. * Tutorial 21: Reading networks in Mata with the nwapi functions . * Requires the core nwcommands 3.24 or later. . . * 1. What is in memory . nwwebuse florentine, nwclear . mata: nwapi_version() 1 . mata: nwapi_count() 2 . mata: nwapi_names()' 1 2 +-----------------------------+ 1 | flobusiness flomarriage | +-----------------------------+ . mata: nwapi_current() flomarriage . mata: nwapi_exists("flomarriage"), nwapi_exists("nonet") 1 2 +---------+ 1 | 1 0 | +---------+ . mata: nwapi_nodes("flomarriage"), nwapi_ties("flomarriage") 1 2 +-----------+ 1 | 16 20 | +-----------+ . mata: nwapi_directed("flomarriage"), nwapi_valued("flomarriage"), nwapi_twomode( > "flomarriage") 1 2 3 +-------------+ 1 | 0 0 0 | +-------------+ . mata: nwapi_nodes(), nwapi_ties() 1 2 +-----------+ 1 | 16 20 | +-----------+ . mata: lab = nwapi_nodelabels("flomarriage") . mata: lab[1..5]' 1 2 3 4 5 +------------------------------------------------------------------+ 1 | acciaiuoli albizzi barbadori bischeri castellani | +------------------------------------------------------------------+ . . * 2. Reading the ties . mata: A = nwapi_matrix("flomarriage") . mata: rows(A), cols(A) 1 2 +-----------+ 1 | 16 16 | +-----------+ . mata: A[1..4, 1..4] [symmetric] 1 2 3 4 +-----------------+ 1 | 0 | 2 | 0 0 | 3 | 0 0 0 | 4 | 0 0 0 0 | +-----------------+ . mata: hasmissing(A) 0 . mata: A = nwapi_matrix("flomarriage") . mata: A[2,1] = 7 . mata: B = nwapi_matrix("flomarriage") . mata: A[1..3, 1..3], B[1..3, 1..3] 1 2 3 4 5 6 +-------------------------+ 1 | 0 0 0 0 0 0 | 2 | 7 0 0 0 0 0 | 3 | 0 0 0 0 0 0 | +-------------------------+ . mata: mata drop A B . . mata: E = nwapi_edgelist("flomarriage") . mata: rows(E), cols(E) 1 2 +-----------+ 1 | 20 3 | +-----------+ . mata: E[1..5, .] 1 2 3 +-------------+ 1 | 1 9 1 | 2 | 2 6 1 | 3 | 2 7 1 | 4 | 2 9 1 | 5 | 3 5 1 | +-------------+ . mata: nwapi_outdegree("flomarriage")' 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 +---------------------------------------------------------------------------- 1 | 1 3 2 3 3 1 4 1 6 1 3 0 3 2 4 +---------------------------------------------------------------------------- 16 ------+ 1 3 | ------+ . mata: nwapi_outdegree("flomarriage") == nwapi_indegree("flomarriage") 1 . mata: id = nwapi_nodeid("flomarriage", "medici") . mata: id 9 . mata: nb = nwapi_neighbors("flomarriage", id) . mata: nb' 1 2 3 4 5 6 +-------------------------------+ 1 | 1 2 3 13 14 16 | +-------------------------------+ . mata: lab[nb]' 1 2 3 4 5 +------------------------------------------------------------------ 1 | acciaiuoli albizzi barbadori ridolfi salviati +------------------------------------------------------------------ 6 --------------+ 1 tornabuoni | --------------+ . mata: lab[nwapi_neighbors("flomarriage", "medici")]' 1 2 3 4 5 +------------------------------------------------------------------ 1 | acciaiuoli albizzi barbadori ridolfi salviati +------------------------------------------------------------------ 6 --------------+ 1 tornabuoni | --------------+ . mata: mata drop E id nb lab . . * directed networks: out- and in-neighbors . nwwebuse glasgow, nwclear . mata: nwapi_directed("glasgow1"), nwapi_nodes("glasgow1"), nwapi_ties("glasgow1" > ) 1 2 3 +-------------------+ 1 | 1 50 113 | +-------------------+ . mata: nwapi_neighbors("glasgow1", 1)' 1 2 +---------+ 1 | 3 6 | +---------+ . mata: nwapi_inneighbors("glasgow1", 1)' 6 . mata: nwapi_outdegree("glasgow1")'[1..10] 1 2 3 4 5 6 7 8 9 10 +---------------------------------------------------+ 1 | 2 3 3 3 0 3 3 2 5 2 | +---------------------------------------------------+ . mata: nwapi_indegree("glasgow1")'[1..10] 1 2 3 4 5 6 7 8 9 10 +---------------------------------------------------+ 1 | 1 3 8 0 0 1 3 2 2 2 | +---------------------------------------------------+ . . * valued networks: the matrix holds the values, the degrees count the ties . nwwebuse sampson, nwclear . mata: nwapi_valued("samplike") 1 . mata: A = nwapi_matrix("samplike") . mata: max(A) 3 . mata: sum(A :> 0) 88 . mata: nwapi_ties("samplike") 88 . nwtomata samplike, mat(S) . mata: sum(S :> 0) 106 . mata: mata drop A S . . * 3. Looping over the networks in memory . nwwebuse glasgow, nwclear . set seed 2111 . nwrandom 30, prob(.1) undirected name(random30) . mata: ------------------------------------------------- mata (type end to exit) -------- : void nwt_table() > { > real scalar i, n, m, dens > string colvector names > string scalar nm > > names = nwapi_names() > printf("{txt}%-14s %6s %6s %-10s %7s\n", "network", "nodes", "ties", "type" > , "density") > for (i = 1; i <= rows(names); i++) { > nm = names[i] > n = nwapi_nodes(nm) > m = nwapi_ties(nm) > dens = m / (nwapi_directed(nm) ? n*(n-1) : n*(n-1)/2) > printf("{res}%-14s %6.0f %6.0f %-10s %7.3f\n", nm, n, m, > nwapi_directed(nm) ? "directed" : "undirected", dens) > } > } : end ---------------------------------------------------------------------------------- . mata: nwt_table() network nodes ties type density glasgow1 50 113 directed 0.046 glasgow2 50 116 directed 0.047 glasgow3 50 122 directed 0.050 random30 30 51 undirected 0.117 . . * 4. A measure of your own . mata: ------------------------------------------------- mata (type end to exit) -------- : // how many other nodes are reached in at most two steps (matrix version) : real colvector nwt_reach2(string scalar net) > { > real matrix A, R > > A = nwapi_matrix(net) > R = ((A + A*A) :> 0) > _diag(R, 0) > return(rowsum(R)) > } : // the mean of x over the neighbors of every node; missing for an isolate (works > for large networks) : real colvector nwt_nbrmean(string scalar net, real colvector x) > { > real scalar i, n > real colvector out, nb > > n = nwapi_nodes(net) > out = J(n, 1, .) > for (i = 1; i <= n; i++) { > nb = nwapi_neighbors(net, i) > if (rows(nb) > 0) out[i] = mean(x[nb]) > } > return(out) > } : end ---------------------------------------------------------------------------------- . nwwebuse florentine, nwclear . mata: nwt_reach2("flomarriage")' 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 +---------------------------------------------------------------------------- 1 | 6 10 9 8 6 3 9 4 11 2 6 0 11 7 8 +---------------------------------------------------------------------------- 16 ------+ 1 10 | ------+ . mata: x = st_data(., "wealth") . mata: nwt_nbrmean("flomarriage", x)' 1 2 3 4 5 +----------------------------------------------------------------------- 1 | 103 47.66666667 61.5 67.66666667 83.33333333 +----------------------------------------------------------------------- 6 7 8 9 10 ----------------------------------------------------------------------- 1 36 42.5 8 31 10 ----------------------------------------------------------------------- 11 12 13 14 15 ----------------------------------------------------------------------- 1 70 . 99 75.5 35 ----------------------------------------------------------------------- 16 ---------------+ 1 46 | ---------------+ . . * 5. Back to Stata: variables, matrices, a new network . mata: st_store(., st_addvar("double", "reach2"), nwt_reach2("flomarriage")) . mata: st_store(., st_addvar("double", "nbrwealth"), nwt_nbrmean("flomarriage", x > )) . list _nwnode wealth reach2 nbrwealth in 1/6 +------------------------------------------+ | _nwnode wealth reach2 nbrwealth | |------------------------------------------| 1. | acciaiuoli 10 6 103 | 2. | albizzi 36 10 47.666667 | 3. | barbadori 55 9 61.5 | 4. | bischeri 44 8 67.666667 | 5. | castellani 20 6 83.333333 | |------------------------------------------| 6. | ginori 32 3 36 | +------------------------------------------+ . mata: st_matrix("R", nwt_reach2("flomarriage")) . matrix list R R[16,1] c1 r1 6 r2 10 r3 9 r4 8 r5 6 r6 3 r7 9 r8 4 r9 11 r10 2 r11 6 r12 0 r13 11 r14 7 r15 8 r16 10 . mata: st_numscalar("meanreach", mean(nwt_reach2("flomarriage"))) . display meanreach 6.875 . mata: ------------------------------------------------- mata (type end to exit) -------- : P = nwapi_matrix("flomarriage") : P = P*P : _diag(P, 0) : end ---------------------------------------------------------------------------------- . mata: st_local("labs", invtokens(nwapi_nodelabels("flomarriage")', ",")) . nwset, mat(P) name(paths2) labs(`labs') undirected . nwsummarize paths2 -------------------------------------------------- Network name: paths2 Network id: 3 Directed: false Valued: true Two-mode: false Nodes: 16 Selfloop: false Edges: 43 Minimum value: 0 Maximum value: 2 Density: .358 Mean degree: 5.38 Temporal: false . mata: mata drop P x . . * 6. A reusable command . capture program drop nwnbrwealth . program nwnbrwealth 1. syntax [anything(name=netname)], GENerate(name) VARiable(varname numeric) 2. _nwsyntax `netname' 3. _nwdatasync `netname' 4. confirm new variable `generate' 5. quietly generate double `generate' = . 6. mata: st_store(., "`generate'", nwt_nbrmean(st_local("netname"), st_data( > ., "`variable'"))) 7. end . nwwebuse florentine, nwclear . nwnbrwealth flomar, generate(nw1) variable(wealth) . nwnbrwealth flomarriage, generate(nw2) variable(wealth) . list _nwnode wealth nw1 nw2 in 1/4 +---------------------------------------------+ | _nwnode wealth nw1 nw2 | |---------------------------------------------| 1. | acciaiuoli 10 103 103 | 2. | albizzi 36 47.666667 47.666667 | 3. | barbadori 55 61.5 61.5 | 4. | bischeri 44 67.666667 67.666667 | +---------------------------------------------+ . . * 7. Guarding against old versions and errors . capture mata: nwapi_version() . display _rc 0 . mata: assert(nwapi_version() >= 1) . capture noisily mata: nwapi_nodes("friends") network friends not found . display _rc 482 . capture noisily mata: nwapi_neighbors("flomarriage", "nobody") node nobody not found in network flomarriage . display _rc 485 . capture noisily mata: nwapi_neighbors("flomarriage", 99) node 99 does not exist in network flomarriage (it has 16 nodes) . display _rc 485 . capture noisily mata: nwapi_neighbors("flomarriage", 1.5) the node number must be a whole number, not 1.5 . display _rc 198 . . * 8. Large networks . clear . quietly set obs 25000 . generate long a = _n . generate long b = mod(_n, 25000) + 1 . nwset a b, edgelist name(ring) undirected . mata: nwapi_nodes("ring"), nwapi_ties("ring") 1 2 +-----------------+ 1 | 25000 25000 | +-----------------+ . mata: E = nwapi_edgelist("ring") . mata: E[1..3, .] 1 2 3 +-------------------------+ 1 | 1 2 1 | 2 | 1 25000 1 | 3 | 2 3 1 | +-------------------------+ . mata: E[rows(E), .] 1 2 3 +-------------------------+ 1 | 24999 25000 1 | +-------------------------+ . mata: d = nwapi_outdegree("ring") . mata: min(d), max(d) 1 2 +---------+ 1 | 2 2 | +---------+ . mata: nwapi_neighbors("ring", 1)' 1 2 +-----------------+ 1 | 2 25000 | +-----------------+ . capture noisily mata: A = nwapi_matrix("ring") network ring has 25000 nodes; a dense 25000 x 25000 matrix would need about 4.66 G > B and is refused (more than 20000 nodes). Use nwapi_edgelist(), nwapi_neighbors( > ) or nwapi_outdegree() instead, which never need the dense matrix. . display _rc 484 . mata: mata drop E d . end of do-file