This is the numerical simulation code for SquareCBwK, the main algorithm in our paper Optimal Contextual Bandits with Knapsacks under Realizibility via Regression Oracles, which is accepted by AISTATS2023. We present simulation results for linear CBwK to demonstrate the dependency on time horizon ($T$), dimension ($m$), and number of arms ($K$) of SquareCBwK utilizing Newtonized GLMtron and Online Gradient Descent oracles. In comparison with the performance of LinUCB (Agrawal and Devanur, 2016), SquareCBwK with Newtonized GLMtron oracles exhibits an improve dependency on dimension $m$ although with additional dependence on K.