Theoretical stellar evolution models are constructed and tailored to the best known , observationally derived characteristics of metal-poor ( [ Fe/H ] \sim - 2.3 ) stars representing a range of evolutionary phases : subgiant HD140283 , globular cluster M92 , and four single , main sequence stars with well-determined parallaxes : HIP46120 , HIP54639 , HIP106924 , and WOLF1137 .
It is found that the use of a solar-calibrated value of the mixing length parameter \alpha _ { \text { MLT } } in models of these objects is ineffective at reproducing their observed properties .
Empirically calibrated values of \alpha _ { \text { MLT } } are presented for each object , accounting for uncertainties in the input physics employed in the models .
It is advocated that the implementation of an adaptive mixing length is necessary in order for stellar evolution models to maintain fidelity in the era of high precision observations .