Well, there are papers claiming that there’s a placental microbiome. I don’t buy it. I’m not aware of any link between maternal microbial exposures and child health outcomes, other than the obvious deleterious ones like infections.
We’re focusing on early life brain development since there’s so much development happening, we’re more likely to be able to see effects if they’re there. I wouldn’t be surprised if your gut microbes continue to have effects throughout life (I’ve seen studies on things like depression, to I’m not sure how convincing they are), but I’d be surprised if the effect sizes were very large, and therefore I’d be surprised if anyone could see the effects with the methods and typical clinical cohorts we have.
Best diet advice I can give is to eat a lot of fiber, but that’s more for inflammation than cognition.
Julia has given me a path back to science, my first love, after a career spent in software.
Julia is expressive and performant. This is a great combination because it means that the interesting bits are neither farmed out to C nor distorted to fit into the language’s syntax. The interesting bits aren’t in a mysterious black box, they’re accessible and understandable. That’s a godsend for me because having an accessible implementation makes it easier for me to learn the science.
Julia’s been a sort of gateway drug. It got me started. I could go further into the science than otherwise. When I reached the limit of what I could accomplish with the math I knew, the community helped me get a little bit further with resources that let me flank the problems I couldn’t attack head on. Buoyed by that success, I resolved to put in the work to learn the math, which I’m doing now.
I still have much to learn, but it’s been very satisfying to be able to pick up an academic paper and understand (at least some of) the math!
So, it’s not so much that I’m doing something interesting with Julia, but that it’s making my life more interesting. I’m really grateful.
I’m using Julia for computer vision, more specifically 2D and 3D pose estimation with the goal of real-time 3D motion tracking. Some early results from an image -> 2D pose model and a 2D pose -> 3D pose model:
I am alsodevelopingsomelibraries to improve the deep learning ecosystem around Flux to make it more approachable for general use (and to make my own life easier of course).
I work in climate research, specifically land-atmosphere exchanges (photosynthesis, evaporation, etc).
I use all sort of datasets, e.g. field or satelites measurements and models, to do analysis, processing, visualization, etc.
I think for us climate scientists, Julia is a great opportunity to unify languages used by modelers (who mostly use Fortran) and empiricists (who mostly use R or Python). The CLIMA project is a nice example. This is great because these two communities often work separately on the same topic…
I hope I can convince colleagues to join, and maybe after my postdoc get the opportunity to teach that language to young scientists =)
SimpleCache invalidates the cache if you change the source code of a function that is called by this function, but not if you change the source code of an independent function.
Memoize.jl will not invalidate the cache if you change the source code.
Sort of. Mom is definitely the greatest microbial exposure, at least for vaginal births (cesarean births have different microbiomes), but infant microbiomes are very different from adult ones. So it’s some combination of exposure and then selection in the infant gut (probably mostly from diet), not quite a 1:1 transfer from mom.
The manifolds feature of Optim made it much easier to relax the orbiting superfluid vortices in this paper. I’ll submit it as an example for the manual one of these days.
Hi have used Julia for particle physics data analysis, including some parts of the analysis for the measurement of the muon magnetic anomaly announced at Fermilab on April 7. Julia has very nice features for particle physics data analysis. I have used so far primarily linear algebra, structs / dictionaries to organize data, plotting library with the Plots.jl interface. The PrettyTables.jl library is very useful to report tabular data. For linear algebra elaborations Julia is the best programming language I know for functionality and logical elegance.
When speaking of “magnetic anomaly”, one must say in advance that the anomaly for the muon (and other particles) means the excess over 2 of the magnetic moment ratio, and it is expected and predicted by the theory. Said that, we measure that the magnetic anomaly is very close to the prediction, yet there is a 3.3 standard deviations measured difference, which grows to 4.2 standard deviations when combined with a previous measurement: it looks like that the measurement does not verify the theory (a fact that confusingly in this case may be reported as “anomaly”), but both the prediction and the measurement are difficult and complicate, so there are still chances that it is a statistical fluctuation or that there is some human error.
I missed the ability that R has to print large tabular data (dataframes and nD matrices) in a readable way when the horizontal space is limited and lines have to be wrapped. R is particularly good in handling properly row and column names in this situation. I wrote some code on top of PrettyTables.jl to do that. Since long time I planned to publish that although my code is not refined and not worth becoming a package.
Other Julia functionality that is quite nice for a physicist is Measurements.jl for dealing with uncertain and correlated quantities, and Covar.jl, with similar functionality. I may do some work in the future to use these packages for doing elaborations, for the time being I still use primarily some come I wrote on R.
Thanks, I am aware of that package, which is very nice. I was meaning a slightly different functionality, i.e. printing for instance a large matrix with line wrapping but keeping alignments, row and column names, in a readable way, like R does, instead of providing ways to interactively exploring data with very long lines.
