Revert

src/solvers/insert/insert.jl

@@ -46,72 +46,4 @@ function default_inserter(
   state[v] *= phi
   state = set_ortho_region(state, [v])
   return state, nothing
-end
-
-function bp_inserter(
-  ψ::AbstractITensorNetwork,
-  ψAψ_bpcs::Vector{<:BeliefPropagationCache},
-  ψIψ_bpc::BeliefPropagationCache,
-  state::ITensor,
-  region;
-  cache_update_kwargs=(;),
-  kwargs...,
-)
-  spec = nothing
-
-  form_network = unpartitioned_graph(partitioned_tensornetwork(ψIψ_bpc))
-  ψAψ_bpcs = BeliefPropagationCache[reset_messages(ψAψ_bpc) for ψAψ_bpc in ψAψ_bpcs]
-  ψIψ_bpc = reset_messages(ψIψ_bpc)
-  @show messages(ψIψ_bpc)
-  if length(region) == 1
-    states = [state]
-  elseif length(region) == 2
-    v1, v2 = region[1], region[2]
-    e = edgetype(ψ)(v1, v2)
-    pe = partitionedge(
-      ψIψ_bpc, ket_vertex(form_network, v1) => bra_vertex(form_network, v2)
-    )
-    stateᵥ₁, stateᵥ₂, spec = factorize_svd(
-      state, uniqueinds(ψ[v1], ψ[v2]); ortho="none", tags=edge_tag(e), kwargs...
-    )
-    states = noprime.([stateᵥ₁, stateᵥ₂])
-    #TODO: Insert spec into the message tensor guess here?!
-  end
-
-  for (i, v) in enumerate(region)
-    state = states[i]
-    state_dag = copy(state)
-    form_bra_v, form_ket_v = bra_vertex(form_network, v), ket_vertex(form_network, v)
-    ψ[v] = state
-    state_dag = replaceinds(
-      dag(state_dag), inds(state_dag), dual_index_map(form_network).(inds(state_dag))
-    )
-    ψAψ_bpcs = BeliefPropagationCache[
-      update_factor(ψAψ_bpc, form_ket_v, state) for ψAψ_bpc in ψAψ_bpcs
-    ]
-    ψAψ_bpcs = BeliefPropagationCache[
-      update_factor(ψAψ_bpc, form_bra_v, state_dag) for ψAψ_bpc in ψAψ_bpcs
-    ]
-    ψIψ_bpc = update_factor(ψIψ_bpc, form_ket_v, state)
-    ψIψ_bpc = update_factor(ψIψ_bpc, form_bra_v, state_dag)
-  end
-
-  ψAψ_bpcs = BeliefPropagationCache[
-    update(ψAψ_bpc; cache_update_kwargs...) for ψAψ_bpc in ψAψ_bpcs
-  ]
-
-  ψIψ_bpc = update(ψIψ_bpc; cache_update_kwargs...)
-
-  updated_ψIψ = unpartitioned_graph(partitioned_tensornetwork(ψIψ_bpc))
-  numerator_terms = [
-    scalar(
-      unpartitioned_graph(partitioned_tensornetwork(ψAψ_bpc));
-      (cache!)=Ref(ψAψ_bpc),
-      alg="bp",
-    ) for ψAψ_bpc in ψAψ_bpcs
-  ]
-  eigval = sum(numerator_terms) / scalar(updated_ψIψ; (cache!)=Ref(ψIψ_bpc), alg="bp")
-  @show eigval
-  @show scalar(only(ψAψ_bpcs); alg="exact") / scalar(ψIψ_bpc; alg="exact")
-  return ψ, ψAψ_bpcs, ψIψ_bpc, spec, (; eigvals=[eigval])
-end
+end