more analysis

1 files changed, 57 insertions, 75 deletions

diff --git a/analysis/vis.livemd b/analysis/vis.livemd
index 30401ca..33e0b54 100644
--- a/analysis/vis.livemd
+++ b/analysis/vis.livemd
@@ -14,6 +14,7 @@ Mix.install([
 
 ```elixir
 require Explorer.DataFrame
+require Explorer.Series
 alias Explorer.DataFrame, as: DF
 alias Explorer.Series, as: SE
 job_id = "1147"
@@ -35,31 +36,13 @@ cost_model_files =
 cost_model_files
 ```
 
-```elixir
-defmodule CostModel do
-  defstruct impl: nil, ops: []
-end
-
-defmodule OpCostModel do
-  defstruct op: nil, x0: nil, x1: nil, x2: nil, x3: nil, nmrse: nil
-end
-
-defmodule Parse do
-  def cost_model_row(row) do
-    [name, x0, x1, x2, x3, nmrse] = String.split(row, " & ")
-    [nmrse | _] = String.split(nmrse)
-
-    %OpCostModel{
-      op: name,
-      x0: String.to_float(x0),
-      x1: String.to_float(x1),
-      x2: String.to_float(x2),
-      x3: String.to_float(x3),
-      nmrse: String.to_float(nmrse)
-    }
-  end
+<!-- livebook:{"reevaluate_automatically":true} -->
 
-  def cost_model_output(fname) do
+```elixir
+# Parse cost model information
+cost_models =
+  cost_model_files
+  |> Enum.map(fn fname ->
     {:ok, contents} = File.read(fname)
     [_, table | _] = String.split(contents, "line\n")
     [rows | _] = String.split(table, "\n\\end")
@@ -68,73 +51,72 @@ defmodule Parse do
     [_, impl] = String.split(fname, "cost-model-")
     impl = String.replace(impl, "-", ":")
 
-    %CostModel{
-      impl: impl,
-      ops: rows |> Enum.map(&cost_model_row/1)
-    }
-  end
-end
+    rows
+    |> Enum.map(fn row ->
+      [name, x0, x1, x2, x3, nmrse] = String.split(row, " & ")
+      [nmrse | _] = String.split(nmrse)
+
+      %{
+        op: name,
+        x0: String.to_float(x0),
+        x1: String.to_float(x1),
+        x2: String.to_float(x2),
+        x3: String.to_float(x3),
+        nmrse: String.to_float(nmrse)
+      }
+    end)
+    |> Enum.map(fn d -> d |> Map.put(:impl, impl) end)
+    |> DF.new()
+  end)
+  |> DF.concat_rows()
 ```
 
-<!-- livebook:{"reevaluate_automatically":true} -->
+```elixir
+Tucan.histogram(cost_models, "nmrse")
+```
 
 ```elixir
-cost_models = cost_model_files |> Enum.map(&Parse.cost_model_output/1)
+cost_models
+|> DF.filter(op == "contains")
 ```
 
 ## Cost model exploratory plots
 
 ```elixir
-defmodule PlotCostModel do
-  @startn 0
-  @endn 20000
-  @resolution 100
-  def gen_ts(ns, %OpCostModel{x0: x0, x1: x1, x2: x2, x3: x3}) do
-    Enum.map(ns, fn n -> %{n: n, t: x0 + n * x1 + n * n * x2 + n * n * n * x3} end)
-  end
-
-  def points_op(op) do
-    ns = @startn..@endn//@resolution
-
-    gen_ts(ns, op)
-    |> Enum.map(fn data -> Map.put(data, :name, op.op) end)
-  end
-
-  def points_model(model) do
-    model.ops
-    |> Enum.map(fn op ->
-      points_op(op)
-      |> Enum.map(fn point -> Map.put(point, :impl, model.impl) end)
-    end)
-    |> List.flatten()
-  end
-
-  def points(models) do
-    models
-    |> Enum.map(&points_model/1)
-    |> List.flatten()
-  end
-
-  def plot_op(op) do
-    points = points_op(op)
-    Tucan.lineplot(points, "n", "t", title: op.op)
-  end
-
-  def plot_all_ops(%CostModel{impl: impl, ops: ops}) do
-    ops
-    |> Enum.map(&plot_op/1)
-    |> Tucan.concat(columns: 3)
-    |> Tucan.set_title(impl)
-  end
+startn = 0
+endn = 200_000
+resolution = 100
+
+points_for = fn impl, op ->
+  %{"x0" => [x0], "x1" => [x1], "x2" => [x2], "x3" => [x3]} =
+    DF.filter(cost_models, impl == ^impl and op == ^op)
+    |> DF.to_columns()
+
+  Enum.map(startn..endn//resolution, fn n ->
+    %{
+      impl: String.split(impl, "::") |> List.last(),
+      op: op,
+      n: n,
+      t: x0 + n * x1 + n * n * x2 + n * n * n * x3
+    }
+  end)
+  |> DF.new()
 end
 ```
 
 <!-- livebook:{"reevaluate_automatically":true} -->
 
 ```elixir
+inspect_op = "contains"
+
 cost_models
-|> Enum.map(&PlotCostModel.plot_all_ops/1)
-|> Tucan.vconcat()
+|> DF.filter(op == ^inspect_op)
+|> DF.distinct(["impl"])
+|> DF.to_rows()
+|> Enum.map(fn %{"impl" => impl} -> points_for.(impl, inspect_op) end)
+|> DF.concat_rows()
+|> Tucan.lineplot("n", "t", color_by: "impl", clip: true)
+|> Tucan.Scale.set_y_domain(0, 200)
 ```
 
 ## Read benchmark data