab to calculate completely?

Thanks.

Answer 1

[Walter Roberson]

solve() is not going to store the solutions into "x": it is only going to return the value (which would then be displayed in the command window.)

Because of that, when you get to the CC(i) = CC(i) + x line, x is still a basic symbolic variable whose "value" is its own name, but CC is double precision, so MATLAB tries to convert the name x to a double precision value and of course fails.

Assign the result of solve() to something, and add that something to CC(i) .

You may also need to use a call to double() to force conversion from symbolic numbers to double precision.

[Joon Jeon]

Thanks for quick response.

If you don't mind, let me raise one more question. I tried what you advised such as followings.

Z=[]

for i=1:22

syms x

solve (log(x(i))-(EU(i)-US(i))/0.67)

Z(i)=Z(i)+double(x)

end

However, now it says "Index exceeds matrix dimensions".

Then, I REMOVED all the (i)s in solve function such as

for i=1:22

syms x

solve (log(x)-(EU(i)-US(i))/0.67)

end

And now it gives answers(but not matrix, you know).

So, I got back to the original and removed Z(i) terms.

Then, it doesn't work.

Hm, doesn't solve save its result in x? Then, is there any other commands you can recommend to me? I need the result in matrix form.

Thanks in advance.

[Walter Roberson]

As I said before "Assign the result of solve() to something, and add that something to CC(i) ."

syms x

for i = 1 : 22

xval = solve(log(x)-(EU(i)-US(i))/0.67);

Z(i) = Z(i) + double(xval);

end

Or, more simply, just use

Z = exp((EU - US)/0.67);

to replace all of that code.

[Joon Jeon]

Walter, it's not.

It keeps saying "Index exceeds matrix dimensions"

And since (EU-US) is sometimes negative, I cannot apply logarithm rule like that.

[Alexander]

Are you sure your variables EU, US, and Z are big enough? This works for me:

EU = rand(22, 1);

US = rand(22, 1);

Z = zeros(22, 1);

syms x

for i = 1 : 22

xval = solve(log(x)-(EU(i)-US(i))/0.67);

Z(i) = Z(i) + double(xval);

end

[Walter Roberson]

What is the difficulty in applying exp() to a negative value?

[Joon Jeon]

It's from the rule of exponential.

X in the log(x) must be positive value when transform to exponential one.

[Walter Roberson]

If log(x)-(EU(i)-US(i))/0.67 = 0 (condition to be solved) then

log(x) = (EU(i)-US(i))/0.67 and then

exp(log(x)) = exp((EU(i)-US(i))/0.67) so

x = exp((EU(i)-US(i))/0.67)

and at no point are you taking the log of a negative number.