The cascaded "local green" is not compatible with the already "assigned red". It creates "shadow in the context" ... Argument doesn't match the expected kind. The k:=k eliminates the "shadow in the context". With the "local green" and considering the Σ expansion, 'k' is computed only once and placed in the Σ stack. Thus, f(x) in the solver runs an 'f stack' already constructed. The "local green" runs the solver in 6.5 sec. ============================================ The "assigned red" is a much different story. Even with k:=k, 'k' is not local unless it is isolated [yellow]. What happens with the "assigned red", the 'f stack' reconstructs itself 36 times accordingly to the solve f(x) thus consuming approximately 36*6 sec=4.5 min =========================================== These observations pertain to Smath 5346. The "red local companion" reduces timing down to 2 min. The retained construction is the ............. => red star <= ...........

Acceptable but not flexible

Explain things in disorder, as it comes to mind: 1. expansion of f(Φ) Σ [only 5 terms] is immensely long ... "immensely long" but just minuscule compared to the 'ksmooth' algorithm . 2. on the range of Φ [i.e: 'x' in the solver], the f(Φ) range is rounded 1.7 ... -1.8 in order to get visual 'Sol'. 3. On that range of 36 demanded solutions, "solve" accepts each recursive solving cycle to start from '0'. However, the end range of the solver must exceed the upper value '1' of the 'Φ' ... 1.01 is typical OK. Here, the rule is "trial/error": must work in minimum time. 4. Inverse(f,L,H,N) is algo style wrt 'f'. The solver will cycle 36 times wrt the incremented ω. Also, it will run 36 times f(x) as we are solving for 'x', where each 'x' depends upon the 36 running 'ω'.