Wolfram Mathematica

« first day (2620 days earlier) ← previous day next day → last day (1874 days later) »

b3m2a1

12:49 AM

@Kuba what do you think about this: mathematica.meta.stackexchange.com/a/2454/38205

is it worth pursuing?

5 hours later…

user21

5:31 AM

@b3m2a1, that for the link. Much appreciated. The FEM is almost always NOT the best method to solve a specific problem, but is almost always a reasonable choice. The major benefit of the FEM is that it works reasonably well on a huge class of problems.

J. M. is slightly pensive

6:15 AM

@b3m2a1 Yes, that's the best approach, as opposed to computing factorials that can overflow before dividing. Nice to see you did it.

@b3m2a1 the Stirling cycle numbers are always nonnegative; they are equivalent to (-1)^(n - k) StirlingS1[n, k].

b3m2a1

6:40 AM

@J.M.isslightlypensive ah I see. I just did something that'd replicate the Mathematica StirlingS1 so I have the negatives as well.

J. M. is slightly pensive

6:57 AM

@b3m2a1 it might be more expedient to just absorb the signs into the overall (-1)^(* something *) factor, so that no subtractions are involved in the recurrence for generating Stirling numbers.

1 hour later…

Kuba

8:26 AM

@b3m2a1 I wouldn't call it lightweight :) I am terrible at judging such cases so I won't say. What I think is worth pursuing is a project type/structure independent IDE. Something I started with DevTools`. So I opt for giving people tools that do well defined things and how they use it is up to them. My experience with your code is that it is great if it fits your style/project setup. So either your idea comes with a really clear documentation or you may end up with low feedback.
That is just an opinion though and I could be completely wrong.

2 hours later…

Feeds

10:34 AM

PackageData.net: Carlson

posted on March 22, 2019

A package for Carlson elliptic integrals in Mathematica.

kirma

10:45 AM

Mma Online seems to be having some issues... wonder what's going on.

Like, extreme slowness even on rendering simple notebooks.

4 hours later…

Edmund

2:16 PM

Any resources out there on how to create an EntityStore from a REST API? I have access to a set of internal API functions on top of relational data. I want to create an EntityStore from them but not too certain where to start.

C. E.

2:27 PM

@Edmund You can start here. Pay special attention to the part about computed properties if you want to get data from the API on demand.

Edmund

2:41 PM

@C.E. Thanks. I don't have an initial list of entities locally so I guess I would need to create a package that does the initial load of the main entities?

3 hours later…

C. E.

5:27 PM

@Edmund You have to create and register the EntityStore somehow, such as in a package, yes.

b3m2a1

@Kuba sounds reasonable to me. My current plan is basically to make something that implements tabbing by saving to files with a left-hand file browser and a mechanism for people to write plugins that can be accessible via a menu in the DockedCells.

All that it'll need will be a root directory to start looking down from

That's why I was calling it lightweight, since I'm not trying to force too much structure on people and so keep things flexible.

b3m2a1

5:45 PM

@Edmund @CarlLange has recently been doing this with his WikiData stuff. You could give that a look too.

Jason B.

6:22 PM

@CarlLange - it is possible to set up an entity store where the entities are not predefined. You set up your entity store as in the link above, defining the computed properties as described, but then also use an "EntityValidationFunction".

I originally had the Molecule stuff set up as an entity store, where you would instantiate an entity via Entity[ "ChemicalStructure", <smiles string>]

and then the entity validation function would verify that the input string is valid

it would be easy to have the validation function call the API and verify the existence of the entity in the database

b3m2a1

@JasonB. You know that's actually a pretty slick way to create objects / do OOP

Jason B.

it is a shame I can't find a documented example of a non-trivial EntityValidationFunction on the web

@b3m2a1 - I think a neat side project would be to set up an entity store for publications, using the DOI resolver to get all the information

the entity standard names would be the DOI, the formatted name could be the paper title, or a formatted citation, or what not

@b3m2a1 it was before I started using the expression store stuff. I did find the custom entity framework to be really easy to work with

(oops, I sent the first message to carl instead of edmund)

@Edmund - search the doc page for EntityStore for the text "Create an entity store of all integers", that shows how to use an entity store without predefining the entity list

2 hours later…

Edmund

8:42 PM

@JasonB. Thanks. I will look that up.

2 hours later…