longemen3000
longemen3000 on Hacker News
Hello,
The textbook seems nice and clear. The only nitpick i have is that it should talk more about equations of state. I understand that it may not be the focus of the text, but mentioning the current state of equations of state (SAFTs, cubics, multiparameter) would help guide readers looking on how to generate their own steam tables for their fluid of interest, even if the advice is just "go use CoolProp"
On the other hand, i really like the ilustrations on turbomachinery, helps ground the theoretical content.
That is one thing i like about the Julia package ecosystem. The general Registry (where package metadata is stored and where you go to register a new package), recommends using explicit names over short acronyms. For example, DifferentialEquations.jl is a package that does differential equations in julia (recognizable via the .jl suffix). What does Garlic.jl do? Exactly, garlic (the vegetable) modelling.
It seems like this uses DWSIM under the hood for the simulations? Cool project
a 3-meter high wall of water is not really good either, due to kinetic energies, if no significant tsunami coastal protections, that water flow can reach far inland
i don't use Erlang (mainly Julia), but this change makes a lot of sense. in Julia, there is the == operator (value equality) and the === operator (value egality). 0.0 == -0.0,because they have the same "value". but they aren't identical (the bits are different), so !(0.0 === -0.0).
tianqi sold all his shares of SQM (one of Chile lithium companies, around 23%) after the announcement.
The situation of the lithium in Chile is very particular. in the Pinochet dictatorship, litium was nationalized, because it was a key ingredient of atomic bombs (and not much else at that time), and USA didn't want other countries to access the resource, so all current mines are in a lease contract (Albemarle until 2043, SQM until 2030). the new proposition opens new zones for exploitation, in exchange of the public-private parnership
"Fun" fact about that, it was nationalized because it is a material for use in atomic bombs, in the 70 and 80 there wasn't another use for large scale lithium mining, except for that
well, thermo is from the greek word thérmē meaning heat. we use the termo- suffix for all things that involve heat (thermodynamics -> termodinámica, thermometer -> termómetro)
I suppose that Twitter has to pay each time a SMS is sent (ar at least, they pay according to the amount of SMS generated), so the bots pump that amount?
for instagram, i just gave up looking at the feed and just look at the posts that a group chat sends
yeah, my systems are really small in comparison (1-20) but with higher order derivatives (up to 4th order), so reverse AD is not the best in that regard
I feel called out on the academic part hahaah. I simply want to code state of the art (thermodynamic) models, and at least julia helps by providing easy testing and publishing infraestructure. but obviously we can't compete with a corporation in code quality (we are trying!)
Unrelated, but for small sizes, i really prefer to use forward mode in julia (Via ForwardDiff.jl) instead of Zygote. the overhead of reverse ADing over an arbitrary function with mutation is not worth it.
Of course! we are urgently in need of a "critical boiling" relation. At the moment we are using critical isochores as a way to "extend" the saturation curve, but it is less than ideal (at least they can be calculated quickly).
I agree with the Widom line, at the moment i haven't found any concrete relation to calculate those from a equation of state functional. if there is something concrete, i would be glad to code it.
We have a solver for the pure critical point, that just finds the (v,T) coordinate that satisfies dP/dV = 0 = d²P/dV². of course, not all EoS can even satisfy this condition. Cubics have the correct temperature and pressure (by definition) but fail on volume. SAFT equations and all EoS with association terms predicts critical points with temperatures higher than the experimental ones. a lot of MultiParameter EoS (like IAWPS95, the standard for pure water) have some anomalies at the vicinity of the critical point. Nevertheless, finding such critical points, however far from reality are from the Real fluid that is trying to represent, helps on finding where are your properties in the phase diagram. In practice, you never stay at the critical point, you pass near it and you can tell by some properties that you are passing near it.
There is also new developments on "Phase equilibria" over the critical point,because it seems that supercritical fluids do present some sort of continuous phase transition between a liquid-like state and a gas-like state. there is this thing named the widom line that marks this separation, But not all EoS even present this line..., "are there phases in the supercritical fluid?" is a research question in that sense.
A lot of new work on supercritical fluids is the ability of calculate properties for those. normally we don't access to those conditions, so it is tricky to measure, but there is always new work on correlations capable of explaining the anomalies present on near critical and supercritical fluids.
On the viscosity front, there is this new framework named Entropy Scaling, that postulates that scaled viscosity ≈ f(residual entropy)^(2/3).
There is also new work on calculation of interfacial tensions from pure equations of state,that are relevant because you could have a two phase liquid-liquid mix and you want to transport something from one liquid phase to another.
Also, as the article says, over the critical point, there is no surface tension, because there is only one phase. (do not take this phrase literally, because that only happens in a pure compound, as soon as you have a mixture, the surface tension will depend on all elements of the mixture)
Furthermore, some fluid mixtures present something called UCST (upper critical solution temperatures) and LCST (lower critical solution temperatures). imagine if you lower the temperature of and oil-water mixture and suddenly they start to mix. and if you keep lowering the temperature, they unmix again!, and of course, not every fluid presents those properties, but we need to understand those who do. and specially, calculate those points. Few if any industrial thermodynamic calculators can do this, because you require fourth order accurate derivatives (can this be solved with automatic differentiation like what they do with neural networks?, of course, but you need a differentiation-ready equation of state)
On more macro properties, supercritical cycles are useful. you can start from a liquid and end with a gas without boiling, if you pressurize your liquid over the critical pressure (isothermically) then lower the temperature (isobarically) and finally releasing pressure (isothermically). one advantage of doing that is that you don't produce bubbles or any kinetic artifacts that are present during a normal phase transition.
i'm working on calculating thermodynamic properties from equations of state [1], and yeah, looking for criteria on how to distinguish from a low pressure, a near critical pressure and a supercritical pressure is not trivial, specially when you don't know the critical point of a fluid (the case for all molecular EoS). one property i've relying on is that the second virial coefficient minimum density is always lower than the saturated gas density.. until a threshold near criticality. a lot of assumptions about a thermodynamic model fail at near-critical points. Understanding and exploiting those properties helps on building more accurate property solvers for any, existing or new, equations of state.
[1] https://github.com/ypaul21/Clapeyron.jl
In Julia at least, there are packages that provide alternatives to Bigfloats:
Logarithmic numbers: more range, less precision
https://github.com/cjdoris/LogarithmicNumbers.jl
Double arithmetic: stitch 2 floats together
https://github.com/JuliaMath/DoubleFloats.jl
Of course, if you are calling BLAS/LAPACK, you are constrained to use floats, but the recommendation on DoubleFloats is clear: if you know you algorithms, use the increased precision only in the parts that matter
I like the discrepancy of technology presented in that counterexample. Euler definitely didn't Have that amount of computing power to just do a naive search,now we can even do those calculations on home computers
i just want to know, do they really use the name "monkey puzzle tree" in english speaking countries? i was a little surprised to see that name, given that araucaria is the accepted name for the tree (in spanish).
If we dig deeper in the rabbit hole, araucaria is the name given to the tree (and the zone where the tree grows, and the indigenous (mapuche) people that inhabited, and keeps inhabiting the land) by the spanish empire.
The original name (newen) was given by the mapuche people. "piñon" is also a mapuche word and while we use "araucaria" when refering to the tree, we use "piñón" instead of "araucaria nut". to refer to the edible part.
just to point out, the Araucaria Araucana gives "piñón" (a type of pine nut, but not the same). they can be made into flour or be eaten after boiling.
There is a very useful function in Julia, named Base.prompt (I don't remember if it's exported or not, but better be safe
height_str=Base.prompt("Height above sea level(m): ")
height =Parse(Float64, height_str)