Hi

Let y=f(x,p), where p is the forwad model parameter.
By assuming there is no error of y, I'd like to estimate the error of
retrieved x due to the assumption of p.
Very often, the error of retrieved x is performed by considering the
range of p from mean_p - stdev_p to mean_p + stdev_p.
The mean_p and stdev_p can be decided by measurements.
The retrieved x is x0 and x1 when p is mean_p - stdev_p and mean_p +
stdev_p.
Should the error in the retrieved x be
1/ x1-x0
2/(x1-x0)/2

And when considered as relative error (%), it is
1/(x1-x0)/x_mean*100
2/(x1-x0)/(x1+x0)*2*100  ?
x_mean is the retrieved value for mean_p.

These two problems have been confusing me for a long time.
Thank you in advance.

Mike

On Apr 7, 3:23 pm, Mike <Sulfate...@gmail.com> wrote:





Hi

Let y=f(x,p), where p is the forwad model parameter.
By assuming there is no error of y, I'd like to estimate the error of
retrieved x due to the assumption of p.
Very often, the error of retrieved x is performed by considering the
range of p from mean_p - stdev_p to mean_p + stdev_p.
The mean_p and stdev_p can be decided by measurements.
The retrieved x is x0 and x1 when p is mean_p - stdev_p and mean_p +
stdev_p.
Should the error in the retrieved x be
1/ x1-x0
2/(x1-x0)/2

And when considered as relative error (%), it is
1/(x1-x0)/x_mean*100
2/(x1-x0)/(x1+x0)*2*100  ?
x_mean is the retrieved value for mean_p.

These two problems have been confusing me for a long time.
Thank you in advance.

Mike

Any suggestions would be very appreciated.

Mike- Hide quoted text -

- Show quoted text -
Any suggestions would be very appreciated.