Changing circuitry of analog computer *During* simulations?

[Kirana Kumara P]

Your misunderstanding of analogue computers explains your weird approach and the title of this thread. You are treating the analogue computer as if it is a discrete multi-tasking digital computer running fixed-time algorithms. It is in fact a continuous function of continuous inputs. There can be no final step in an equation solution as the state variables will never really be stable. The output of an analogue processor that you display will be the result of a low-pass filtering process of the recent values of state variables. You cannot afford to wait for the solution to settle, nor can you waste time reloading the state variable values, or restarting the circuit for every frame.

This could mean that an analog computer is infinitely fast. One gets the output (result) as soon as the input is applied. Am I right?

 

[Baluncore]

This could mean that an analog computer is infinitesimally fast. One gets the output (result) as soon as the input is applied. Am I right?

Probably not.
"infinitesimally fast" actually means so slow it appears to never change.
"infinitely fast" actually means so fast it appears to be instant.

The integrators take time to change. The immediate value will be whatever you initialise it with.
That would suggest that the immediate output is not the next state, but is actually the previous state.

 

[Kirana Kumara P]

Probably not.
"infinitesimally fast" actually means so slow it appears to never change.
"infinitely fast" actually means so fast it appears to be instant.

The integrators take time to change. The immediate value will be whatever you initialise it with.
That would suggest that the immediate output is not the next state, but is actually the previous state.

I had corrected it to "infinitely fast". What would be the typical time lag?

 

[Baluncore]

I had corrected it to "infinitely fast". What would be the typical time lag?

You must define what you mean by "time lag".
If the inputs did not continue to change, the Time Constant of the integrators would get it to within 37% of a theoretical final value. Two TC will get to within 13.5%, 3TC to 4.9%, 4TC to 1.83%, 5TC to within 0.67%...

 

[FactChecker]

You must define what you mean by "time lag".
If the inputs did not continue to change, the Time Constant of the integrators would get it to within 37% of a theoretical final value. Two TC will get to within 13.5%, 3TC to 4.9%, 4TC to 1.83%, 5TC to within 0.67%...

In modeling a physical process in time, there should be no integrators, filtering, or signal latency except those that model the real physical process. There are no "final values", there are just the correct values as a function of time.

 

[FactChecker]

This could mean that an analog computer is infinitely fast. One gets the output (result) as soon as the input is applied. Am I right?

Yes. You should instantly get the result that the physical process would have at that time, given the same inputs in time. And the results should change with time as the physical process would. (After all, the analog computer is an electrical physical process made to mimic your problem.) I consider an analog process to be equivalent to a digital process with an infinitely small time step. There is no data latency internal to the analog computer.

Suppose you have a step input with steps every 33 milliseconds. An analog computer will instantly start to respond to the step changes as a physical process would. The step inputs are not realistic, because your real physics had a continuous analog input that went smoothly from the prior step value to the new step value. So the simulation has some delay due to the "sample and hold" nature of your inputs. The essential difference between the analog computer and a digital one is that the analog computer outputs will start to respond immediately whereas the digital computer will hold its output constant till it can compute new outputs (about 33 milliseconds later). Here I am assuming that the digital calculations would be done in a "hard loop" with as small a frame time as possible and that its new outputs would not be available till 33 milliseconds after the inputs change. The signal latency of the simulation may be compensated for with predictors (lead filters), but that will introduce noise which may not be acceptable. The results of the analog simulation can be improved by giving it inputs that are as close to continuous as possible (a higher rate than 30 Hz). That will not help a digital simulation unless it can run at the faster rate.

That being said, I do not see any way to set up an analog computer for your problem. You do not have a simple problem of switching between a small number of signal paths. Your simulation will have to adapt to far more situations than I have ever dealt with on an analog computer. I don't see how you could use analog computers, unless there are some massively parallel ones. And I am not aware of any.

 

[anorlunda]

@Kirana Kumara P , how many times do you need to be told, "No," before you accept the answer?

You should present your desires to your engineers and then accept their advice about what is feasible and how to go about it. If the situation were reversed and the engineer said to you, "Treat the patient with non-surgical methods," I'm sure you would quickly say "Leave the medicine to me."

 

[FactChecker]

@Kirana Kumara P , how many times do you need to be told, "No," before you accept the answer?

You should present your desires to your engineers and then accept their advice about what is feasible and how to go about it. If the situation were reversed and the engineer said to you, "Treat the patient with non-surgical methods," I'm sure you would quickly say "Leave the medicine to me."

Maybe he has accepted the answer and still has some intellectual curiosity on the subject. In my opinion this thread has been a little confusing.

 

[Baluncore]

In my opinion there has been some misleading information in this thread.

That is probably true of almost every thread as the OP question is rarely fully explained.

I believe this may be a case of the Dunning–Kruger effect; the experts are concerned with doubts and details that only they understand, while the OP does not understand the depth of the advanced technology, but continues to grasp for hope of a solution, long after Pandora's box has been well ventilated.

 

[FactChecker]

That is probably true of almost every thread as the OP question is rarely fully explained.

I believe this may be a case of the Dunning–Kruger effect; the experts are concerned with doubts and details that only they understand, while the OP does not understand the depth of the advanced technology, but continues to grasp for hope of a solution, long after Pandora's box has been well ventilated.

I would like to restate my post. I think the information has been correct, but maybe not exactly answering the right question. I reworded it to say that I think the thread has been a little confusing.

 

[Kirana Kumara P]

Some of the replies in the thread are clear that it will not be possible to build an analog computer that can solve the concerned problem, while the others worry about the complexity, cost, reliability, risk (probability of success), future use, possibility of not being able to come up with an (suitable) analogy/circuit etc.

I am not convinced still that it is not possible to build an analog computer that can solve the concerned problem. However, I am less likely to design/build the said analog computer, at least for the time being.

I would like to once again express my since gratitude to all those who have spent their precious time answering my questions.