Running a mechanical simulation in your mind

[CrikeyBobs]

Many moons ago, my math teacher once told the class about a guy who could run physical simulations in his mind. Basically he would imagine a machine: its construction, dimensions, materials etc. and then "set it running". At some later point he would return to the machine and take various measurements (eg. wear and tear).

At the time it sounded incredible, but I didn't really question it (he was a very good math teacher). Looking back, it sounds completely implausible, or at least severely exaggerated. So was teach feeding us a line of BS, or is there a kernel of truth to his tale?

[Ladewig]

I don't understand why such a savant would have to come back to it later, couldn't such a person imagine the future condition immediately?

[madurobob]

Tesla is said to have done this sort of thing. Not that its verifiable or testable. But, if he said that's how he figured things out I've no reason not to believe him.

[CrikeyBobs]

I don't understand why such a savant would have to come back to it later, couldn't such a person imagine the future condition immediately?

My memory is that it was a real time simulation. Of course there's plenty of room for error in my recollection, or possibly even in the original telling (assuming that the teacher was being sincere, of course!)

[Alkatran]

I don't understand why such a savant would have to come back to it later, couldn't such a person imagine the future condition immediately?

Because you can't just skip ahead in time when simulating some physical system. There's no way to get the right answer without simulating the intermediate states.

http://en.wikipedia.org/wiki/n-body_problem

That being said, I also find it hard to believe that anyone could run an accurate simulation of a complicated machine in their head. Especially to measure something complex like "wear and tear". It's certainly *possible*, but it's the type of thing you use a computer to do.

[CrikeyBobs]

Tesla is said to have done this sort of thing. Not that its verifiable or testable. But, if he said that's how he figured things out I've no reason not to believe him.

Excellent! Nikola_Tesla mentions this ability. As you say, it cannot be verified or tested (on Tesla anyway). So even if it has moved into the realms of urban legend, at least I know I wasn't being fed a line by my teacher!

[Beanbag]

I doubt you could measure "wear" in a hypothetical machine. You can estimate where wear would be likely, and allow for it, but wear is based on a large number of variables, like lubrication, surface characteristics, loading, etc, and changes over time.

Visualizing a mechanical system isn't that uncommon. A good watchmaker (we're talking old-school here) has a mental image of the entire train of wheels laid out in their mind in linear fashion. A good deal of diagnostic work involves looking at the information given by a watch timing machine, and interpreting the periodicity of unwanted events to take you to the individual wheel (gear for non-watchmakers) that might have a miniscule defect on one of the teeth. I use the same technique these days for diagnosing problems with electromechanical aircraft instruments. Based on how often the system "hiccups" per 360 degree revolution of the display, I can pretty much lay my finger on the exact gear that needs attention. Something that has a jerk or hang that has a fairly regular occurrence, but doesn't happen at the same regular points of display revolution is most likely a bad ball bearing assembly.

Beanbag

[macdoc]

Yeah I'd say a good estimator and visual imagination.

I don't think it's all that uncommon and we ca "set our minds" to accomplish things below the conscious levels.

If you consider how easy it is to catch a baseball for us and how hard to teach a robot to do that - we have a strong Baysesian comparator in our heads and with real world experience I can see how a talented engineer could do that.

I can visualize how many physical systems work without having a clue as to the math to quantify it.

Einstein did the thought experiment first, then the math - how's that for a monumental task of visualization.
He had the tools for it....yet was far enough outside the mainstream thinking to get to the breakthrough.
Remarkable.

[Ladewig]

Because you can't just skip ahead in time when simulating some physical system. There's no way to get the right answer without simulating the intermediate states.

I'll agree with that statement, but I do not see how that statement has anything to do with measuring wear and tear.

In order to imagine a system running in real time, one must start with knowledge about how much wear and tear will occur. Why not simply use that knowledge rather than fantasize about a real-time simulation?

[technoextreme]

Excellent! Nikola_Tesla mentions this ability. As you say, it cannot be verified or tested (on Tesla anyway). So even if it has moved into the realms of urban legend, at least I know I wasn't being fed a line by my teacher!
Tesla was ******* crazy at times.

[I Ratant]

The savant could be computing the known rates of wear of the materials in those applications for those lengths of times.
Doesn't seem too arcane.
Like a mechanic disassembling a motor and checking for the wear on the rotating parts. He knows where to look, what to expect to see, and whether or not the wear indicates a problem.

[Ladewig]

The savant could be computing the known rates of wear of the materials in those applications for those lengths of times.
Doesn't seem too arcane.

But isn't absurd to say, "I will now calculate how much wear the machine will experience in three days of constant use; it will take me exactly three days to come up with an answer."

[I Ratant]

That's how teleportation works, I believe.
You think you're in Bakersfield 3 days from now, you travel to Bakersfield and voila.. you're there!
Now, staying in Bakersfield for 3 days, that takes REAL wlllpower! :)

[macdoc]

The savant could be computing the known rates of wear of the materials in those applications for those lengths of times.
Doesn't seem too arcane.
Like a mechanic disassembling a motor and checking for the wear on the rotating parts. He knows where to look, what to expect to see, and whether or not the wear indicates a problem.


I don't even think savant is needed - the subconscious is incredible at "estimating" - that's one reason I prefer the Bayseian brain general theory