Monte-Carlo for REW

[basscleaner]

Hello! I'm a new in this theme, but for REW use experience. This is the invitation to discuss for some improving (to my opinion) REW in low frequencies problem solving. Sorry in advance, if it will be not new for you. Briefly, if to define the space of variables, which are responsible for peaks&dips "good behavior" in frequency responce for real room geometry, source parameters, antropology and design limits, it means, that we have typical task with multiple variables. We, as a rule, can reduce the number of them due to obvious demands and conditions and simplify calculations. For instance, the room height, egulateral triangle for listener and front distance range, LF acoustic center location(s), etc. Hence, one may to use Monte-Carlo methods to find the best system (LS+Listener) position for the room, if preliminary to set value of FR deviation as a target function. I had a deal with many projects in past, where I did such a work with big loss of time. "Random walk" into the task space with good "histories statisitics", sure, allow to find number of best positions and to do the right choice. This proposal at least deserves it be taken into consideration, don't you?

 

[John Mulcahy]

That's not something I have in mind for REW due to the complications of accurately simulating the room, its geometry and its furnishing and specifying the positional constraints for speakers and listeners. REW is focussed on measurement.

 

[basscleaner]

I've got something may change your mind. I know nobody, who did research about all important parameters, which are capable to influence strongly at initial FR for rectangular rooms and even more so for arbitrary geometry. Although FEM-algorithms are quiet ready to solve it for arbitrary room geometry. It just happened that way, but REW turned out to be very convenient tool for acoustical simulations, trust me. I tried CaraCad and some others to compare. For every my project for stereo or home cinema I tried to understand, which and how many parameters take part for FR forming, in addition to strict wave equation with boundary conditions, which as a rule define the solution to be needed. Step by step I had a clear understanding of the sequence for such tasks and to get satisfactory result for system positioning for any small room. Indeed you always must to begin in accordance to the boundary conditions like a group of limits for to change room geometry; number of LF sources with its acoustical center(s) points; system positions as a function of equilateral triangle; absorption surfaces; spread (or swing) for FR "peaks&dips"; zones of system place positions and dimensions. After you get some proposals by calculations for smooth FR, one needs to evaluate influence for acoustical performance, furnishing like sofas, cabinets and shelves and people inside. If you have the experience to evaluate, it is not difficult. At least four my last projects were successful with this technology. The main feature for this is to understand, that for target function in LF range one need to change room geometry. It means, that you must to reduce room dimensions (not necessarily rectangle!) within permitted limits to select the best system position inside the room. Of course, such a solution must be stable at small deviations for listener. That's why, Monte-Carlo seems to be capable there, because we can use millions "histories" to get the reliable result.