Currently, plot_2D_FTLE plots all FTLE values at the grid location - that is, the location corresponding to t0 (start time for the FTLE algorithm). This is incorrect if the idea is to use the smallest FTLE to identify attracting LCS at time T (finish time for the algorithm) - these data should be plotted at the spatial location given by time T. The environment.plot_2D_FTLE should likely do this automatically.
Of course, if T is large, many agents will have left the domain and we should not plot these values. If the goal is to get decent data for both attracting and repelling LCS, the user will need to find an integration time that balances these goals.
In this context, it may be worth implementing a backward-time routine for better resolution when looking for attracting LCS.
Currently, plot_2D_FTLE plots all FTLE values at the grid location - that is, the location corresponding to t0 (start time for the FTLE algorithm). This is incorrect if the idea is to use the smallest FTLE to identify attracting LCS at time T (finish time for the algorithm) - these data should be plotted at the spatial location given by time T. The environment.plot_2D_FTLE should likely do this automatically.
Of course, if T is large, many agents will have left the domain and we should not plot these values. If the goal is to get decent data for both attracting and repelling LCS, the user will need to find an integration time that balances these goals.
In this context, it may be worth implementing a backward-time routine for better resolution when looking for attracting LCS.