The Padé fit appears valid over the full range of the data  (0 to 18 tesla). The change in curvature above 18 T in the extrapolated fit seems unlikely but data at higher fields would be needed to be sure. Higher field data might also allow the use of additional terms in the denominator and  better constrain the high field extrapolation.

Complete Padé fit

We fit the remaining coefficients to Eq. \ref{eq:expand_y_31} in the same way as before, with one difference: we use the final numerical values for the Padé coefficients for the previous fit as the initial values for the  Padé coefficients for the new fit. Thus the coefficient values given above for \(y_0\) become the initial values for the fit to \(y_1\), the final values for \(y_1\) become the initial values for \(y_2\), etc. This approach is justified here because each  \(y_i\left(B\right)\) curve is similar to the previous curve for \(y_{i-1}\left(B\right)\), as shown in Fig. \ref{245853}
In the following figures, we show the results for \(y_1\left(B\right)\) through \(y_5\left(B\right)\).  With the exception of \(\kappa_{5,3}\), all of the coefficients are significantly different from zero (which we therefore set equal to zero). 
In general, the fits agree with the data over the entire field range but in the case of \(y_5\left(B\right)\), the fit begins to diverge from the data above 125 kG (12.5 tesla). Since this is the smallest of the coefficients in the orthogonal Chebyshev expansion, the deviation of the fit from the data will be relatively unimportant in the calculation of \(R\left(T,B\right)\). Additional data at higher fields should allow the inclusion of  higher order terms in this fit (and those for the other \(y_i\) coefficients), improving the fit to the high field behavior.