Abstract
The existing interpolation and regression methods are highly data-specific, challenge-specific, or approach-specific. Peridynamic approach provides a single mathematical framework for diverse data-sets and multi-dimensional data manipulation and model order reduction. The mathematical framework based on the Peridynamic Differential Operator (PDDO) provides a unified approach to transfer information within a set of discrete data, and among data sets in multi-dimensional space. The robustness and capability of this approach have been demonstrated by considering various real or fabricated data concerning two- or three-dimensional applications. The numerical results concern interpolation of real data in two and three dimensions, interpolation to approximate a three-dimensional function, adaptive data recovery in three-dimensional space, recovery of missing pixels in an image, adaptive image compression and recovery, and free energy evaluation through model reduction.
Similar content being viewed by others
References
Franke R (1982) Scattered data interpolation: Tests of some methods. Math Comput 38:181–200
Mittas L, Mitasova H (1999) Spatial interpolation. In: Longley PI, Goodchild MF, Maguire DJ, Rhind DW (eds) Geographical Information Systems: Principles. Wiley, Techniques, Management and Applications, GeoInformation International, pp 481–492
Steffensen JF (2006) Interpolation, 2nd edn. Dover Publications
Cressie N (2015) Statistics for spatial data, Revised. John Wiley & Sons
Liszka T, Orkisz J (1980) The finite difference method at arbitrary irregular grids and its applications in applied mechanics. Comp Struct 11:83–95
Liszka T (1984) An interpolation method for irregular net of nodes. Int J Numer Meth In Engr 20:1599–1612
Madenci E, Barut A, Futch M (2016) Peridynamic differential operator and its applications. Comp Meth Appl Mech Eng 306:408–451
Madenci E, Barut A, Dorduncu M (2019) Peridynamic differential operator for numerical analysis. Springer, New York
Madenci E, Barut A, Dorduncu M (2019) Peridynamics for predicting thermal expansion coefficient of grapheme. 69th Electronic Components & Technology Conference, Las Vegas, NV
Madenci E, Barut A, Prohit P (2020) A Peridynamic approach to computation of elastic and entropic interactions of inclusions on a lipid membrane. J Mech Phys Solids 104046
Su T, Purohit PK (2010) Thermomechanics of a heterogeneous fluctuating chain. J Mech Phys Solids 58:164–186
Boal D (2002) Mechanics of the cell. Cambridge University Press, Cambridge, UK
Zhang Y, Crothers DM (2003) Statistical mechanics of sequence-dependent circular DNA and its application for DNA cyclization. Biophys J 84:136–153
Funding
This study was performed as part of the ongoing research at the MURI Center for Material Failure Prediction through Peridynamics at the University of Arizona (AFOSR Grant No. FA9550-14-1-0073).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of Interest
The authors declare no competing interests.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendices
Appendix 1 Peridynamic differential operator
According to the 2-order TSE in a 2-dimensional space, the following expression holds
where R is the small remainder term. Multiplying each term with PD functions, \({g}_{2}^{{p}_{1}{p}_{2}}(\xi )\), and integrating over the domain of interaction (family), \({H}_{\mathbf{x}}\), results in
in which the point x is not necessarily symmetrically located in the domain of interaction. The initial relative position, \(\xi\), between the material points x and \({\mathbf{x}}^{{\prime}}\) can be expressed as \(\xi =\mathbf{x}-{\mathbf{x}}^{{\prime}}\). This ability permits each point to have its own unique family with an arbitrary position. Therefore, the size and shape of each family can be different, and they significantly influence the degree of non-locality. The degree of the interaction between the material points in each family is specified by a non-dimensional weight function, \(w(\left|\xi \right|)\), which can vary from point to point. The interactions become more local with a decreasing family size. Thus, the family size and shape are important parameters. In general, the family of a point can be non-symmetric due to non-uniform spatial discretization. Each point has its own family members in the domain of interaction (family), and occupies an infinitesimally small entity such as volume, area, or a distance.
The PD functions are constructed such that they are orthogonal to each term in the TS expansion as
with \(({n}_{1},{n}_{2},p,q=\mathrm{0,1},2)\) and \({\delta }_{{n}_{i}{p}_{i}}\) is a Kronecker symbol.
Enforcing the orthogonality conditions in the TSE leads to the non-local PD representation of the function itself and its derivatives as
The PD functions can be constructed as a linear combination of polynomial basis functions:
where \({a}_{{q}_{1}{q}_{2}}^{\,{p}_{1}{p}_{2}}\) are the unknown coefficients, \({w}_{{q}_{1}{q}_{2}}(\left|\xi \right|)\) are the influence functions, and \({\xi }_{1}\) and \({\xi }_{2}\) are the components of the vector \(\xi\). Assuming \({w}_{{p}_{1}{p}_{2}}(\left|\xi \right|)=w(\left|\xi \right|)\) and submitting the PD functions into the orthogonality equation lead to a system of algebraic equations for the determination of the coefficients as
in which
and
After determining the coefficients \({a}_{{q}_{1}{q}_{2}}^{{p}_{1}{p}_{2}}\) via \(\mathbf{a}={\mathbf{A}}^{-1}\mathbf{b}\), then the PD functions \({g}_{2}^{{p}_{1}{p}_{2}}(\xi )\) can be constructed. The detailed derivations and the associated computer programs can be found in [8].
Appendix 2 Ordinary Kriging
As described by Cressie [4], the observed (known) data at spatial locations, \({s}_{1},.....,{s}_{n}\), are modeled as a random process denoted by \(Z(s)\) in ordinary Kriging. It is assumed that the random process satisfies the decomposition
where \(\mu\) is the expected value or mean of \(Z(s)\) and \(\delta (s)\) is the correlated error process. The mean, \(\mu\), is unknown; however, it is assumed to be a constant. The predictor is defined as
subject to the constraint of
The parameter, B, is defined as a block of spatial area whose location and geometry are known. The optimal predictor is obtained by minimizing the mean-squared prediction error defined as
with respect to the coefficients \({\lambda }_{i}\) with \(i=1,...,n\).
Appendix 3 Temperature Data from Weather Stations in Arizona
Weather station name | Elevation (m) | Latitude | Longitude | Temp. (°F) |
---|---|---|---|---|
Robson Ranch | 455.7 | 32.8118 | −111.6313 | 65 |
Tacna 3 NE | 98.8 | 32.7225 | −113.9191 | 68 |
Fry | 2194.6 | 35.07 | −111.84 | 52 |
Gunsight | 1609.3 | 36.7044 | −112.5833 | 50 |
Williams | 2105.9 | 35.2413 | −112.1929 | 56 |
Guthrie | 1097.3 | 32.8819 | −109.3092 | 53 |
Rincon | 2511.6 | 32.2056 | −110.5481 | 60 |
Frazier Wells | 2063.5 | 35.8456 | −113.055 | 52 |
Bisbee 1 WNW | 1694.7 | 31.4475 | −109.9288 | 61 |
Tempe ASU | 355.7 | 33.4258 | −111.9216 | 71 |
Wittmann 1 SE | 513 | 33.7776 | −112.523 | 71 |
Painted Desert National Park | 1755.6 | 35.068 | −109.7688 | 50 |
Union Pass | 1072.9 | 35.2247 | −114.3747 | 62 |
Humbug Creek | 1600.2 | 34.1164 | −112.3006 | 60 |
Hurricane | 1659.6 | 36.6992 | −113.2072 | 47 |
Saguaro National Park | 938.8 | 32.1794 | −110.7363 | 57 |
Willcox | 1271 | 32.2553 | −109.8369 | 58 |
Oak Creek | 1500.8 | 34.9417 | −111.7517 | 56 |
Greer | 2499.4 | 34.06 | −109.45 | 57 |
Safford Agricultural Center | 900.4 | 32.815 | −109.6808 | 63 |
Betatakin | 2220.8 | 36.6778 | −110.5411 | 44 |
Tohono Chul | 770.2 | 32.3391 | −110.9808 | 63 |
Empire | 1417.3 | 31.7806 | −110.6347 | 69 |
Four Springs | 1999.5 | 36.7939 | −112.0422 | 41 |
Teec Nos Pos | 1612.4 | 36.9233 | −109.09 | 42 |
Patagonia Paton Center | 1232.6 | 31.53923 | −110.76028 | 72 |
Payson | 1516.4 | 34.2431 | −111.3028 | 61 |
Chiricahua | 1645.9 | 32 | −109.35 | 62 |
Paria Point | 2205.2 | 36.7278 | −111.8219 | 49 |
Cherry | 1554.5 | 34.5964 | −112.0481 | 58 |
Sanders | 1784 | 35.2239 | −109.3222 | 52 |
San Carlos Reservoir | 771.8 | 33.1819 | −110.5261 | 46 |
Goodwin Mesa | 1280.2 | 34.75 | −113.3 | 63 |
Sunset Crater National Monument | 2127.5 | 35.3694 | −111.5436 | 42 |
Sunrise Mountain | 2856 | 33.9733 | −109.563 | 42 |
Flagstaff | 2133.6 | 35.145 | −111.675 | 56 |
Hopi | 1885.5 | 35.8103 | −110.2069 | 53 |
Selles | 721.2 | 31.91 | −111.8975 | 69 |
Iron Springs | 1804.4 | 34.5853 | −112.5019 | 61 |
Hopkins | 2170.2 | 31.6753 | −110.88 | 66 |
Tucson International Airport | 776.9 | 32.1313 | −110.9552 | 70 |
Mormon Mountain | 2286 | 34.94 | −111.52 | 53 |
Catalina State Park | 825.1 | 32.4177 | −110.9302 | 61 |
Sunset Point | 902.2 | 34.1953 | −112.1417 | 64 |
Tweeds Point | 1585 | 36.5819 | −113.7319 | 51 |
Douglas Bisbee Inter. Airport | 1251.2 | 31.4583 | −109.6061 | 66 |
St. Johns Industrial Air Park | 1747.4 | 34.51833 | −109.37917 | 52 |
Nixon Flats | 1981.2 | 36.39 | −113.1522 | 54 |
Black Rock | 2158 | 36.7944 | −113.7567 | 45 |
Tusayan | 2042.2 | 35.99 | −112.12 | 55 |
Stray Horse | 2139.7 | 33.5406 | −109.3169 | 56 |
Snowslide Canyon | 2965.7 | 35.34 | −111.65 | 47 |
Stanton | 1097.3 | 34.1667 | −112.7333 | 64 |
Olaf Knolls | 883.9 | 36.5072 | −113.8161 | 60 |
Kingman Airport | 1042.4 | 35.2577 | −113.933 | 63 |
Douglas | 1231.4 | 31.345 | −109.5394 | 66 |
Nogales 6 N | 1054.9 | 31.4554 | −110.968 | 73 |
Truxton Canyon | 1630.7 | 35.7825 | −113.7942 | 56 |
Lindbergh Hill | 2682.2 | 36.2858 | −112.0794 | 48 |
Bagdad | 1199.4 | 34.5975 | −113.1745 | 57 |
White Horse Lake | 2188.5 | 35.14 | −112.15 | 56 |
Alamo Dam | 393.2 | 34.228 | −113.5777 | 65 |
Walnut Creek | 1551.4 | 34.9281 | −112.8097 | 62 |
Walnut Canyon National Monument | 2040.6 | 35.1721 | −111.5097 | 43 |
Dry Lake | 2264.1 | 33.3597 | −109.8331 | 60 |
Casa Grande | 426.7 | 32.8875 | −111.7147 | 67 |
Duncan | 1115.6 | 32.748 | −109.1213 | 57 |
Dry Park | 2653.6 | 36.45 | −112.24 | 49 |
Heber Black Mesa Ranger Station | 2008.6 | 34.3925 | −110.558 | 56 |
Alpine | 2447.8 | 33.8417 | −109.1222 | 57 |
Prescott Love Field | 1536.8 | 34.65167 | −112.42083 | 62 |
Phoenix Airport | 337.4 | 33.4277 | −112.0038 | 68 |
Benson 6 SE | 1124.7 | 31.8805 | −110.2403 | 58 |
Yuma Proving Ground | 98.8 | 32.8356 | −114.3942 | 68 |
Winslow Airport | 1489.3 | 35.0281 | −110.7208 | 51 |
Fort Valley | 2240.3 | 35.27 | −111.74 | 56 |
Happy Jack Ranger Station | 2279.9 | 34.7433 | −111.4139 | 47 |
Bellemont Weather Forecast Office | 2179.9 | 35.2302 | −111.8221 | 50 |
Beaver Dam | 588.6 | 36.9139 | −113.9423 | 60 |
Sasabe | 1094.2 | 31.483 | −111.5436 | 75 |
Show Low Airport | 1954.1 | 34.2639 | −110.0075 | 50 |
Window Rock Airport | 2054 | 35.6575 | −109.06139 | 52 |
Moss Basin | 1804.4 | 35.0336 | −113.8925 | 58 |
Picacho 8 SE | 603.8 | 32.6463 | −111.4017 | 63 |
Jerome | 1508.8 | 34.7522 | −112.1114 | 49 |
Quartzsite | 266.7 | 33.665 | −114.2272 | 69 |
Page Municipal Airport | 1313.7 | 36.92611 | −111.44778 | 43 |
Warm Springs Canyon | 2441.4 | 36.7 | −112.23 | 51 |
Kitt Peak | 2069.6 | 31.96018 | −111.59787 | 60 |
Workman Creek | 2103.1 | 33.81 | −110.92 | 55 |
Yellow John Mountain | 1877.6 | 36.1542 | −113.5417 | 54 |
Carefree | 771.1 | 33.8161 | −111.9019 | 67 |
Grand Canyon National Park Airport | 2013.5 | 35.94611 | −112.15472 | 49 |
Montezuma Castle Nat.Monument | 969.3 | 34.6105 | −111.838 | 59 |
Scottsdale Municipal Airport | 449 | 33.62278 | −111.91056 | 68 |
Havasu | 144.8 | 34.7872 | −114.5617 | 69 |
Limestone Canyon | 2072.6 | 34.1789 | −110.2736 | 51 |
Kartchner Caverns | 1429.5 | 31.8352 | −110.3552 | 53 |
Buckskin Mountain | 1950.7 | 36.9306 | −112.1997 | 48 |
Mount Lemmon Willow Canyon | 2141.8 | 32.3859 | −110.69799 | 47 |
Apache Junction 5 NE | 630.9 | 33.4625 | −111.4813 | 66 |
Anvil Ranch | 840.9 | 31.9793 | −111.3837 | 63 |
Canyon de Chelly | 1709.9 | 36.1533 | −109.5394 | 47 |
Green Valley | 883.9 | 31.893 | −110.9977 | 59 |
Bright Angel Ranger Station | 2438.4 | 36.2147 | −112.0619 | 41 |
Organ Pipe Cactus Nat. Monument | 511.5 | 31.9555 | −112.8002 | 73 |
Fort Huachuca Pioneer Airfield | 1453.3 | 31.60722 | −110.42806 | 66 |
Globe | 1112.5 | 33.3503 | −110.6519 | 56 |
San Simon | 1091.2 | 32.29329 | −109.22691 | 61 |
Phantom Ranch | 771.1 | 36.1066 | −112.0947 | 58 |
Petrified Forest National Park | 1659.9 | 34.7994 | −109.885 | 49 |
Saint Johns | 1764.8 | 34.5172 | −109.4028 | 49 |
Meteor Crater | 1687.1 | 35.0364 | −111.0231 | 47 |
Music Mountain | 1652 | 35.6147 | −113.7939 | 56 |
Muleshoe Ranch | 1272.5 | 32.4 | −110.2708 | 67 |
Tombstone | 1420.1 | 31.7119 | −110.0686 | 56 |
Sierra Vista | 1403.9 | 31.53699 | −110.28073 | 53 |
Hilltop | 1743.5 | 33.6183 | −110.42 | 61 |
Elgin 5 S | 1466.4 | 31.5907 | −110.5087 | 67 |
Ajo | 533.7 | 32.3698 | −112.8599 | 66 |
Smith Peak | 762 | 34.1158 | −113.3472 | 65 |
Rights and permissions
About this article
Cite this article
Madenci, E., Barut, A., Willmarth, E. et al. Peridynamics for Data Estimation, Image Compression/Recovery, and Model Reduction. J Peridyn Nonlocal Model 4, 159–200 (2022). https://doi.org/10.1007/s42102-021-00072-z
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s42102-021-00072-z