Closed davidrpugh closed 10 years ago
The "better fix" required changing the way A0 and L0 were calibrated (specifically calibrating them based on the first available year of data rather that A0=A(0) and L0=L(0)); and changing the ylims to be +- 5% of the min/max values of the variable being plotted.
Currently, the y-scale for the plot_impulse response method of the SolowModel class has hard coded ylims for kind='per_capita' and kind='levels'. This was a quick fix for some deeper issue to do with setting 'log=True' leading to ylims of 1e-20, 1e10 which was undesirable. Need to find a better fix.