cannot use SymPyPythonCall.↓

[thorek1]

I would like to avoid:

using SymPyPythonCall

or

import SymPyPythonCall: ↓

my preferred solution which currently does not work would be:

import SymPyPythonCall as SPPC
SPPC.@syms x
SPPC.↓(x)
# ERROR: UndefVarError: `↓` not defined

I don't know if this is Julia specific that you cannot use unicode symbols in this way or if it is SymPyPythonCall

Either way, it would be nice to have my preferred solution working as well :)

[jverzani]

The issue is SPPC.↓ doesn't parse right, but SPPC.:↓ does:

import SymPyPythonCall as SPPC
SPPC.@syms x
SPPC.:↓(x)

Now, let me ask, are you using this downarrow function for anything? It really is intended as an escape hatch when some overrides don't work, not as a daily driver. If you find yourself reaching for it often, it seems like I should be adding a convenience. (Which to some level, this package is all about.)

[thorek1]

thanks for the hint. that solved it.

I use it to have a common type on which I can do set operations.

I have one object (vars_to_solve ) of type SymPyCore.Sym from this part of my code:

    for pos in symbols_pos
        eval(:($pos = SPyPyC.symbols($(string(pos)), real = true, finite = true, positive = true)))
    end
    for neg in symbols_neg
        eval(:($neg = SPyPyC.symbols($(string(neg)), real = true, finite = true, negative = true)))
    end
    for none in symbols_none
        eval(:($none = SPyPyC.symbols($(string(none)), real = true, finite = true)))
    end

    Set(eval(:([$(𝓂.vars_in_ss_equations...)])))

and another object (soll_free_syms ) of type SymPyCore.Sym{PythonCall.Py} from this part of the code:

soll = SPyPyC.solve(eqs_to_solve,vars_to_solve)
soll_free_syms = union(SPyPyC.free_symbols.(soll[1])...)

and with those I want to do:

intersect(soll_free_syms, vars_to_solve)
# Internal Error: StackOverflowError:

instead this works:

intersect((soll_free_syms .|> SPyPyC.:↓), (vars_to_solve .|> SPyPyC.:↓))

[jverzani]

Hmm, this shouldn't have a stack overflow (intersect(soll_free_syms, vars_to_solve)). If you had an MWE to share I'd be most appreciative.

[thorek1]

here is the essence of the code (which works fine):

import SymPyPythonCall as SPPC
symbols_pos = [:Pibar, :phi_pi, :beta]

for pos in symbols_pos
    eval(:($pos = SPPC.symbols($(string(pos)), real = true, finite = true, positive = true)))
end

vars_to_solve = Set(eval(:([$(symbols_pos...)])))

eqs_to_solve = Pibar^(phi_pi/(phi_pi-1))/beta

soll = SPPC.solve(eqs_to_solve,collect(vars_to_solve)[2])

soll_free_syms = SPPC.free_symbols.(soll[1])

intersect(vars_to_solve,soll_free_syms)

what is missing is that all of this is set up within a macro and I don't have time to include also that component now. I am afraid that is where he gets confused for reasons beyond me. I hope this helps

[jverzani]

After changing collect(vars_to_solve)[2] to collect(vars_to_solve)[1] it all ran as expected. If/when you get time to run the whole thing, you can likely just insert some debug code just before the call to intersect (@show vars_to_solve, soll_free_syms) and that will hopefully point the right direction to try an adjustment.