array_map returns a new array of the callback's return values; the input array's keys are preserved when called on a SINGLE array. array_filter keeps elements where the callback is truthy. The classic gotcha is that array_filter keeps the ORIGINAL keys even though the values are now sparse: when JSON-encoding the result, PHP will emit an object literal because keys are no longer 0..n-1. Wrap the result in array_values(...) whenever you want a clean numerically-indexed list. Arrow functions (fn($x) => $x * 2) require PHP 7.4+; pre-7.4 you used function ($x) use ($outer) { return ...; }.

array_reduce($arr, $callback, $initial) walks the array left-to-right with an accumulator, returning the final accumulator. The seed is mandatory if you want a known starting value; pass null to mean 'no seed' but be ready for a null first iteration. The pattern of building an associative array from a list of pairs comes up constantly when normalising parsed CSV or query results; array_combine($keys, $values) is shorter when you already have parallel arrays. For numeric sums, array_sum($nums) is even shorter and avoids the callback overhead.

array_combine zips two parallel arrays into an associative array using the first as keys; both must be the same length. array_keys and array_values give you a clean numerically-indexed list of either side, useful for re-indexing after a filter. array_flip swaps keys and values, collapsing duplicates silently, which is the fastest way to dedupe a list (then call array_keys on the result). Distinguish in_array($v, $arr) (linear scan over values) from array_key_exists($k, $arr) and isset($arr[$k]): isset returns false on a key whose value is null, while array_key_exists returns true; pick based on whether null is a meaningful state in your data.