Rcpp::List run(const data_set& data, MODEL& model, SGD& sgd) {
unsigned n_samples = data.n_samples;
unsigned n_features = data.n_features;
unsigned n_passes = sgd.get_n_passes();
bool good_gradient = true;
bool good_validity = true;
bool averaging = false;
if (sgd.name() == "asgd" || sgd.name() == "ai-sgd") {
averaging = true;
}
// TODO these should really be vec's
mat theta_new;
mat theta_new_ave;
mat theta_old = sgd.get_last_estimate();
mat theta_old_ave = theta_old;
unsigned max_iters = n_samples*n_passes;
bool do_more_iterations = true;
bool converged = false;
if (sgd.verbose()) {
Rcpp::Rcout << "Stochastic gradient method: " << sgd.name() << std::endl;
Rcpp::Rcout << "SGD Start!" << std::endl;
}
for (unsigned t = 1; do_more_iterations; ++t) {
theta_new = sgd.update(t, theta_old, data, model, good_gradient);
if (averaging) {
if (t != 1) {
theta_new_ave = (1. - 1./(double)t) * theta_old_ave +
1./((double)t) * theta_new;
} else {
theta_new_ave = theta_new;
}
sgd = theta_new_ave;
} else {
sgd = theta_new;
}
good_validity = validity_check(data, theta_new, good_gradient, t, model);
if (!good_validity) {
return Rcpp::List();
}
// Check if satisfy convergence threshold.
if (averaging) {
converged = sgd.check_convergence(theta_new_ave, theta_old_ave);
} else {
converged = sgd.check_convergence(theta_new, theta_old);
}
if (converged) {
sgd.end_early();
do_more_iterations = false;
}
// Stop if hit maximum number of iterations.
if (t == max_iters) {
//if (!sgd.pass()) {
// Rcpp::Rcout
// << "Informational Message: The maximum number of iterations is "
// << "reached! The algorithm has not converged."
// << std::endl
// << "Estimates from this stochastic gradient descent are not "
// << "guaranteed to be meaningful."
// << std::endl;
//}
do_more_iterations = false;
}
// Set old to new updates and repeat.
if (averaging) {
theta_old_ave = theta_new_ave;
}
theta_old = theta_new;
}
Rcpp::List model_out = post_process(sgd, data, model);
return Rcpp::List::create(
Rcpp::Named("model") = model.name(),
Rcpp::Named("coefficients") = sgd.get_last_estimate(),
Rcpp::Named("converged") = converged,
Rcpp::Named("estimates") = sgd.get_estimates(),
Rcpp::Named("pos") = sgd.get_pos(),
Rcpp::Named("times") = sgd.get_times(),
Rcpp::Named("model.out") = model_out);
}
static int validate_device(const device_class *devclass)
{
int error = 0;
int is_invalid, i;
const char *s;
INT64 devcount, optcount;
char buf[256];
char *s1;
char *s2;
iodevice_t devtype;
int (*validity_check)(const device_class *devclass);
/* critical information */
devtype = (iodevice_t) (int) device_get_info_int(devclass, DEVINFO_INT_TYPE);
devcount = device_get_info_int(devclass, DEVINFO_INT_COUNT);
/* sanity check device type */
if (devtype >= IO_COUNT)
{
printf("%s: invalid device type %i\n", devclass->gamedrv->name, (int) devtype);
error = 1;
}
/* sanity check device count */
if ((devcount <= 0) || (devcount > MAX_DEV_INSTANCES))
{
printf("%s: device type '%s' has an invalid device count %i\n", devclass->gamedrv->name, device_typename(devtype), (int) devcount);
error = 1;
}
/* File Extensions Checks
*
* Checks the following
*
* 1. Tests the integrity of the string list
* 2. Checks for duplicate extensions
* 3. Makes sure that all extensions are either lower case chars or numbers
*/
s = device_get_info_string(devclass, DEVINFO_STR_FILE_EXTENSIONS);
if (!s)
{
printf("%s: device type '%s' has null file extensions\n", devclass->gamedrv->name, device_typename(devtype));
error = 1;
}
else
{
memset(buf, 0, sizeof(buf));
strcpy(buf, s);
/* convert to be null delimited */
s1 = buf;
while(*s1)
{
if (*s1 == ',')
*s1 = '\0';
s1++;
}
s1 = buf;
while(*s1)
{
/* check for invalid chars */
is_invalid = 0;
for (s2 = s1; *s2; s2++)
{
if (!isdigit(*s2) && !islower(*s2))
is_invalid = 1;
}
if (is_invalid)
{
printf("%s: device type '%s' has an invalid extension '%s'\n", devclass->gamedrv->name, device_typename(devtype), s1);
error = 1;
}
s2++;
/* check for dupes */
is_invalid = 0;
while(*s2)
{
if (!strcmp(s1, s2))
is_invalid = 1;
s2 += strlen(s2) + 1;
}
if (is_invalid)
{
printf("%s: device type '%s' has duplicate extensions '%s'\n", devclass->gamedrv->name, device_typename(devtype), s1);
error = 1;
}
s1 += strlen(s1) + 1;
}
}
/* enforce certain rules for certain device types */
switch(devtype)
{
case IO_QUICKLOAD:
case IO_SNAPSHOT:
if (devcount != 1)
{
printf("%s: there can only be one instance of devices of type '%s'\n", devclass->gamedrv->name, device_typename(devtype));
error = 1;
}
/* fallthrough */
case IO_CARTSLOT:
if (!device_get_info_int(devclass, DEVINFO_INT_READABLE)
|| device_get_info_int(devclass, DEVINFO_INT_WRITEABLE)
|| device_get_info_int(devclass, DEVINFO_INT_CREATABLE))
{
printf("%s: devices of type '%s' has invalid open modes\n", devclass->gamedrv->name, device_typename(devtype));
error = 1;
}
break;
default:
break;
}
/* check creation options */
optcount = device_get_info_int(devclass, DEVINFO_INT_CREATE_OPTCOUNT);
if ((optcount < 0) || (optcount >= DEVINFO_CREATE_OPTMAX))
{
printf("%s: device type '%s' has an invalid creation optcount\n", devclass->gamedrv->name, device_typename(devtype));
error = 1;
}
else
{
for (i = 0; i < (int) optcount; i++)
{
if (!device_get_info_string(devclass, DEVINFO_STR_CREATE_OPTNAME + i))
{
printf("%s: device type '%s' create option #%d: name not present\n",
devclass->gamedrv->name, device_typename(devtype), i);
error = 1;
}
if (!device_get_info_string(devclass, DEVINFO_STR_CREATE_OPTDESC + i))
{
printf("%s: device type '%s' create option #%d: description not present\n",
devclass->gamedrv->name, device_typename(devtype), i);
error = 1;
}
if (!device_get_info_string(devclass, DEVINFO_STR_CREATE_OPTEXTS + i))
{
printf("%s: device type '%s' create option #%d: extensions not present\n",
devclass->gamedrv->name, device_typename(devtype), i);
error = 1;
}
}
}
/* is there a custom validity check? */
validity_check = (int (*)(const device_class *)) device_get_info_fct(devclass, DEVINFO_PTR_VALIDITY_CHECK);
if (validity_check)
{
if (validity_check(devclass))
error = 1;
}
return error;
}