cumsum error

[weymouth]

I'm just trying to make a function for the arclength along a space-curve.

curve(u) = [1.,0.,cos(u)]
dist = arclength(curve)
speed = Fun(u->hypot(derivative(curve,u)...),0..pi)
sum(speed) # 2.0 works!
cumsum(speed)(pi) # 0.0 doesn't work!

[jishnub]

Does pi lie within the domain? I suspect this isn't the case because of rounding

[weymouth]

I specify the interval to be 0..pi.

1. I assumed the interval was closed, but maybe the periodic nature of the function is a problem? Is this something that can be inferred from the speed object?
2. If you evaluate "outside" the domain, is the correct response to throw an error, or return a random(?) value?

[jishnub]

Currently, evaluating a Fun outside the domain returns 0. This is a questionable choice that leads to a lot of confusion, and maybe it's better to throw an error.
We also have

julia> pi in 0..pi
false

As the endpoint of the interval isn't exactly pi.

[weymouth]

Ha. That would be the issue! I guess that's a floating point thing?

…

On Sun, May 25, 2025, 18:22 Jishnu Bhattacharya ***@***.***> wrote: *jishnub* left a comment (JuliaApproximation/ApproxFun.jl#955) <#955 (comment)> Currently, evaluating a Fun outside the domain returns 0. This is a questionable choice that leads to a lot of confusion, and maybe it's better to throw an error. We also have julia> pi in 0..pifalse As the endpoint of the interval isn't exactly pi. — Reply to this email directly, view it on GitHub <#955 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AADSKJ7LL3IMJYFAQCZROJD3AHU57AVCNFSM6AAAAAB5NUIO6OVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDSMBXHEZDINZWGE> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[jishnub]

Yes, indeed

[weymouth]

Can/should that be fixed?