Support mit-mots in the JAX backend

Author: rlouf

aesara/link/jax/dispatch/scan.py

@@ -13,6 +13,7 @@
 def jax_funcify_Scan(op, node, **kwargs):
     scan_inner_fn = jax_funcify(op.fgraph)
     input_taps = {
+        "mit_mot": op.info.mit_mot_in_slices,
         "mit_sot": op.info.mit_sot_in_slices,
         "sit_sot": op.info.sit_sot_in_slices,
         "nit_sot": op.info.sit_sot_in_slices,
@@ -32,9 +33,6 @@ def parse_outer_inputs(outer_inputs):
             "shared": list(op.outer_shared(outer_inputs)),
             "non_sequences": list(op.outer_non_seqs(outer_inputs)),
         }
-        if len(outer_in["mit_mot"]) > 0:
-            raise NotImplementedError("mit-mot not supported")
-
         return outer_in
 
     if op.info.as_while:
@@ -323,7 +321,7 @@ def build_jax_scan_inputs(outer_in: Dict):
         sequences = outer_in["sequences"]
         init_carry = {
             name: outer_in[name]
-            for name in ["mit_sot", "sit_sot", "shared", "non_sequences"]
+            for name in ["mit_mot", "mit_sot", "sit_sot", "shared", "non_sequences"]
         }
         init_carry["step"] = 0
         return n_steps, sequences, init_carry
@@ -340,7 +338,7 @@ def build_inner_outputs_map(outer_in):
         [+ while-condition]
 
         """
-        inner_outputs_names = ["mit_sot", "sit_sot", "nit_sot", "shared"]
+        inner_outputs_names = ["mit_mot", "mit_sot", "sit_sot", "nit_sot", "shared"]
 
         offset = 0
         inner_output_idx = defaultdict(list)
@@ -415,6 +413,9 @@ def scan_inner_in_args(carry, x):
             current_step = carry["step"]
 
             inner_in_seqs = x
+            inner_in_mit_mot = from_carry_storage(
+                carry["mit_mot"], current_step, input_taps["mit_mot"]
+            )
             inner_in_mit_sot = from_carry_storage(
                 carry["mit_sot"], current_step, input_taps["mit_sot"]
             )
@@ -427,6 +428,7 @@ def scan_inner_in_args(carry, x):
             return sum(
                 [
                     inner_in_seqs,
+                    inner_in_mit_mot,
                     inner_in_mit_sot,
                     inner_in_sit_sot,
                     inner_in_shared,
@@ -439,6 +441,7 @@ def scan_new_carry(carry, inner_outputs):
             """Create a new carry value from the values returned by the inner function (inner-outputs)."""
             step = carry["step"]
             new_carry = {
+                "mit_mot": [],
                 "mit_sot": [],
                 "sit_sot": [],
                 "shared": [],
@@ -452,6 +455,14 @@ def scan_new_carry(carry, inner_outputs):
                 ]
                 new_carry["shared"] = shared_inner_outputs
 
+            if "mit_mot" in inner_output_idx:
+                mit_mot_inner_outputs = [
+                    inner_outputs[idx] for idx in inner_output_idx["mit_mot"]
+                ]
+                new_carry["mit_mot"] = to_carry_storage(
+                    mit_mot_inner_outputs, carry["mit_mot"], step, input_taps["mit_mot"]
+                )
+
             if "mit_sot" in inner_output_idx:
                 mit_sot_inner_outputs = [
                     inner_outputs[idx] for idx in inner_output_idx["mit_sot"]
@@ -486,6 +497,10 @@ def scan_new_outputs(inner_outputs):
 
             """
             outer_outputs = []
+            if "mit_mot" in inner_output_idx:
+                outer_outputs.append(
+                    [inner_outputs[idx] for idx in inner_output_idx["mit_mot"]]
+                )
             if "mit_sot" in inner_output_idx:
                 outer_outputs.append(
                     [inner_outputs[idx] for idx in inner_output_idx["mit_sot"]]

tests/link/jax/test_scan.py

@@ -3,7 +3,7 @@
 from packaging.version import parse as version_parse
 
 import aesara.tensor as at
-from aesara import function
+from aesara import function, grad
 from aesara.compile.mode import Mode
 from aesara.configdefaults import config
 from aesara.graph.fg import FunctionGraph
@@ -23,6 +23,7 @@
 # Disable all optimizations
 opts = RewriteDatabaseQuery(include=[None], exclude=["cxx_only", "BlasOpt"])
 jax_mode = Mode(JAXLinker(), opts)
+py_mode = Mode("py", opts)
 
 
 def test_while_cannnot_use_all_outputs():
@@ -66,8 +67,8 @@ def test_sit_sot():
         n_steps=3,
     )
 
-    jax_fn = function((a_at,), res, updates=updates, mode="JAX")
     fn = function((a_at,), res, updates=updates)
+    jax_fn = function((a_at,), res, updates=updates, mode="JAX")
     assert np.allclose(fn(1.0), jax_fn(1.0))
 
 
@@ -413,3 +414,29 @@ def power_step(prior_result, x):
 
     for output_jax, output in zip(jax_res, res):
         assert np.allclose(jax_res, res)
+
+
+def test_mitmots_basic():
+
+    init_x = at.dvector()
+    seq = at.dvector()
+
+    def inner_fct(seq, state_old, state_current):
+        return state_old * 2 + state_current + seq
+
+    out, _ = scan(
+        inner_fct, sequences=seq, outputs_info={"initial": init_x, "taps": [-2, -1]}
+    )
+
+    g_outs = grad(out.sum(), [seq, init_x])
+
+    out_fg = FunctionGraph([seq, init_x], g_outs)
+
+    seq_val = np.arange(3)
+    init_x_val = np.r_[-2, -1]
+    (seq_val, init_x_val)
+
+    fn = function(out_fg.inputs, out_fg.outputs)
+    jax_fn = function(out_fg.inputs, out_fg.outputs, mode="JAX")
+    print(fn(seq_val, init_x_val))
+    print(jax_fn(seq_val, init_x_val))