Add compile and run argument classes

examples/coin.py

@@ -4,15 +4,30 @@
 import matplotlib.pyplot as plt
 import seaborn as sns
 
-
-with treeppl.Model(filename="coin.tppl", samples=100000) as coin:
+with treeppl.Model(filename="coin.tppl", m="smc-bpf", particles=100000) as coin:
     res = coin(
         outcomes=[
-            True, True, True, False, True, False, False, True, True, False,
-            False, False, True, False, True, False, False, False, False, False,
+            True,
+            True,
+            True,
+            False,
+            True,
+            False,
+            False,
+            True,
+            True,
+            False,
+            False,
+            False,
+            True,
+            False,
+            True,
+            False,
+            False,
+            False,
+            False,
+            False,
         ]
     )
-    sns.histplot(
-        x=res.samples, weights=res.nweights, bins=100, stat="density", kde=True
-    )
+    sns.histplot(x=res.samples, weights=res.nweights, bins=100, stat="density", kde=True)
     plt.show()

examples/crbd.py

@@ -6,18 +6,16 @@
 import matplotlib.pyplot as plt
 import seaborn as sns
 
-
 alcedinidae = treeppl.Tree.load("trees/Alcedinidae.phyjson", format="phyjson")
 samples = None
-with treeppl.Model(filename="crbd.tppl", samples=10000, subsamples=10) as crbd:
+with treeppl.Model(
+    filename="crbd.tppl", m="smc-bpf", particles=10000, resample="align", subsample=True, subsample_size=10
+) as crbd:
     for i in range(1000):
         res = crbd(tree=alcedinidae)
-        samples = pd.concat([
-            samples,
-            pd.DataFrame({
-                "lambda": res.items(0), "mu": res.items(1), "lweight": res.norm_const
-            })
-        ])
+        samples = pd.concat(
+            [samples, pd.DataFrame({"lambda": res.items(0), "mu": res.items(1), "lweight": res.norm_const})]
+        )
         weights = np.exp(samples.lweight - samples.lweight.max())
         plt.clf()
         sns.kdeplot(data=samples, x="lambda", weights=weights)

examples/generative_crbd.py

@@ -3,13 +3,10 @@
 import treeppl
 from Bio import Phylo
 
+args = {"time": 5.0, "lambda": 1.0, "mu": 0.1}
 
-params = {"time": 5.0, "lambda": 1.0, "mu": 0.1}
-
-with treeppl.Model(filename="generative_crbd.tppl", samples=1) as generative_crbd:
-    result = generative_crbd(**params)
+with treeppl.Model(filename="generative_crbd.tppl", particles=1) as generative_crbd:
+    result = generative_crbd(args)
     tree = result.samples[0]
-    tree = Phylo.BaseTree.Clade(
-        branch_length=params["time"] - tree.age, clades=[tree.to_biopython()]
-    )
+    tree = Phylo.BaseTree.Clade(branch_length=args["time"] - tree.age, clades=[tree.to_biopython()])
     Phylo.draw(tree)

examples/treeppl_in_jupyter.ipynb

@@ -57,9 +57,7 @@
    "id": "a67a0418",
    "metadata": {},
    "source": [
-    "Once the extension is loaded, users can use the `%%treeppl` cell magic to write and compile a TreePPL program. Executing the cell creates a `treeppl.Model` object, which allows interaction with the compiled program directly in Python. This object is stored in a Python variable specified as an argument to the `%%treeppl` magic (i.e., immediately following `%%treeppl` on the first line). Optionally, the variable name can be followed by parameters and values. Most of these are passed to the TreePPL compiler, except for the `samples` parameter, which specifies the number of samples to draw when the program is executed. Examples are provided below.\n",
-    "\n",
-    "The extension also supports basic syntax highlighting for TreePPL programs.\n",
+    "Once the extension is loaded, users can use the `%%treeppl` cell magic to write and compile a TreePPL program. Executing the cell creates a `treeppl.Model` object, which allows interaction with the compiled program directly in Python. This object is stored in a Python variable specified as an argument to the `%%treeppl` magic (i.e., immediately following `%%treeppl` on the first line). The variable name can be followed by arguments and values, all of which are passed directly to the TreePPL compiler. Examples are provided below.\n",
     "\n",
     "For example, the following cell demonstrates a simple TreePPL program for simulating the flip of a fair coin:"
    ]
@@ -71,7 +69,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "%%treeppl flip samples=10\n",
+    "%%treeppl flip --particles 10\n",
     "\n",
     "model function flip() => Bool {\n",
     "  assume p ~ Bernoulli(0.5);\n",
@@ -84,9 +82,9 @@
    "id": "d71e9d44",
    "metadata": {},
    "source": [
-    "In this example, a `treeppl.Model` instance is created, and the program is compiled. The variable name `flip` (specified after `%%treeppl`) provides an interface for interacting with the model. The argument `samples=10` specifies the number of samples to generate when the program is executed.\n",
+    "In this example, a `treeppl.Model` instance is created, and the program is compiled. The variable name `flip` (specified after `%%treeppl`) provides an interface for interacting with the model. The argument `--particles 10` sets the number of particles used when executing the program. If no inference method is specified via the `--m` parameter, the default is importance sampling.\n",
     "\n",
-    "To run the TreePPL program, simply call the variable as a function (e.g., `flip()`). This executes the program and returns a `treeppl.InferenceResult` object, which includes a `samples` attribute containing the generated samples. While samples may have different weights in more complex programs, they are equally weighted in this simple example. We will cover programs with weighted samples later.\n",
+    "To run the TreePPL program, simply call the variable as a function (e.g., `flip()`). This executes the program and returns a `treeppl.InferenceResult` object, which includes a `samples` attribute containing the generated samples. While samples may have different weights in more complex programs, they are equally weighted in this simple example. Programs with weighted samples will be demonstrated later.\n",
     "\n",
     "Here’s an example of how to use the compiled program:"
    ]
@@ -125,7 +123,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "%%treeppl coin samples=100000\n",
+    "%%treeppl coin -m smc-bpf --particles 100000\n",
     "\n",
     "model function coin(outcomes: Bool[]) => Real {\n",
     "  assume p ~ Uniform(0.0, 1.0);\n",
@@ -169,7 +167,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "%%treeppl generative_crbd samples=1\n",
+    "%%treeppl generative_crbd --particles 1\n",
     "\n",
     "model function generativeCrbd(time: Real, lambda: Real, mu: Real) => Tree {\n",
     "  assume waitingTime ~ Exponential(lambda + mu);\n",
@@ -198,16 +196,16 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "params = {\n",
+    "args = {\n",
     "    \"time\": 5.0,\n",
     "    \"lambda\": 1.0,\n",
     "    \"mu\": 0.1\n",
     "}\n",
     "\n",
-    "result = generative_crbd(**params)\n",
+    "result = generative_crbd(args)\n",
     "tree = result.samples[0]\n",
     "tree = Phylo.BaseTree.Clade(\n",
-    "    branch_length=params[\"time\"] - tree.age,\n",
+    "    branch_length=args[\"time\"] - tree.age,\n",
     "    clades=[tree.to_biopython()]\n",
     ")\n",
     "Phylo.draw(tree)"
@@ -228,7 +226,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "%%treeppl crbd samples=10000 subsamples=10\n",
+    "%%treeppl crbd -m smc-bpf --particles 10000 --resample align --subsample --subsample-size 10\n",
     "\n",
     "function simulateExtinctSubtree(time: Real, lambda: Real, mu: Real) {\n",
     "  assume waitingTime ~ Exponential(lambda + mu);\n",
@@ -275,9 +273,7 @@
    "cell_type": "code",
    "execution_count": null,
    "id": "00bfd918",
-   "metadata": {
-    "scrolled": false
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "alcedinidae = treeppl.Tree.load(\"trees/Alcedinidae.phyjson\", format=\"phyjson\")\n",
@@ -306,9 +302,9 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "Python 3 (ipykernel)",
+   "display_name": "Python [conda env:base] *",
    "language": "python",
-   "name": "python3"
+   "name": "conda-base-py"
   },
   "language_info": {
    "codemirror_mode": {
@@ -320,7 +316,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.11.5"
+   "version": "3.13.9"
   }
  },
  "nbformat": 4,

setup.py

@@ -1,13 +1,13 @@
 from setuptools import setup
 
 setup(
-    name='treeppl',
-    version='0.1',
-    description='Python Interface to TreePPL',
-    author='Jan Kudlicka',
-    author_email='github@kudlicka.eu',
-    packages=['treeppl'],
+    name="treeppl",
+    version="0.1",
+    description="Python Interface to TreePPL",
+    author="Jan Kudlicka",
+    author_email="github@kudlicka.eu",
+    packages=["treeppl"],
     install_requires=[
-        'numpy',
+        "numpy",
     ],
 )

tests/test_all.py

@@ -5,23 +5,20 @@
 
 def test_compilation_error():
     with pytest.raises(treeppl.CompileError):
-        with treeppl.Model(
-            source="""\
+        with treeppl.Model(source="""\
 model incorrect() {
   return result;
 }
-""",
-            samples=1,
-        ) as model:
+""") as model:
             model()
 
 
 def test_coin():
-    nsamples = 100
-    with treeppl.Model(filename="examples/coin.tppl", samples=nsamples) as coin:
+    particles = 100
+    with treeppl.Model(filename="examples/coin.tppl", m="smc-bpf", particles=particles) as coin:
         res = coin(outcomes=(np.random.random(10) < 0.5).tolist())
         assert isinstance(res, treeppl.InferenceResult)
-        assert len(res.samples) == nsamples
+        assert len(res.samples) == particles
 
 
 def test_matrix_mul():
@@ -35,17 +32,15 @@ def test_matrix_mul():
   return a *@ b;
 }
 """,
-        samples=1,
+        particles=1,
     ) as matrix_mul:
         result = matrix_mul(a=a, b=b).samples[0]
     assert (result == expected).all()
 
 
 def test_tree_input():
     tree = treeppl.Tree.Node(
-        left=treeppl.Tree.Node(
-            left=treeppl.Tree.Leaf(age=0.0), right=treeppl.Tree.Leaf(age=0.0), age=0.5
-        ),
+        left=treeppl.Tree.Node(left=treeppl.Tree.Leaf(age=0.0), right=treeppl.Tree.Leaf(age=0.0), age=0.5),
         right=treeppl.Tree.Leaf(age=0.0),
         age=1.0,
     )
@@ -62,7 +57,7 @@ def test_tree_input():
   return count_leaves(tree);
 }
 """,
-        samples=1,
+        particles=1,
     ) as count_tree_leaves:
         res = count_tree_leaves(tree=tree)
         assert res.samples[0] == 3

treeppl/__init__.py

@@ -1,4 +1,4 @@
-from .base import Model, InferenceResult
+from .base import Model, InferenceResult, CompileArguments, RunArguments
 from .exceptions import CompileError, InferenceError
 from .serialization import Object, constructor
 from .stdlib import Tree