x1 (amount/value (li/ni) or labelled entity) is the (mutual) difference between/change in (states/realizations of/output from) general and relevant idea/object or subject of interest or function/aspect/property/topic/dependent parameter/dependent variable x2 under or due to a change in (independent) parameter/variable/index/identification/ownership/property/aspect x3 (generic name) to (later) specific state/realization/value/person/entity/meaning x4 (type and dimensions match those of x3) from (earlier) specific state/realization/value/person/entity/meaning x5 (type and dimensions match that of x3), under circumstances/conditions/with(in) scope/with caveats x6.
Especially in reference to incremental or quantized (or discretely observed) changes. x2 is an x1-type-valued function of x3; x4 and x5 are realizations of x3 (therefore, all three share type). For example, x1 could be some value with dimensions of currency/money/monetary value measured in real numbers, x2 could be a specific business' profit from certain sales, x3 may be time (so, x2 would be the profit over time, measured in a real number of units of currency), and x4 and x5 would be specific times - such as today and yesterday, respectively. Ideally, all factors, variables, conditions, etc. other than x3 should remain constant or there should be a control group/population which records the effects of changes in other factors, variables, conditions, etc.; however, in real life, things can be more complicated or messy and it is hard to ensure that all factors are accounted for (especially in social experiments and business); x5 records those other conditions or specifies the scope (for example: subjects, time, and location) of the 'experiment'. Typically, x4 will be temporally later than x5, but this need not be the case; in any case, x4 is taken to be the latter state (etc.) under consideration in the present analysis in the context of this word's usage (in some sequence). x1 could be "0" (whatsoever that means) - it could even be identically so; in other words, there need be no change or mutual difference, even in principle (the function/object could be constant); and this fact could even be known a priori. See also: "sumji" / "selsumji".