docs and const arr

src/solver/postprocess.jl

@@ -104,8 +104,7 @@ function _postprocess_root(x::SymbolicUtils.BasicSymbolic)
         end
     end
 
-    trig_simplified = check_trig_consts(x)
-    !isequal(trig_simplified, x) && return trig_simplified
+    x = convert_consts(x)
 
     if oper === (+)
         args = arguments(x)
@@ -136,28 +135,49 @@ function postprocess_root(x)
     x # unreachable
 end
 
-function check_trig_consts(x)
-    !iscall(x) && return x
 
-    oper = operation(x)
-    inv_opers = [asin, acos, atan]
-    inv_exacts = [0, Symbolics.term(*, pi),
+inv_exacts = [0, Symbolics.term(*, pi),
         Symbolics.term(/,pi,3),
         Symbolics.term(/, pi, 2),
         Symbolics.term(/, Symbolics.term(*, 2, pi), 3),
         Symbolics.term(/, pi, 6),
         Symbolics.term(/, Symbolics.term(*, 5, pi), 6),
         Symbolics.term(/, pi, 4)
-    ]
+]
+inv_evald = Symbolics.symbolic_to_float.(inv_exacts)
+
+const inv_pairs = collect(zip(inv_exacts, inv_evald))
+"""
+    function convert_consts(x)
+This function takes BasicSymbolic terms as input (x) and attempts
+to simplify these basic symbolic terms using known values.
+Currently, this function only supports inverse trignometric functions.
+
+## Examples
+```jldoctest
+julia> Symbolics.convert_consts(Symbolics.term(acos, 0))
+π / 2
+
+julia> Symbolics.convert_consts(Symbolics.term(atan, 0))
+0
+
+julia> Symbolics.convert_consts(Symbolics.term(atan, 1))
+π / 4
+```
+"""
+function convert_consts(x)
+    !iscall(x) && return x
+
+    oper = operation(x)
+    inv_opers = [asin, acos, atan]
 
     if any(isequal(oper, o) for o in inv_opers) && isempty(Symbolics.get_variables(x))
         val = Symbolics.symbolic_to_float(x)
-        for i in eachindex(inv_exacts)
-            exact_val = Symbolics.symbolic_to_float(inv_exacts[i])
-            if isapprox(exact_val, val, atol=1e-6)
-                return inv_exacts[i]
-            elseif isapprox(-exact_val, val, atol=1e-6)
-                return -inv_exacts[i]
+        for (exact, evald) in inv_pairs
+            if isapprox(evald, val)
+                return exact
+            elseif isapprox(-evald, val)
+                return -exact
             end
         end
     end