Kirill Zubov

It looks like the problem is that methods crash if it doesn't converge enough (to `reltol and abstol` )and it ends on the `maxiters`. Since it all works with large...

@vaishnavtv NeuralPDE on this moment doesn't work with vector input `f([xV, xα]) ` `Differential` work only with neural network(NN) input ```julia @parameters xV @variables u(..) Differential(xV)(u(xV)) ``` where u is...

if it is symbolic form, ModelingToolkit handles it: ```julia @parameters x @variables u(..) D = Differential(x) f(x) = x^2 julia> eq_l = D(u(x) + fx(x)) Differential(x)(u(x) + x^2) julia> expand_derivatives(eq_l)...

NeuralPDE parser issue with derivative of @register's function ```julia @parameters x @variables u(..) D = Differential(x) f(x) = x^2 @register f(x) eq = D(u(x) + f(x)) ~ 0. expand_derivatives(eq) bcs...

https://github.com/JuliaGizmos/Interact.jl

@oashour you need `@register_symbolic` https://symbolics.juliasymbolics.org/dev/manual/functions/ ```julia @parameters t x @variables c₁(..) c₂(..) rand_f(x) = rand(size(x)...) @register_symbolic rand_f(x) Base.Broadcast.broadcasted(::typeof(rand_f), x) = rand_f(x) vec_ = ones(1,10) rand_f(vec_) """ julia> rand_f(vec_) 1×10 Matrix{Float64}:...

@ChrisRackauckas do you mean something like this algorithm? ``` for every x in x_vector create a random vector - rand(N), N - size of vector xi ~ rand(N) vector of...

why do we use random numbers how initial conditions? is it an unknown initial condition and we try to pick one from a random set which is better? sort of...

that is, we initialize random vectors `r1, r2` once and use them without generating new ones at each iteration?

ok, got it