Fix plotting to stop using methods/attributes removed from models (#491)

Summary:

Dim grid no longer exists in models, plotting code updated to use standalone dim_grid.

Bounds are gone from models so get_lse_interval needs to separately define the bounds of the grid to sample from.

Differential Revision: D67535775

aepsych/plotting.py

@@ -11,7 +11,7 @@
 import matplotlib.pyplot as plt
 import numpy as np
 from aepsych.strategy import Strategy
-from aepsych.utils import get_lse_contour, get_lse_interval, make_scaled_sobol
+from aepsych.utils import dim_grid, get_lse_contour, get_lse_interval, make_scaled_sobol
 from matplotlib.axes import Axes
 from matplotlib.image import AxesImage
 from scipy.stats import norm
@@ -174,7 +174,7 @@ def _plot_strat_1d(
     assert x is not None and y is not None, "No data to plot!"
 
     if strat.model is not None:
-        grid = strat.model.dim_grid(gridsize=gridsize).cpu()
+        grid = dim_grid(lower=strat.lb, upper=strat.ub, gridsize=gridsize).cpu()
         samps = norm.cdf(strat.model.sample(grid, num_samples=10000).detach())
         phimean = samps.mean(0)
     else:
@@ -196,9 +196,6 @@ def _plot_strat_1d(
     if target_level is not None:
         from aepsych.utils import interpolate_monotonic
 
-        lb = strat.transforms.untransform(strat.lb)[0]
-        ub = strat.transforms.untransform(strat.ub)[0]
-
         threshold_samps = [
             interpolate_monotonic(grid, s, target_level, strat.lb[0], strat.ub[0])
             for s in samps
@@ -300,7 +297,7 @@ def _plot_strat_2d(
     if strat.model is not None:
         strat.model.fit(train_x=x, train_y=y, max_fit_time=None)
 
-        grid = strat.model.dim_grid(gridsize=gridsize)
+        grid = dim_grid(lower=strat.lb, upper=strat.ub, gridsize=gridsize).cpu()
         fmean, _ = strat.model.predict(grid)
         phimean = norm.cdf(fmean.reshape(gridsize, gridsize).detach().numpy()).T
     else:
@@ -341,6 +338,8 @@ def _plot_strat_2d(
             model=strat.model,
             mono_grid=mono_grid,
             target_level=target_level,
+            grid_lb=strat.lb,
+            grid_ub=strat.ub,
             cred_level=cred_level,
             mono_dim=1,
             lb=mono_grid.min(),
@@ -513,7 +512,12 @@ def plot_slice(
     """
     extent = np.c_[strat.lb, strat.ub].reshape(-1)
     if strat.model is not None:
-        x = strat.model.dim_grid(gridsize=gridsize, slice_dims={slice_dim: slice_val})
+        x = dim_grid(
+            lower=strat.lb,
+            upper=strat.ub,
+            gridsize=gridsize,
+            slice_dims={slice_dim: slice_val},
+        ).cpu()
     else:
         raise RuntimeError("Cannot plot without a model!")
     if lse:

aepsych/utils.py

@@ -168,13 +168,19 @@ def interpolate_monotonic(
     y1 = y[idx]
 
     x_star = x0 + (x1 - x0) * (z - y0) / (y1 - y0)
-    return x_star.cpu().item()
+
+    if isinstance(x_star, torch.Tensor):
+        return x_star.cpu().item()
+    else:
+        return x_star
 
 
 def get_lse_interval(
     model: GPyTorchModel,
     mono_grid: Union[torch.Tensor, np.ndarray],
     target_level: float,
+    grid_lb: torch.Tensor,
+    grid_ub: torch.Tensor,
     cred_level: Optional[float] = None,
     mono_dim: int = -1,
     n_samps: int = 500,
@@ -189,11 +195,17 @@ def get_lse_interval(
         model (GPyTorchModel): Model to use for sampling.
         mono_grid (Union[torch.Tensor, np.ndarray]): Monotonic grid.
         target_level (float): Target level.
+        grid_lb (torch.Tensor): The lower bound of the grid to sample from to calculate
+            LSE.
+        grid_ub (torch.Tensor): The upper bound of the grid to sample from to calculate
+            LSE.
         cred_level (float, optional): Credibility level. Defaults to None.
         mono_dim (int): Monotonic dimension. Defaults to -1.
         n_samps (int): Number of samples. Defaults to 500.
-        lb (float): Lower bound. Defaults to -float("inf").
-        ub (float): Upper bound. Defaults to float("inf").
+        lb (float): Theoreticaly true lower bound for the parameter. Defaults to
+            -float("inf").
+        ub (float): Theoretical true uppper bound for the parameters. Defaults to
+            float("inf").
         gridsize (int): Grid size. Defaults to 30.
 
     Returns:
@@ -203,7 +215,7 @@ def get_lse_interval(
     xgrid = torch.stack(
         torch.meshgrid(
             [
-                torch.linspace(model.lb[i].item(), model.ub[i].item(), gridsize)
+                torch.linspace(grid_lb[i].item(), grid_ub[i].item(), gridsize)
                 for i in range(model.dim)
             ]
         ),