The reason I installed an older version of linux to use FB again is the access to assembly language instructions. Then you can use Basic as an efficient scripting language to glue machine code together. I tried to keep modern with the D programming language but their assembly language compiler is br...

I am trying the xoshiro256 type algorithms with a different mixing/tempering function to the original. You can change the timestamp function for non linux amd64. I'm not looking for the algorithm to pass the usual RNG tests (though it may or may nearly), just to have a large state space and avoid co...

I actually installed an older version of a linux amd64 to use FB again, to do algorithm research in "no share" mode. Then JavaScript and HTML 5 for any commercial things. It's a very positive simplification to stick to those 2 programming languages. If you are not fixated on programming la...

I spend a few hours creating a .bi file for the EGGX graphic library for Linux AMD64 only. That should enable me to use Freebasic again, since the built in graphics driver doesn't work for me anymore. \ Also the eggx library looks good for giving many programming language a Basic look and feel. For ...

Well, you can quantize 10.68372892 to 10.6 or 10 or 11, whatever way you want to do it. I was just saying there are some reasons to avoid anything like if possible when trying to solve problems by evolution. ..change of topic.. I was looking at Fortran as an alternative to FB. However I read that Ba...

I'm not using FB at the moment because of the graphic problem I have on linux amd64. I might try using FB with EGGX graphics library: https://www.ir.isas.jaxa.jp/~cyamauch/eggx_procall/ but that would mean writing an extensive .bi file. It seems easier to just use it from C. Anyway I had this to say...

I'm using XenialPup, where you are permanently administrator! Obviously I don't care very much about security.

What you do is create a symbolic link named libtinfo.so.5 to the ncurses library. For whatever reason that often fixes the problem. My bigger issue is the FB graphics library doesn't work with my current version of Linux, XenialPup. I can use the fltk library available on the forum but it is a bit O...

The gfxlib doesn't work on some versions of Linux now. Lack of maintenance? I'm using java and javafx gui instead. Oracle though are looking for ways to commercially exploit java to the maximum, which is a big negative. For all the programming languages out there, there aren't too many light weight,...

The problem with FB on Xenial Linux is that the graphics are always displayed in full screen mode. The problem with array slices in D is they become invalid if you increase the length of the underlying array in a very messy way. I think you can map everything in FB to D in a straight forward way, in...

Some of the D language is really badly though out. Like array views being total incompatible increasing array length. And some of it is overly complex or simply too much. However guy who designed it probably has an IQ of 150. Nevertheless a subset of the D language seems like it would map to a Basic...

I'm doing a review of Java currently. I was looking at a number of different languages but none appealed to me very much. The main reason I have dropped FB for the moment is I don't need access to assembly language for a number of planned projects. I can share more widely in Java and at least remain...

I was jumped on and assaulted by two bullies on the nvidia developers forum recently, or that's what it felt like. Isn't the internet a wonderful thing? The secret id of every human can come creeping out to wreck damage and destruction and seek revenge, even for another person having a slightly diff...

Probing the response of a randomly initialized deep neural network at different depths and number of weight parameters: https://drive.google.com/open?id=0BwsgMLjV0BnhYUlZOHhrZVB6cVU You can get beautiful cloud like patterns or swirl like patterns depending on the activation function you use and the ...

Fitting a 2D Chebyshev function approximation using evolution: type cheb2d xt(any) as single yt(any) as single coeffP(any) as single coeffC(any) as single precision as double iterations as ulong parentcost as single declare sub init(order as ulong,precision as double,iterations as ulong) declare sub...