Re: pade function in Octave?

[Przemek Klosowski]

On 11/1/19 1:56 PM, vivek guajri wrote:
> Hi Allen,
>
> Thank you for your email. I have attached the excel file (includes x and y
> values and pdf file of the fit and observed curve)
>    x&y.xlsx 
> <https://gcc01.safelinks.protection.outlook.com/?url=https%3A%2F%2Foctave.1599824.n4.nabble.com%2Ffile%2Ft373519%2Fx%2526y.xlsx&amp;data=02%7C01%7Cprzemek.klosowski%40nist.gov%7C052da038cec047eabf0808d75ef4c953%7C2ab5d82fd8fa4797a93e054655c61dec%7C1%7C1%7C637082277749877919&amp;sdata=DDhJqUN9oL9mHlip9pcIqw1gFKyNpz2ruqPxXlXYtUo%3D&amp;reserved=0>
>    curvefitting.pdf
> <https://gcc01.safelinks.protection.outlook.com/?url=https%3A%2F%2Foctave.1599824.n4.nabble.com%2Ffile%2Ft373519%2Fcurvefitting.pdf&amp;data=02%7C01%7Cprzemek.klosowski%40nist.gov%7C052da038cec047eabf0808d75ef4c953%7C2ab5d82fd8fa4797a93e054655c61dec%7C1%7C1%7C637082277749877919&amp;sdata=JWvH%2BDTY1CPTHBbIL7%2FplIxk7uwY4QlhhJ2MJg6jSWU%3D&amp;reserved=0>
>   -

But the data in the .pdf are completely different from the data in your 
previous email---the former look like they may be fitted by a polynomial 
of a reasonably low degree, but the latter aren't. You of course can fit 
any data if you allow high order polynomials, but it's almost always a 
bad idea because such polynomials will behave crazily between and 
outside your data points.

Even the data from your .xslx spreadsheet look funny---there are 
discontinuities in the x values (between point 14 and 15, and between 19 
and 20) that don't look real but suggest some sort of measurement error. 
They could be fit by a third-order polynomial but it looks wrong.

What is the data, what are the uncertainities of the data, and what is 
the physical model that you expect to predict the data? If you can 
answer these questions, then the fitted parameters of this model will 
have a physical meaning. Otherwise it'll be just more-less random numbers.