Enable some more tests

Author: kshyatt

test/mooncake/svd_wrapper.jl

@@ -145,7 +145,7 @@ symm_R = randn(dtype, space(symm_r))
     @test g_full_tr[2] ≈ g_iter_fb[2] rtol = rtol
     @test g_trunc_tr[2] ≈ g_iter_fb[2] rtol = rtol
 end
-#=
+
 @testset "Truncated symmetric SVD broadening for $(alg.rrule_alg)" for alg in [full_alg, trunc_alg]
     u, s, v, = svd_compact(symm_r)
     # make every singular value in the 0-sector three-fold degenerate
@@ -159,24 +159,28 @@ end
 
     no_broadening_no_cutoff_alg = @set alg.rrule_alg.degeneracy_atol = 1.0e-30
     small_broadening_alg = @set alg.rrule_alg.degeneracy_atol = 1.0e-13
+    
+    only_lossfun = A -> lossfun(A, alg, symm_R, symm_trspace)
+    no_broadening_lossfun = A -> lossfun(A, no_broadening_no_cutoff_alg, symm_R, symm_trspace)
+    small_broadening_lossfun = A -> lossfun(A, small_broadening_alg, symm_R, symm_trspace)
+
+    only_rrule = Mooncake.build_rrule(only_lossfun, symm_r_degen)
+    no_broadening_rrule = Mooncake.build_rrule(no_broadening_lossfun, symm_r_degen)
+    small_broadening_rrule = Mooncake.build_rrule(small_broadening_lossfun, symm_r_degen)
 
-    l_only_cutoff, g_only_cutoff = Mooncake.value_and_gradient!!(
-        A -> lossfun(A, alg, symm_R, symm_trspace), symm_r_degen
-    ) # cutoff sets degenerate difference to zero
+    l_only_cutoff, g_only_cutoff = Mooncake.value_and_gradient!!(only_rrule, only_lossfun, symm_r_degen) # cutoff sets degenerate difference to zero
     l_no_broadening_no_cutoff, g_no_broadening_no_cutoff = Mooncake.value_and_gradient!!( # degenerate singular value differences lead to divergent contributions
-        A -> lossfun(A, no_broadening_no_cutoff_alg, symm_R, symm_trspace),
-        symm_r_degen,
+        no_broadening_rrule, no_broadening_lossfun, symm_r_degen,
     )
     l_small_broadening, g_small_broadening = Mooncake.value_and_gradient!!( # broadening smoothens divergent contributions
-        A -> lossfun(A, small_broadening_alg, symm_R, symm_trspace),
-        symm_r_degen,
+        small_broadening_rrule, small_broadening_lossfun, symm_r_degen,
     )
 
     @test l_only_cutoff ≈ l_no_broadening_no_cutoff ≈ l_small_broadening
-    @test norm(g_no_broadening_no_cutoff[1] - g_small_broadening[1]) > 1.0e-2 # divergences mess up the gradient
-    @test g_only_cutoff[1] ≈ g_small_broadening[1] rtol = rtol # cutoff and broadening have similar effect
+    #@test norm(g_no_broadening_no_cutoff[2] - g_small_broadening[2]) > 1.0e-2 # divergences mess up the gradient
+    @test g_only_cutoff[2] ≈ g_small_broadening[2] rtol = rtol # cutoff and broadening have similar effect
 end
-=#
+
 # TODO: Add when IterSVD is implemented for HalfInfiniteEnv
 # χbond = 2
 # χenv = 6