pukpr

High-Speed, Low-Cost Workstation for Computation-Intensive Statistics
NSF SBIR Phase I Report (1990)

Summary: An Early Precursor to Accelerator-Based Scientific Computing

NSF SBIR Program Context

The National Science Foundation (NSF) Small Business Innovation Research (SBIR) program awards contracts to small businesses to stimulate technological innovation, meet federal research and development needs, and increase private-sector commercialization of innovations derived from federal research.

Phase I awards (typically ~$50,000 in the late 1980s/early 1990s) fund feasibility studies — typically 6–12 months — to determine the technical merit and commercial potential of an idea. Successful Phase I projects may lead to Phase II for prototype development.

pukpr

C2M2L and the Context Server: Pre-LLM Knowledge-Based Modeling Support

The Component, Context, and Manufacturing Model Library (C2M2L) was developed under DARPA’s Adaptive Vehicle Make (AVM) program around 2012–2015. It created reusable, semantically rich libraries for components, manufacturing processes, and environmental contexts to accelerate model-based design, verification, and fabrication of complex vehicles (with applications far broader than military uses, including general engineering, environmental simulation, and systems design).

A core element was the Dynamic Context Server (DCS)—a Prolog-based modeling support tool designed for environmental context modeling (land, atmospheric, and aquatic domains). It enabled knowledge-based inferencing for vehicle-environment interactions, supporting formal terminology, abstraction levels, and reusable patterns.

Key Features of the Context Server (from the White Paper)

• Architecture: Built on a semantic web foundation using triple-s

pukpr

#!/usr/bin/env python3
"""
timeseries_modeler.py

Read a two-column CSV (time, value). Read a JSON file with the same filename
appended by ".p" (e.g. data.csv.p) into a data dictionary. Create a clone of the
time-series and run a loop that steps through timestamps, calling a user-
customizable model-step function to produce a modeled series. At the end the
script computes Pearson's correlation coefficient and the variance of the
squared errors (and also MSE), using library routines.

pukpr

import numpy as np
from sklearn.linear_model import Lasso

# ----------------------------
# USER INPUT: hidden layer & target
# ----------------------------
# Example synthetic data (replace with your arrays)
N = 1000
t = np.linspace(0, 10, N)
hidden_layer = np.sin(2*np.pi*1*t) + 0.5*np.sin(2*np.pi*3*t)  # latent forcing

pukpr

#!/usr/bin/env python3
"""
timeseries_modeler.py

Read a two-column CSV (time, value). Read a JSON file with the same filename
appended by ".p" (e.g. data.csv.p) into a data dictionary. Create a clone of the
time-series and run a loop that steps through timestamps, calling a user-
customizable model-step function to produce a modeled series. At the end the
script computes Pearson's correlation coefficient and the variance of the
squared errors (and also MSE), using library routines.

pukpr

{"PeriodsPhase":[13.599781479999999,7.510001926736354,18.679732370000025,7.195787246118572,20.998850680000004,9.392972953999996,6.001334823,10.345732619999998,9.26752208,6.276528506318169,6.158543069000001,15.797059490000018,5.047887753499271,7.347247053426061,1.7632983608814414,6.514613283064518,7.845674634],"Hold":0.0015847482104409999,"Initial":0.019102685632566375,"Imp_Stride":2,"LTE_Freq":182.73199150000377,"Periods":[27.2122,27.3216,27.5545,13.63339513,13.69114014,13.6061,13.6608,13.71877789,6795.985773,1616.215172,2120.513853,13.77725,3232.430344,9.132931547,9.108450374,9.120674533,27.0926041],"LTE_Amp":1.2282362293246383,"DeltaTime":0.0,"Imp_Amp":38.41771511999972,"AliasedPhase":[12.936492162500015,10.819234671777158,17.28825828537955,9.83624739277385,24.275678314901782,15.228945466500011,8.477116474739738,14.315824278046344,9.824004490952355,6.960279582531864,4.2964023424706665,19.297607654892147,3.967560759570126,22.113385778146025,8.444613035937136,4.885017823159738,10.475305999774953,6.49431729103

pukpr

pukpr

from vpython import *

Running = True

def exit_app(evt):
    global Running
    if evt.key == 'q':  # Press 'q' to quit
        print("Exiting simulation...")
        Running = False

pukpr

import numpy as np
import matplotlib.pyplot as plt
from scipy.signal import periodogram, windows

# Parameters
years = 500
dt = 1 / 365.25  # daily resolution
t = np.arange(0, years, dt)
fs = 1 / dt  # sampling frequency in cycles per year

pukpr

with Text_IO;
with GEM.LTE.Primitives;

function GEM.Mix_Regression (File_Name : in String) return Gem.LTE.Period_Set is
   use GEM.LTE, GEM.LTE.Primitives;
   D : Data_Pairs := Make_Data(File_Name);
   First, Last : Integer;
   Singular : Boolean;
   Forcing : Data_Pairs := D;
   Model : Data_Pairs := D;