GRRRR part 2

On 03/06/2022 15:30, Phil Hobbs wrote:
> Chris Jones wrote:
>> On 02/06/2022 23:33, jlarkin@highlandsniptechnology.com wrote:
>>> There have been attempts to use computers to actually design 
>>> circuits, or at least to optimize values in a given topology. They
>>>  tended to be ludicrous failures.
>>
>> I used an optimizer for some chip designs, it was very very good for
>>  things like choosing the size of the transistors in flipflops for 
>> best toggle frequency per supply current, and optimising a low pass 
>> filter for best noise and in-band error-vector-magnitude and stop-band 
>> rejection etc. all at the same time. It did way better than
>> I could have.
> 
> Sounds like a super useful tool.  I did something similar for optimizing
> plasmonic nanoantennas 15 or so years ago, and like yours, it found good
> solutions that weren't at all obvious.  So I'm a fan of the general 
> approach.

Simulated annealing or simplex are pretty good for that sort of thing. 
Then conjugate gradients once you get somewhere near an optimum.

You may never be sure you have the global optimum but it will probably 
find a solution that is better than any human can in a reasonable time.

>> The trick was to write a script that runs the right simulations and 
>> results in an expression (or several) that correctly describes how 
>> well a circuit meets the goals. Once you've done that, it can twiddle
>> the knobs much better than any human, and I don't mean because it
>> could do it faster and spam the simulations across a thousand CPUs
>> whilst you could look at only one at a time, it was also better in
>> that it could remember many sets of parameters that were good in
>> various ways, and combine them more efficiently than a human. It was
>> a company internal tool and they will surely have kept it that way.
> 
> Numerical optimization based on merit / penalty functions has been well
> known for 200 years, since Gauss iirc.  That's a far cry from actual 
> computer-based design.
> 
> Even lenses, which you'd think would be a natural application, have been 
> resistant to fully-automated design--it's all about finding a suitable 
> starting point.

Optical lens designs have been pretty much solved now at least in the 
domains that I frequent. There could only be a handful of truly weird 
configurations remaining that haven't been tried already.

The last interesting one in terms of being very different to orthodoxy 
was Willstrop's three mirror telescope (no chromatic or sphereical 
aberration, precise focus and very fast).

https://www.ast.cam.ac.uk/about/three-mirror.telescope

To the best of my knowledge no full scale one has ever been built.

I expect the odd novelty still lurks in the shadows. The search for 
eyepieces with ever more wide angle views remains the Holy grail - they 
are getting a bit ridiculous now with some offering 120 degree AFOV.

https://www.telescopehouse.com/Telescope-Accessories/EXPLORE-SCIENTIFIC-120-Ar-Eyepiece-9mm-2.html

I think that is the current record holder but I could be wrong on that. 
Price and weight are both a bit on the high side (lots of glass in it).

> There are various approaches that modify topologies, of which the best 
> known are genetic algorithms.

Modifying the topology is best done by humans. Optimising the components 
against some library of available materials and shapes is now the domain 
of sophisticated ray tracing programs. Zemax is probably the best known:

https://www.zemax.com/pages/try-opticstudio-for-free

Learning curve could best be described as STEEP...

-- 
Regards,
Martin Brown

Samsung 980 Pro NVMe SSD 1TB PCIe 4.0 (Amazon)

Oom, baby.

The 2 TB version jumps to $270. But if money is no object, those two would 
make an amazing primary and secondary drive combination. Primary being 
smaller.

I'm only using half of my 960 Pro (primary drive) and I already have the 980 
original (secondary drive).









John Doe <always.look@message.header> wrote:

> John Larkin wrote: 
> 
>> I'll need to get a new PC soon. People say that a screaming CPU and lots
>> of ram and solid-state C drive would really speed things up. 
> 
> I have had NVMe drives for ages. Get good ones, like from Samsung. Spend
> more than you used to spend on hard drives, it's well worth it. Primary
> (and secondary) storage is not whizbang fun, but as IBM used to put it, it
> affects "throughput" more than anything else. NVMe is what I have always
> wanted. 
> 
> If you use a secondary drive, the transfer rate between two premium NVMe
> drives is OUTRAGEOUS. Transferring movies from your downloads folder to
> the secondary drive is quick. Making backups of Windows is also quick. 

The 1 TB is $155. Great price.