From ddbcba30cd48a35f210daa67139c7ce21fe28765 Mon Sep 17 00:00:00 2001
From: odow <o.dowson@gmail.com>
Date: Sun, 14 Jul 2024 12:20:11 +1200
Subject: [PATCH 1/4] [Nonlinear.ReverseAD] fix NLPBlock and final_touch
 bridges

---
 src/Nonlinear/ReverseAD/reverse_mode.jl | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)

diff --git a/src/Nonlinear/ReverseAD/reverse_mode.jl b/src/Nonlinear/ReverseAD/reverse_mode.jl
index ca7fa1e0f8..9b09a74a82 100644
--- a/src/Nonlinear/ReverseAD/reverse_mode.jl
+++ b/src/Nonlinear/ReverseAD/reverse_mode.jl
@@ -54,6 +54,26 @@ function _reverse_mode(d::NLPEvaluator, x)
     for con in d.constraints
         _reverse_eval(con)
     end
+    # If a JuMP model uses the legacy nonlinear interface, then JuMP constructs
+    # a NLPEvaluator at the start of a call to `JuMP.optimize!` and it passes in
+    # the list of variables in the JuMP model to `.ordered_variables`.
+    #
+    # During `MOI.initialize`, `.last_x` gets filled with `NaN` to match the
+    # length of `ordered_variables`.
+    #
+    # However, if the model includes a bridge that adds new decision variables
+    # in `Utilities.final_touch`, then the total number of variables (in `x`)
+    # might be larger than the cache in `last_x`.
+    #
+    # It is safe to resize `last_x` because only the variables in
+    # `ordered_variables` can appear in the NLPBlock.
+    #
+    # I don't think we need any other fixes becasue callers to things like
+    # `eval_objective` can pass in a longer input `x` vector without fear
+    # because the excess elements won't be used.
+    if length(d.last_x) < length(x)
+        resize!(d.last_x, length(x))
+    end
     copyto!(d.last_x, x)
     return
 end

From a3654f27e73fd8b9d6e066187135097254f65755 Mon Sep 17 00:00:00 2001
From: odow <o.dowson@gmail.com>
Date: Sun, 14 Jul 2024 15:55:22 +1200
Subject: [PATCH 2/4] Update

---
 test/Nonlinear/ReverseAD.jl | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/test/Nonlinear/ReverseAD.jl b/test/Nonlinear/ReverseAD.jl
index 9f3cc632fd..8eb597f873 100644
--- a/test/Nonlinear/ReverseAD.jl
+++ b/test/Nonlinear/ReverseAD.jl
@@ -1121,6 +1121,20 @@ function test_timers()
     return
 end
 
+function test_varying_length_x()
+    model = MOI.Nonlinear.Model()
+    x = MOI.VariableIndex(1)
+    MOI.Nonlinear.set_objective(model, :(sin($x)))
+    evaluator =
+        MOI.Nonlinear.Evaluator(model, MOI.Nonlinear.SparseReverseMode(), [x])
+    MOI.initialize(evaluator, Symbol[:Jac])
+    ∇f = [NaN]
+    MOI.eval_objective_gradient(evaluator, ∇f, [1.0, 2.0])
+    @test length(∇f) == 1
+    @test ∇f[1] ≈ cos(1.0)
+    return
+end
+
 end  # module
 
 TestReverseAD.runtests()

From 3494ae1e38d3943b1df55b308a66f2c1f7e655a5 Mon Sep 17 00:00:00 2001
From: Oscar Dowson <odow@users.noreply.github.com>
Date: Sun, 14 Jul 2024 17:44:33 +1200
Subject: [PATCH 3/4] Update src/Nonlinear/ReverseAD/reverse_mode.jl

---
 src/Nonlinear/ReverseAD/reverse_mode.jl | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/Nonlinear/ReverseAD/reverse_mode.jl b/src/Nonlinear/ReverseAD/reverse_mode.jl
index 9b09a74a82..b1ea0f9773 100644
--- a/src/Nonlinear/ReverseAD/reverse_mode.jl
+++ b/src/Nonlinear/ReverseAD/reverse_mode.jl
@@ -68,7 +68,7 @@ function _reverse_mode(d::NLPEvaluator, x)
     # It is safe to resize `last_x` because only the variables in
     # `ordered_variables` can appear in the NLPBlock.
     #
-    # I don't think we need any other fixes becasue callers to things like
+    # I don't think we need any other fixes because callers to things like
     # `eval_objective` can pass in a longer input `x` vector without fear
     # because the excess elements won't be used.
     if length(d.last_x) < length(x)

From 1044004772738d82c51bdbcaa0a972b746bf0782 Mon Sep 17 00:00:00 2001
From: Oscar Dowson <odow@users.noreply.github.com>
Date: Mon, 15 Jul 2024 11:02:51 +1200
Subject: [PATCH 4/4] Update reverse_mode.jl

---
 src/Nonlinear/ReverseAD/reverse_mode.jl | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/src/Nonlinear/ReverseAD/reverse_mode.jl b/src/Nonlinear/ReverseAD/reverse_mode.jl
index b1ea0f9773..5487da2237 100644
--- a/src/Nonlinear/ReverseAD/reverse_mode.jl
+++ b/src/Nonlinear/ReverseAD/reverse_mode.jl
@@ -59,11 +59,12 @@ function _reverse_mode(d::NLPEvaluator, x)
     # the list of variables in the JuMP model to `.ordered_variables`.
     #
     # During `MOI.initialize`, `.last_x` gets filled with `NaN` to match the
-    # length of `ordered_variables`.
+    # length of `ordered_variables`, that is, the number of variables in the
+    # JuMP model.
     #
     # However, if the model includes a bridge that adds new decision variables
-    # in `Utilities.final_touch`, then the total number of variables (in `x`)
-    # might be larger than the cache in `last_x`.
+    # then the total number of variables in the optimizer (in `x`) will be
+    # larger than the cache in `last_x`.
     #
     # It is safe to resize `last_x` because only the variables in
     # `ordered_variables` can appear in the NLPBlock.