Like many people I'm re-evaluating my career in this new year and it's occurred to me that coding would be a useful skill to develop. I have a biochemistry PhD (lab based) and now work in admin in an academic setting. My institution does a lot of software development and data analysis and is always on the lookout for staff in those areas, particularly women, so could be a useful avenue for me to explore.
I have zero coding experience so am not sure where to start. I need to test the waters to see if this is something I could be good at/would enjoy. What language would be best to learn for scientific/'big data' type work? Can anyone recommend courses? Ones that cost money would be fine if it's not extortionate. But it can't be too intense as I need to maintain work-life balance due to having a young child. I'm based in the UK.
EDIT: Thank you all so much for your help! Looks like Python is the winner, maybe R too
What’s the best languages/software for science (math, physics, astronomy, chemistry, biology etc) related tasks?
I will like to create something like this in 3D:
Basically, the title. I am currently working as a scientific software developer for a tiny company in a Bangalore. But, I am always looking for newer opportunities for the future. Hence, I am asking here.
I am reading a paper where the authors developed a tool that uses GPU software acceleration [via multithreading (?) ] to cluster data. They then compared it to conventional software that utilizes CPU multithreading and found that GPU software acceleration does the clustering task faster. My question is if this is the case, why isn't GPU software acceleration being utilized more in ML?
> I am aware of the basics that there are tasks that are better suited for CPUs (e.g. fewer, longer/complex tasks) and some tasks that are better suited for GPUs (numerous, shorter/repetitive tasks). But aside from these factors, what other factors affect why GPU isn't the more commonly used processing unit in ML?
I was looking around for mods and OpenEye has a specific version listed that I can't find nowhere else. There's no download button.
I'm looking for a software that allows you to tag phrases/quotes so, when I need them, I can find all the phrases/quotes that I've tagged with a particular tag. Something like Maxdqa, but for scientific articles
I tried Obsidian by exporting the underlines and tagging the phrases but then I can't find a window that allows you to see all the phrases/quotes with a given tag. Probably I'm doing something wrong
Thanks in advance for the help!
Hi! Does anyone have suggestions the easiest/best software to create illustrations for publication. I notice that a lot of reviews in the frontiers journal group have beautiful diagrams.
Hi all, I am an engineering mathematics masters graduate and have experience writing code for mathematical modelling research problems. I am looking into a career in software engineering and am wondering if anyone could tell me the key differences and things to look for in traditional software engineering as opposed to scientific programming.
I have a fairly basic understanding of arrays and pointers. But, I need to learn better about these arrays & pointers, as well as data structure, linked lists, memory management etc. So, what would you suggest for learning C.
I came across these two MIT courses for learning C :-
Also, I found hackerrank has some practice tests. But, what would be your suggestion to maximize learning in the shorter time frame for my requirement.
Thank you
Hi, our client is the expert and manufacturer of healthcare diagnostics equipment for scientific and medical purposes and has its own custom-developed software complex developed 22 years ago. Nowadays these technologies are obsolete and it was too risky to keep the software on legacy development tools.
What's the best solution - perform software actualization and migration to an updated version of the framework OR develop the software from scratch with a totally different technology?
Check the new case study here: https://www.softacom.com/healthcare_and_scientific_diagnostics_software/
I'm currently writing an article in which I need to create/recreate a figure. I essentially want to create a figure like the one shown in this link. I can't use that figure since I do not have copyright for it, and also I wish to recreate the figure in better resolution.
What software would you suggest for creating such a figure?
First, it’s surprising to me how much raw capital is poured into startups, how few studies are out there to try to quantify what causes failure.
One of the studies I reviewed was a study on studies — it tried to aggregate all the studies it could find in an attempt to add some structure to how software startups should behave, but in its own words:
“…there is a lack of relevant primary studies on software development in the startup context.”
(I get that software development isn’t the same as studying why they fail, but my research concluded the same).
Not only aren’t there a whole lot, but the ones I found were imperfect, mostly relating to a lack of substantial sample size.
Despite that, some decent studies did their best with what they had.
I recommend reading the studies themselves because I’m oversimplifying here, but here’s what I found:
- They mismanaged their resources
- They did a poor job of collecting and integrating customer feedback
- Lack of working chemistry between founders
These seem plenty obvious (and I provided more details here), but one thing I found the studies conflicted on was experience.
Some studies marked a lack of experience as a failure reason, but others didn't and found many first-time founders finding success.
In my personal experience, going through one startup as a founder was enough for me to "get" it (my first startup checked all three of these steps, but my second one was "successful").
Anyway, I still thought going through this was pretty useful.
Here are the studies themselves
Study 1, 2, [3](https://pdf.sciencedirectassets.com/277811/1-s2.0-S1877042816X00213/1-s2.0-S1877042816315889/main.pdf?X-Amz-Security-Token=IQoJb3JpZ2luX2VjEGwaCXVzLWVhc3QtMSJHMEUCIQDr6ij6Nt7PsgbLDhnZtH81d91Vnj25iijifsLWdBXgxwIgOGUTcunqRV69nfhs%2BdeA1Krw3N%2B%2FH8YUDb5IWhvIcdUq%2BgMIdRAEGgwwNTkwMDM1NDY4NjUiDJRsV23vfEMSKaPW7CrXA4RITB4f3GkdeyEtH743SoyCvCsnF1oHvXacqXh%2FUB8%2FPWPwAxRkAXBUbMYfo3IQVd1ytUBOEh7WZFOeJIoqy0BkfE7%2FnDWNWIxop4NWM2lCvAgn4wqag0W75d4hSGoKNuWDAHXsr44g%2Bx5mCmw86%2BLI1TiViE%2FdeI6ZMorgnAQL9mv8oBCTFPLSUliKBOa4pjKeotDWNKO8VqZ9kyamBGyTore%2FTh3BGye%2FwfVWVdtPqsbL3COqnPwSwRtis0ugEBkhbFFjG99zCRwPn7rH7NGAIVUAPQAwL8E0JjrGhSdRWgL156el1OLrARkimxW7VVeNPWgVbsgPN7HghV8gFTj%2FmWPCyKAqlzP57Y1ut2R5kiQmoxuCYklKarXM8JpKKb
... keep reading on reddit ➡Hi, I'd like to know what programming languages/software frameworks are particularly well-suited to developing software to control scientific instruments and receive data from them. That would presumably involve communicating with microprocessor(s) to do things like control motors, read sensor data etc.
Primarily thinking about a Windows system, would something like WPF be best? I can't seem to find much on the internet about this topic as opposed to a purely embedded system, but perhaps I'm not using the right terms (what would you call such a program?)
Any help would be greatly appreciated.
Edit: Thanks for your responses, you've given me some useful things to Google!
Im assuming its not Microsoft Word.
Hi all,
My name is Jonah, I'm the founder of sciugo.com a LIMS system hoping revolutionize the way we share experiment design metadata and results data internally.
I'm looking for a cofounder to extend upon the vision and drive business ops.
Here's some demo videos of previous versions which I'd love to discuss at your earliest convenience:
I recently got a recommendation to use Grammarly for final corrections as I worked on an article in team (I had to write one chapter), and I did not agree with some usage of definite / indefinite articles as well as commas. I am not a native English speaker, but what are your thoughts of Grammarly (or alternative SWs) in general?
This is a continuation post to the VkFFT announcement. Here I present an example of scientific application, that outperforms its CUDA counterpart, has no proprietary code behind it and is crossplatform - Vulkan Spirit. This is a fully GPU version of the computational magnetism package Spirit, developed at FZ Jülich. I hope this post can motivate other scientists (including machine learning researchers) to explore the world of Vulkan for scientific GPU computing, as right now it is heavily dominated by CUDA.
From mathematical point of view, simulation of a magnetic system in micromagnetics can be described as a system of differential equations (LLG) on a finite-difference mesh. Each cell's position is influenced by positions of its neighbors, material parameters, external effects and many other things. Successful iterative integration of the LLG system can yield time dynamics, resembling experimentally observed evolutuon of magnetics.
From the programming point of view, simulation software is simpler than the one that has to communicate with the user during runtime. There are no calculations performed on the CPU during the execution, so it is only used to create a command buffer before launch, which is not modified afterwards. Combining multiple iterations in a single command buffer significantly reduces initialization overhead and is one of the main benefits of using Vulkan due to its low-level nature.
The Vulkan Spirit includes many algorithms written in SPIR-V shaders, such as LBFGS, VP and CG energy minimizers, RK4 and Depondt integrators of differential equations. The VkFFT library was primarily developed to compute the Dipole-Dipole interaction part of the gradient, which is one of the most time consuming parts of the iteration. It was possible to optimize every single part of the command buffer to reduce memory transfers to the minimum due to the explicit memory handling of Vulkan. This allowed to get up to 3x performance increase in comparison to CUDA based micromagnetics code mumax3. More information can be found on the github repository: https://github.com/DTolm/spirit
Thanks for the read!
As a side note, the VkFFT has been improved in the past month - it su
... keep reading on reddit ➡I'm the founder of sciugo.com a LIMS system hoping revolutionize the way we share experiment design metadata and results data internally.
I'm looking for a cofounder to extend upon the vision and drive business ops.
Here's some demo videos of previous versions which I'd love to discuss at your earliest convenience:
Sneak peak of newest Western blot automated quantification workflow:
https://www.youtube.com/watch?v=-ucbwBkPJEo
Experiment design workflow:
