Also, say I want to end up at -0.7 or any arbitrary negative point (will always be some “positive” multiple of -0.1) but the set of preceding terms need to start at a point and go up (+0.1 ) every other term or alternatingly after the descents) then down the next term but the slope for the descents “increases” -0.1 every other term and the upwards slope is always a flat +0.1. It basically alternates between steepingly descending going then going up a flat +0.1 to attenuate the increasingly steep drops. Would prefer the 2nd last terms subsequent slope to be a descent to the final arbitrary point selected for

Its like one of those zigzag type graphs that you see in psychophysical measurements or other experiments that try to find thresholds of sensation

Kinda feels like linear algebra or something but I probably forgot a lot. Is this kind of thing actually directly solvable or is it consistent with my theory here? Feels like it should be but i dont know enough to have any confidence either which way

    • sopularity_fax@sopuli.xyzOP
      1·
      3 months ago

      Ooh, havent stumbled on that phrase before :) What would it be if the ascent was replaced with plateaus or like it stays at the same after the descent?

      Gradient plateau or something?

      • scrollo@lemmy.worldEnglish
        3·
        3 months ago

        Hmmm, not sure I understand what you mean. However, there are things called “saddles” in numerical methods that kinda look an optimum point, but really they’re just intersections of funky surfaces.

  • BCsven@lemmy.ca
    2·
    3 months ago

    I’ve used something in engineering software and excel for optimization by goal seeking. You give some variable parameters and the value of another and the software manipulates the variables and as it approaches the goal it stops making wild ± guesses and starts narrowing them to meet your target (withing a tolerance)