Why are symbolic expressions being rounded?

Question

Bill Tubbs 2020-9-2

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/588046-why-are-symbolic-expressions-being-rounded

回答： Bill Tubbs 2020-9-3

I'm deriving some expressions symbolically and then substituting values and computing the results and I only recently noticed that the results are being rounded.

>> format long
>> syms x; double(subs(10/x,x,3))
ans =
   3.333300000000000
   

or simply,

>> x = sym(3.33333333)
 
x =
 
3.3333
 
>> double(x)
ans =
   3.333300000000000

I can't even reproduce results from the documentation it seems:

>> syms x
>> p = sym(pi);
>> piVpa = vpa(p)
 
piVpa =
 
3.1416
 

According to the documentation, this is supposed to produce:

piVpa =

3.1415926535897932384626433832795

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Bill Tubbs 2020-9-2

编辑：Bill Tubbs 2020-9-2

在 MATLAB Online 中打开

Actually, I think it is double that is doing the rounding:

>> syms x; 10/3 - subs(10/x,x,3)
 
ans =
 
0
 
>> 10/3 - double(subs(10/x,x,3))
ans =
     3.333333333355171e-05

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Bill Tubbs 2020-9-3

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/588046-why-are-symbolic-expressions-being-rounded#answer_489355

在 MATLAB Online 中打开

Mathworks technical support have diagnosed the problem. Don't know what caused it originally but the 'FloatingPointOutput' setting in the symbolic preferences was set to 1.

Here is the solution to fix it:

>> sympref
ans = 
  struct with fields:
           FourierParameters: [1×2 sym]
           HeavisideAtOrigin: [1×1 sym]
            AbbreviateOutput: 1
               TypesetOutput: 1
         FloatingPointOutput: 1
      PolynomialDisplayStyle: 'default'
    MatrixWithSquareBrackets: 0
>> sympref('FloatingPointOutput','default');
>> double(sym('pi'))
ans =
   3.141592653589793
>> p = sym(pi);
>> piVpa = vpa(p)
 
piVpa =
 
3.1415926535897932384626433832795
 