Subtract "v2" from "v1". */

Read zero-terminated string into z */

Return the offset of the coefficient of the constant term of "node" * within the (I)LP. * * Within each node, the coefficients have the following o

Convert node->sched into a map and return this map. * * The result is cached in node->sched_map, which needs to be released * whenever node->sched

Compute a basis for the rows in the linear part of the schedule * and extend this basis to a full basis. The remaining rows * can then be used to f

Return the offset of the coefficients of the variables of "node" * within the (I)LP. * * Within each node, the coefficients have the following orde

Given the index of a column "c", return the index of a row * that can be used to pivot the column in, with either an increase * (sgn > 0) or a decre

Print the "node" to "p". * * "in_block" is set if we are currently inside a block. * If so, we do not print a block around the children of a block

Print the constraints of "bmap" to "p". * The names of the variables are taken from "space". * "latex" is set if the constraints should be printed i

Print the quasi-polynomial "qp" to "p" in C format, with the variable names * taken from the domain space "space". */

Return a set containing those elements in the domain * of "pwaff" where it satisfies "fn" (if complement is 0) or * does not satisfy "fn" (if comple

First read the domain of the affine expression, which may be * a parameter space or a set. * The tricky part is that we don't know if the domain is

Read a piecewise affine expression from "s" with domain (space) "dom". */

Given a basic set that is known to be bounded, find and return * an integer point in the basic set, if there is any. * * After handling some trivia

Set the indentation of the innermost active YAML element to "indent". * If "indent" is equal to ISL_YAML_INDENT_FLOW, then this means * that the cur

Move down to next level and push callback onto context tableau * to decrease the level again when it gets rolled back across * the current state. T

Check if any of the currently active sample values satisfies * the inequality "ineq" (an equality if eq is set). */

Check that a schedule can be constructed on the given domain * with the given validity and proximity constraints. */

Drop the parameters in "uset" that are not also in "space". * "n" is the number of parameters in "space". */

Return the universe in the space of "map". */

Add "f" times the number of equality and inequality constraints of "bset" * to "n_eq" and "n_ineq" and free "bset". */

Compute the convex hull of a set without any parameters or * integer divisions. Depending on whether the set is bounded, * we pass control to the w

Given a "sink" access, a list of n "source" accesses, * compute for each iteration of the sink access * and for each element accessed by that iterat

TODO: fix this function

Helper function to add a guarded polynomial to either pwf_tight or pwf, * depending on whether the result has been determined to be tight. */

Given two constraints "a" and "b" on the variable at position "abs_pos" * (in "a" and "b"), add a constraint to "bset" that ensures that the * bound

Generate code for the schedule domain "bounds" * and add the result to "list". * * We mainly detect strides here and check if the bounds do not *

Move "sample" to a point that is one up (or down) from the original * point in dimension "pos". */

Is any domain element of "umap" scheduled after any of * the corresponding image elements by the tree rooted at "node"? * * If "umap" is empty, the

Align the parameters of the to isl_pw_aff arguments and * then apply a function "fn" on them that returns an isl_map. */

Subtract expr2 from expr1. * * If expr2 is zero, we simply return expr1. * If expr1 is zero, we return * * (isl_ast_op_minus, expr2) * * Otherw

Compute an explicit representation for the existentially quantified * variables in "bset" by computing the "minimal value" of the set * variables.

Compute the set difference between bmap and map and call * dc->add on each of the piece until this function returns * a negative value. * Return 0

Check if there is any lower bound (if lower == 0) and/or upper * bound (if upper == 0) on the specified dim. */

Given two basic sets i and j, * check if relaxing all the cut constraints of i by one turns * them into valid constraint for j and check if we can w

Check that the result of an allocation ("p") is not NULL and * complain if it is. * The only exception is when allocation size ("size") is equal to

Construct the space of the cluster with index "i" containing * the strongly connected component "scc". * * In particular, construct a space called

Check if the union of "bmap_i" and the basic map represented by info[j] * can be represented by a single basic map, * after aligning the divs of "bm

Check if the union of the given pair of basic maps * can be represented by a single basic map. * If so, replace the pair by the single basic map and

Combine the initial two partial solution of "sol" into * a partial solution with the current context domain of "sol" and * the function description

Don't assume equalities are in order, because align_divs * may have changed the order of the divs. */

Compute a schedule for the given dependence graph and insert it at "node". * Return the updated schedule node. * * We first check if the graph is c

Compute a schedule for the given instance set, validity constraints and * proximity constraints and return a corresponding union map representation.

Examine the current band (the rows between graph->band_start and * graph->n_total_row), deciding whether to drop it or add it to "node" * and then c

Construct a band of schedule rows for a connected dependence graph. * The caller is responsible for determining the strongly connected * components

Compute a schedule for a subgraph of "graph". In particular, for * the graph composed of nodes that satisfy node_pred and edges that * that satisfy

Construct a piecewise quasipolynomial that is constant on the given * domain. In particular, it is * 0 if cst == 0 * 1 if cst == 1 * infinity if

Copy constraint from src to dst, putting the vars of src at offset * dim_off in dst and the divs of src at offset div_off in dst. * If both sets are

Create an AST node for the current dimension based on * the schedule domain "bounds" and return the node encapsulated * in an isl_ast_graft. * * "

Given a non-parametric tableau, add cuts until an integer * sample point is obtained or until the tableau is determined * to be integer infeasible.

Delete a given equality, moving all the following equalities one up. */

Use Tarjan's algorithm for computing the strongly connected components * in the dependence graph only considering those edges defined by "follows".

Apply Tarjan's algorithm to detect the strongly connected components * in the dependence graph. * Only consider the (conditional) validity dependenc

Does set dimension "pos" of "set" have an obviously fixed value? */

Given a dimension type for an isl_qpolynomial_fold, * return the corresponding type for the domain. */